Command<\/th>\n	Kind<\/th>\n	Function<\/th>\n<\/tr>\n<\/thead>\n
\/spdd-story<\/a><\/td>\n	Non-compulsory<\/td>\n	Breaks a big requirement into unbiased, deliverable \n consumer tales following the INVEST precept.<\/td>\n<\/tr>\n
\/spdd-analysis<\/a><\/td>\n	Core<\/td>\n	Extracts area key phrases from necessities, scans related \n code, and produces a strategic evaluation protecting area ideas, \n dangers, and design course.<\/td>\n<\/tr>\n
\/spdd-reasons-canvas<\/a><\/td>\n	Core<\/td>\n	Generates the complete REASONS Canvas \u2014 an executable blueprint \n from high-level rationale all the way down to method-level operations.<\/td>\n<\/tr>\n
\/spdd-generate<\/a><\/td>\n	Core<\/td>\n	Reads the Canvas and generates code job by job, strictly \n following the operations, norms, and safeguards outlined within the \n immediate.<\/td>\n<\/tr>\n
\/spdd-api-test<\/a><\/td>\n	Non-compulsory<\/td>\n	Generates a cURL-based API check script with structured check \n circumstances protecting regular, boundary, and error situations.<\/td>\n<\/tr>\n
\/spdd-prompt-update<\/a><\/td>\n	Core<\/td>\n	Incrementally updates the Canvas when necessities change \n (necessities \u2192 immediate \u2192 code).<\/td>\n<\/tr>\n
\/spdd-sync<\/a><\/td>\n	Core<\/td>\n	Synchronizes code-side adjustments (refactoring, fixes) again \n into the Canvas so the immediate stays an correct file of the \n present code (code \u2192 immediate).<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/section>\n<\/section>\n \n Enhancing a billing engine with SPDD<\/h2>\n A sophisticated workflow is obscure within the summary, so \n now we have ready an instance workflow of enhancing an present software program \n system. This technique, and its enhancement, are neccessarily small so as \n to be understandable inside a tutorial article. That mentioned the enhancement \n instance is a full end-to-end instance: from creating preliminary \n necessities, to analyzing enterprise necessities, to producing and reviewing a structured immediate, \n to producing and verifying code, to ultimate cleanup and testing.<\/p>\n You possibly can observe together with this instance by putting in \n openspdd<\/a> in your individual surroundings.<\/p>\n \n The present system<\/h3>\n The present system is an easy billing engine that calculates payments \n for utilizing a large-language mannequin. It accepts a file that captures how \n many tokens are utilized in a session and calculates a invoice.<\/p>\n The whole codebase for this preliminary model is \n accessible on GitHub<\/a>. \n The repository contains the preliminary necessities story<\/a> \n and all of the SPDD artifacts used to generate it<\/a>. \n For brevity, we do not describe that preliminary technology right here, but it surely follows primarily the identical steps as that for the enhancement. \n We deal with describing the enhancement as a result of most work on a system are enhancements.<\/p>\n<\/section>\n \n The enhancement<\/h3>\n Pushed by evolving enterprise necessities and direct consumer suggestions, we’re enhancing the billing engine to transition from a static pricing mannequin to a extra refined, \n versatile infrastructure. This replace goals to help various subscription methods and variable, model-specific pricing via the next key adjustments:<\/p>\n \n API enhancement: replace the present `POST \/api\/utilization<\/code> \n endpoint to just accept a brand new, required modelId<\/code> parameter \n (e.g., \u201cfast-model\u201d, \u201creasoning-model\u201d).<\/li>\n` `Mannequin-aware pricing: shift from a single world fee to dynamic \n pricing, the place prices fluctuate relying on the precise AI mannequin \n invoked.<\/li>\n` `Multi-plan billing logic: introduce distinct billing behaviors \n primarily based on the client’s subscription tier:<\/li>\n`\nCustomary plan (optimized): retains the worldwide month-to-month quota, \n however any overage utilization is now calculated utilizing model-specific \n charges.<\/li>\n Premium plan (new): operates with out a quota restrict. It \n introduces cut up billing, the place immediate tokens and completion \n tokens are charged individually at totally different charges relying \n on the mannequin used.<\/li>\n<\/ul>\nArchitectural scalability: implement an extensible design sample \n (similar to Technique or Manufacturing facility) to cleanly isolate the calculation \n formulation for various plans, making certain the system can simply \n accommodate future pricing fashions.<\/li>\n<\/ul>\nSince this new part encompasses each enterprise necessities and technical particulars, \n it’s sometimes accomplished collaboratively via a pairing session between a PO (or BA) and a developer.<\/i><\/p>\n<\/section>\n\nStep 1: Create preliminary necessities<\/h3>\nTo kick off the method rapidly, we will use the \/spdd-story<\/code> \n command to generate a consumer story instantly primarily based on the enhancement. Usually, consumer tales are offered by the PO or BA. \n Nevertheless, in our workflow, we will remodel tales of any type right into a constant format and dimension. \n So long as there’s shared alignment on the ultimate acceptance standards, this step might be carried out by a PO, BA, or developer, relying on the crew’s versatile division of labor.<\/p>\n Instruction:<\/p>\n\nHow spdd-story<\/a> works<\/h3>\nThis command breaks a big requirement into unbiased, \n deliverable consumer tales following the INVEST precept (1\u20135 days \n of labor every). Every story contains acceptance standards written in \n enterprise language, able to function enter for \n \/spdd-analysis<\/code>.<\/p>\n Its goal is to make giant necessities manageable and to make sure a standardized, \n predictable format for the subsequent steps.<\/p>\n<\/div>\n The AI analyzed the enhancement description and cut up it into two \n consumer tales:<\/p>\n The auto-generated tales are detailed sufficient to function a baseline \n for a proper venture. For this walkthrough we consolidate them right into a \n single simplified story so the instance stays self-contained.<\/p>\n Instruction:<\/p>\n\nConsolidate the next two consumer tales right into a single, simplified \n story: @[User-story-1-1-initial]Multi-Plan-Billing-Basis-&-Customary-Plan-Mannequin-Conscious-Pricing.md @[User-story-1-2-initial]Premium-Plan-Break up-Price-Billing.md<\/p>\n Necessities: 1. Merge each plans (Customary and Premium) into one coherent story. 2. Hold solely the sections: Background, Enterprise Worth, Scope In, Scope Out, and Acceptance Standards. 3. Strip implementation-level element \u2014 deal with what the system ought to do, not how. 4. Acceptance Standards should use Given\/When\/Then format with concrete numeric examples. 5. Hold the outcome concise \u2014 no multiple web page. 6. Solely maintain three high-level ACs.\n <\/p>\n<\/div>\n Directions of this sort hardly ever produce similar textual content on each run \u2014 \n fashions and sampling introduce small variations \u2014 so we nonetheless count on to \n evaluation and tweak the output earlier than treating it as ultimate. The mixed \n story under is the model we refined for this walkthrough: a intentionally \n simplified consolidation of the 2 preliminary tales.<\/p>\n<\/section>\n\nStep 2: Make clear evaluation<\/h3>\nEarlier than leaping into implementation, the developer opinions the consumer \n story to construct a shared understanding of what it means in apply. If \n there are apparent business-level points, that is the purpose to align with \n the BA or PO. On this case the story is obvious sufficient, so we transfer \n straight to breaking it down alongside three dimensions: core logic, scope \n boundaries, and definition of finished.<\/p>\n Core logic<\/b><\/p>\n One new required subject on the API: modelId<\/code>. The client \n now tells us which AI mannequin they used \u2014 that is the important thing that unlocks the \n proper value.<\/p>\n\nCustomary Plan:<\/i> Buyer has a month-to-month token quota. Utilization \n inside quota is free. Overage is charged at a model-specific fee \n (e.g., fast-model $0.01\/1K vs. reasoning-model $0.03\/1K). Current \n quota logic stays; solely the speed lookup adjustments.<\/li>\n Premium Plan:<\/i> No quota. Each token is billed from the \n first one. Immediate tokens and completion tokens are charged individually, \n every at a model-specific fee. Invoice = immediate cost + completion \n cost. This plan is totally new.<\/li>\nRouting:<\/i> The system determines the client’s plan and \n dispatches to the matching billing components. The design have to be simple to \n lengthen \u2014 Enterprise plans (Story 2) are subsequent.<\/li>\n<\/ul>\nScope boundaries<\/b><\/p>\n We’re solely calculating the present invoice. We’re NOT constructing buyer \n CRUD, NOT querying historic payments, NOT managing subscriptions, and NOT \n including\/eradicating fashions.<\/p>\n Definition of finished<\/b><\/p>\n The next situations restate the story’s acceptance standards \n with the implementation element the crew must confirm. The fourth \n merchandise (Response format) shouldn’t be a brand new enterprise AC \u2014 it captures the \n non-functional contract the developer provides to make the factors \n testable end-to-end.<\/p>\n\nValidation:<\/i> Lacking modelId<\/code> \u2192 HTTP 400. \n Unknown buyer \u2192 HTTP 404. Damaging tokens \u2192 HTTP 400. All \n present validations stay intact.<\/li>\n Customary Plan billing:<\/i> A buyer with a 100K quota and \n 90K already used submits 30K tokens for fast-model ($0.01\/1K). \n Anticipated outcome: 10K coated by quota, 20K overage, cost $0.20. \n The identical request with reasoning-model ($0.03\/1K) yields $0.60 \u2014 \n similar quota logic, totally different fee.<\/li>\n Premium Plan billing:<\/i> A buyer submits 10K immediate \n tokens + 20K completion tokens for reasoning-model (immediate $0.03\/1K, \n completion $0.06\/1K). Anticipated outcome: $0.30 + $1.20 = $1.50. No \n quota, no overage \u2014 immediate and completion are billed individually.<\/li>\nResponse format:<\/i> HTTP 201 returning invoice ID, buyer ID, \n token counts, timestamp, modelId<\/code>, and a plan-appropriate \n cost breakdown.<\/li>\n<\/ul>\nIf all these situations go, we have conquered this story.<\/p>\n<\/section>\n\nStep 3: Generate evaluation context<\/h3>\nWith the necessities and scope clarified, we use the \n \/spdd-analysis<\/code> command. By feeding it the enterprise \n necessities, we instruct the AI to generate a complete evaluation \n context.<\/p>\n \nHow spdd-analysis<\/a> works<\/h3>\nThis command extracts area key phrases from the enterprise \n necessities (e.g. \u201cbilling\u201d, \u201cquota\u201d, \u201cplan\u201d) and makes use of them to \n scan solely the related elements of the codebase \u2014 not all of it. It \n identifies present ideas, new ideas, key enterprise guidelines, \n and technical dangers.<\/p>\n The output is a context-rich doc protecting area idea \n recognition, strategic course, and danger evaluation. It serves as \n enter for the subsequent step: producing the REASONS Canvas.<\/p>\n<\/div>\n Instruction:<\/p>\n \/spdd-analysis \n @[User-story-1]Multi-Plan-Billing-Basis-&-Mannequin-Conscious-Pricing.md<\/p>\nGenerated artifact: the preliminary evaluation context doc<\/a>.<\/p>\n This command produces a strategic-level evaluation grounded in precise \n codebase exploration. The output focuses totally on the \u201cwhat\u201d and \n \u201cwhy,\u201d intentionally avoiding granular implementation particulars at this \n stage. It sometimes covers:<\/p>\n \nArea ideas: present vs. new, relationships, enterprise \n guidelines<\/li>\n Strategic method: answer course, design choices, \n trade-offs<\/li>\nDangers & gaps: ambiguities, edge circumstances, technical dangers, \n acceptance-criteria protection<\/li>\n<\/ul>\n\nEvaluation and refine the evaluation context<\/h4>\nWith our personal understanding of the enterprise necessities in thoughts, we \n evaluation the generated evaluation doc\u2014specializing in the areas \n highlighted within the alignment<\/a> talent. This evaluation serves \n two functions: confirming that our understanding aligns with the AI’s \n interpretation, and discovering edge circumstances or boundary situations the AI \n may floor that we hadn’t thought of.<\/p>\n On this particular occasion, the evaluation targeted on a number of vital \n areas:<\/p>\n \nWhether or not the Technique Sample was appropriately thought of.<\/li>\n Adherence to the OOP ideas established within the present system, \n particularly ISP and SRP.<\/li>\n The validity of the proposed technique for including new fields.<\/li>\n Figuring out edge circumstances not beforehand anticipated.<\/li>\nUncovering potential technical dangers.<\/li>\n<\/ul>\nUpon finishing the evaluation, the AI’s evaluation largely aligned with \n our architectural intent; in reality, its issues have been much more \n complete than ours in sure areas.<\/p>\n<\/p>\nEdge circumstances and dangers from the \n evaluation doc<\/a><\/p>\n<\/div>\n To be clear, at this stage we solely possess a high-level \n conceptual alignment. Whereas we will rapidly envision the implementation \n for areas the place now we have prior expertise, we can’t utterly map out \n all of the granular technical particulars for the unfamiliar elements proper \n now.<\/p>\n Nevertheless, that’s completely wonderful. The overarching course is aligned. \n We will proceed to the subsequent step: observing how the AI \u201csimulates\u201d the \n concrete implementation particulars inside our established framework and \n context. As soon as now we have these tangible particulars, we will uncover deeper, \n hidden points and make knowledgeable trade-offs primarily based on the precise \n state of affairs\u2014adopting the approaches the place the advantages outweigh the \n drawbacks, and discarding the remainder.<\/p>\n Determination: settle for the evaluation as-is and proceed.<\/p>\n<\/section>\n<\/section>\n \nStep 4: Generate structured immediate<\/h3>\n\nHow spdd-reasons-canvas<\/a> works<\/h3>\nThis command reads enterprise context (the output of \n \/spdd-analysis<\/code>, or a direct necessities description) \n and combines it with the present state of the codebase. It then \n generates a design specification throughout all seven REASONS \n dimensions \u2014 from \u201cwhy are we doing this\u201d to \u201cwhat should we not \n do.\u201d<\/p>\n The output is an executable blueprint. The Operations \n part is exact all the way down to technique signatures, parameter sorts, and \n execution steps.<\/p>\n<\/div>\n Instruction:<\/p>\n \/spdd-reasons-canvas \n @GGQPA-001-202603191100-[Analysis]-multi-plan-billing-model-aware-pricing.md<\/p>\nGenerated artifact: the preliminary structured immediate<\/a>.<\/p>\n By this level, we have already gone via high-level technique throughout \n the evaluation part\u2014so when reviewing the structured immediate, we’re not \n ranging from scratch. As a substitute, we’re inspecting how the AI has \n translated our shared understanding into the REASONS Canvas construction: \n from technique to abstraction to concrete particulars.<\/p>\n Consider it as a development: the evaluation part gave us strategic \n readability; now we’re checking whether or not that readability has been faithfully \n carried via into the architectural abstractions and implementation \n specifics. That is intent alignment at a deeper degree\u2014making certain that \n earlier than any code is generated, the AI has successfully \u201csimulated\u201d the \n whole answer inside our outlined framework. We get to evaluation from a \n world perspective quite than getting misplaced in particulars from the \n begin.<\/p>\n Focus the evaluation on the areas highlighted within the abstraction-first<\/a> talent. On this case, \n this foundational context is already embedded within the codebase and the earlier \n structured immediate<\/a>. Consequently, when producing the structured \n immediate for this iteration, the AI naturally elements in these \n architectural tips and OO ideas. Consequently, although the \n generated content material is extremely advanced, there are remarkably few main \n points. We will choose to proceed with producing the code utilizing this \n structured immediate first, after which conduct a deeper evaluation to determine \n any potential code-level anomalies later.<\/p>\n To date, now we have reached a robust consensus on the intent degree, \n clarifying each the core downside and the decision path. Whereas there \n could also be slight omissions within the particulars, this isn’t a priority; having \n aligned on the general scope with the AI makes native optimizations \n extremely controllable. Now, we transition into the code technology \n part.<\/p>\n<\/section>\n \nStep 5: Generate code<\/h3>\nThis step is extra concerned as we’re producing the product code, \n checks, and our opinions have two different outcomes.<\/p>\n\nGenerate product code<\/h4>\nAs soon as our structured immediate is locked in, use it to generate the \n product code.<\/p>\n\nHow spdd-generate<\/a> works<\/h3>\nThis command reads the REASONS Canvas and generates code job \n by job, following the order outlined in Operations. It strictly \n adheres to the coding requirements in Norms and the constraints in \n Safeguards \u2014 no improvisation, no options past what the spec \n defines.<\/p>\n The core precept: the immediate captures the intent, and the code \n is the implementation of that intent. Generated code should correspond \n one-to-one with this specification.<\/p>\n<\/div>\n Instruction:<\/p>\n \/spdd-generate \n @GGQPA-001-202603191105-[Feat]-multi-plan-billing-model-aware-pricing.md<\/p>\nGenerated artifact: code generated primarily based on the structured \n immediate<\/a>.<\/p>\n Due to the a number of rounds of logical deduction we did earlier \n utilizing structured prompts, we method the code evaluation with a transparent \n focus and set of priorities:<\/p>\n \nStructure: does the code strictly observe our anticipated 3-tier \n structure?<\/li>\n Enterprise logic: does the Service layer implementation completely \n align with our preliminary intent?<\/li>\nScope of change: are the modifications strictly confined to the \n boundaries outlined by the structured immediate, avoiding unrelated \n adjustments or scope creep?<\/li>\n<\/ol>\nOn this particular case, due to the extremely exact context, the \n generated code largely met our expectations, except for just a few \n potential \u201cmagic numbers.\u201d We are going to optimize these out as soon as the \n purposeful verification is full.<\/p>\n The important thing takeaway right here is: don’t be concerned about making errors, and \n do not stress over not catching each single element completely on the \n first strive. So long as we maintain iterating and advancing via the SPDD \n workflow, there are many alternatives to course-correct. Minor \n code smells are wonderful for now\u2014we confirm the core performance first, \n then circle again to optimize.<\/p>\n<\/section>\n\nFunction verification<\/h4>\nThroughout function validation, the SPDD workflow gives the \n \/spdd-api-test<\/code> command to generate purposeful testing \n scripts. <\/p>\n\nHow spdd-api-test<\/a> works<\/h3>\nThis command extracts API endpoint data from the code \n implementation or acceptance standards and generates a cURL-based \n check script. The script features a structured test-case desk \n protecting regular situations, boundary situations, and error \n situations. When executed, it outputs expected-vs-actual comparability \n outcomes.<\/p>\n<\/div>\n Instruction:<\/p>\n \/spdd-api-test<\/p>\nGenerated artifact: the API check script<\/a>.<\/p>\n Guided by the outlined guidelines within the command, the AI generates a \n script that formulates the required check situations utilizing curl instructions. \n We will evaluation these AI-generated situations within the \u201cTEST CASE OVERVIEW\u201d \n part of the script.<\/p>\n <\/p>\nGenerated API Check Script<\/p>\n<\/div>\n Execution: as soon as the script is generated, run it:<\/p>\n sh scripts\/test-api.sh<\/code><\/p>\n Outcome: all purposeful checks handed efficiently.<\/p>\n<\/p>\nAPI Check Outcomes<\/p>\n<\/div>\n<\/section>\n\nCode evaluation & ultimate changes<\/h4>\nDue to the rigorous intent alignment within the first a number of steps, the \n heavy lifting is already finished. At this stage, the remaining points are \n often minor logic discrepancies or surface-level code smells.<\/p>\n To keep up precision in our engineering practices, we categorize \n these ultimate changes into two distinct sorts\u2014primarily based on whether or not they \n change the system’s observable habits\u2014and deal with them utilizing \n totally different methods throughout the SPDD workflow:<\/p>\n<\/p>\nTwo responses to code evaluation adjustments<\/p>\n<\/div>\n<\/section>\n\nLogic corrections (habits adjustments)<\/h4>\nTechnique: replace the immediate first, then generate code. For points \n associated to enterprise guidelines or logic mismatches (which inherently change \n the observable habits of the software program), at all times replace the structured \n immediate to lock within the appropriate intent earlier than touching the code. That is \n an replace or bug repair, not a refactoring.<\/p>\n For example, when persisting modelId<\/code> within the invoice, we \n at the moment permit this subject to be nullable. The underlying cause is the \n want to take care of backward compatibility with historic knowledge, making \n this workaround an inexpensive architectural choice.<\/p>\n<\/p>\nImmediate wants replace<\/p>\n<\/div>\n Nevertheless, there’s an alternate. If the enterprise stakeholders can \n verify what the modelId<\/code> worth needs to be previous to this \n change, we will unify the system’s habits and remove this potential \n technical debt. Let’s assume that, after confirming with the enterprise, \n the modelId<\/code> for all historic payments needs to be set to \n fast-model<\/code>.<\/p>\n With this clear intent, we work together with the AI:<\/p>\n\nHow spdd-prompt-update<\/a> works<\/h3>\nThis command incrementally updates the present Canvas. It \n modifies solely the sections affected by the change and preserves \n all the pieces else. Based mostly on the kind of change \u2014 new requirement, \n architectural adjustment, or constraint change \u2014 it mechanically \n determines which REASONS dimensions want updating.<\/p>\n This differs from \/spdd-sync<\/code>: sync flows from \n code to spec when code has modified; prompt-update flows from \n necessities to spec when necessities have modified.<\/p>\n<\/div>\n Instruction:<\/p>\n\n\/spdd-prompt-update @GGQPA-001-202603191105-[Feat]-multi-plan-billing-model-aware-pricing.md<\/p>\n model_id is a required subject, and its default worth is fast-model. \n Based mostly on this choice, replace the corresponding elements of the structured immediate.<\/p>\n<\/div>\n The AI updates the structured immediate primarily based on this instruction.<\/p>\nUp to date artifact: the up to date structured immediate<\/a>.<\/p>\n As soon as confirmed, use the \/spdd-generate<\/code> command to \n replace the corresponding code primarily based on the newly up to date structured \n immediate:<\/p>\n \/spdd-generate \n @GGQPA-001-202603191105-[Feat]-multi-plan-billing-model-aware-pricing.md<\/p>\n The AI, guided by the principles outlined throughout the \n \/spdd-generate<\/code> command, comprehends the required adjustments \n and performs focused updates completely on the affected \n codebase.<\/p>\n Up to date artifact: the up to date code<\/a>.<\/p>\n It is very important observe that we don’t regenerate your entire \n codebase. We proceed utilizing the present structured immediate and the AI \n handles focused diffs:<\/p>\n \nDetermine the mismatch: discover that the habits of \n modelId<\/code> throughout persistence is inconsistent with the brand new \n enterprise requirement (it have to be obligatory with a default).<\/li>\n Goal the immediate snippet: copy the precise part from the \n structured immediate that defines the outdated logic.<\/li>\n Replace the immediate: paste the extracted snippet into the chat \n alongside the revised enterprise rule, instructing the AI to replace \n the structured immediate first.<\/li>\nGenerate focused code updates: as soon as the immediate displays the brand new \n fact, run \/spdd-generate<\/code> pointing to the up to date file. \n The AI mechanically performs focused diffs completely on the \n affected codebase, quite than regenerating all the pieces from \n scratch.<\/li>\n<\/ol>\n<\/section>\n\nRefactoring (clear code & fashion)<\/h4>\n\n\u201cA change made to the inner construction of software program to make it simpler \n to know and cheaper to switch with out altering its observable \n habits.\u201d<\/p>\n — Martin Fowler<\/p>\n<\/blockquote>\n Technique: refactor the code first, then sync again to the immediate. \n For structural or stylistic points that don’t change observable \n habits, instruct the AI to refactor the code instantly, after which use \n a sync command to replace the immediate documentation.<\/p>\n For instance, the AI-generated BillingServiceImpl<\/code> class \n accommodates some hardcoded magic numbers that have to be extracted into \n significant constants.<\/p>\npersonal int calculateRemainingQuota(String customerId, PricingPlan plan) {\n if (plan.getMonthlyQuota() == null \|\| plan.getMonthlyQuota() == 0<\/span>) {\n return 0<\/span>;\n }\n\n LocalDate currentDate = LocalDate.now(ZoneOffset.UTC);\n LocalDateTime monthStart = currentDate.withDayOfMonth(1<\/span>).atStartOfDay();\n LocalDateTime monthEnd = currentDate.plusMonths(1<\/span>).withDayOfMonth(1<\/span>).atStartOfDay();\n\n Integer currentMonthUsage = billRepository.sumIncludedTokensUsedForMonth(customerId, monthStart, monthEnd);\n return plan.getMonthlyQuota() - currentMonthUsage;\n }\n<\/pre>\nInstruction 1:<\/p>\n @BillingServiceImpl.java Within the calculateRemainingQuota technique, \n there are some magic numbers that have to be processed as \n constants<\/p>\n The AI executes the code refactoring primarily based on this instruction \n (bear in mind the golden rule: at all times refactor in small, incremental \n steps). If the output meets our expectations, we use the \n \/spdd-sync<\/code> command to synchronize these newly up to date \n code particulars again to their corresponding areas throughout the \n structured immediate.<\/p>\n Instruction 2:<\/p>\n\nHow spdd-sync<\/a> works<\/h3>\nThis command compares the present code in opposition to the Canvas \n specification, then synchronizes code-side adjustments (refactoring, \n bug fixes, new parts) again into the Canvas.<\/p>\n The purpose is to maintain the Canvas as an correct design doc \n for the present code, quite than an outdated historic \n file.<\/p>\n<\/div>\n \/spdd-sync<\/p>\n The AI summarizes the adjustments primarily based on the principles outlined within the \n \/spdd-sync<\/code> command. It then follows the structural \n necessities of the REASONS Canvas to write down the detailed code \n description updates again into the corresponding sections of the \n structured immediate.<\/p>\nAs soon as each instructions are executed, we will see all of the immediate and code \n adjustments right here<\/a>.<\/p>\n For any deeper or hidden code smells, merely repeat these steps. \n The golden rule is to at all times maintain the structured immediate synchronized \n along with your newest codebase.<\/p>\n<\/section>\n \nRegression check<\/h4>\nAs soon as all optimizing is full, restart the service and run the \n API check script another time to make sure no core performance was \n damaged in the course of the cleanup.<\/p>\n Outcome: all handed.<\/p>\n<\/p>\nRegression Check Outcomes<\/p>\n<\/div>\n<\/section>\n<\/section>\n\nStep 6: generate unit checks<\/h3>\nUseful testing alone is inadequate for strong validation; it \n acts primarily as an auxiliary test and isn’t factored into code \n protection metrics. The ultimate sign-off on core logic requires \n complete unit checks. At the moment, the SPDD workflow doesn’t have \n devoted testing instructions finalized (these can be launched in \n future iterations). As an interim answer, we make the most of a \n template-driven method to generate structured prompts for unit \n testing.<\/p>\n\nGenerate the preliminary check immediate<\/h4>\nWe start by combining the implementation particulars with our \n standardized testing template to generate a baseline check immediate.<\/p>\n Instruction:<\/p>\nBased mostly on the implementation particulars immediate \n @GGQPA-001-202603191105-[Feat]-multi-plan-billing-model-aware-pricing.md, \n mixed with the template @TEST-SCENARIOS-TEMPLATE.md<\/a>, please \n generate a check immediate file.<\/div>\n<\/section>\n\nDeduplicate and refine situations<\/h4>\nAfter producing the preliminary structured check immediate, a number of the \n proposed check situations have been duplicates of what we already had. To \n deal with this, we continued the dialogue, instructing the AI to \n cross-reference the generated immediate with the present check suite, \n determine the genuinely new situations, and take away any \n redundancies.<\/p>\n Instruction:<\/p>\n @GGQPA-001-202603191105-[Test]-multi-plan-billing-model-aware-pricing.md \n There are checks which can be duplicated with present ones, evaluate the \n related checks that exist, after which solely add checks for brand spanking new \n situations<\/p>\nUp to date artifact: the check structured immediate<\/a>.<\/p>\n<\/section>\n \nGenerate the unit check code<\/h4>\nAs soon as the refined check situations are reviewed and confirmed, use the \n finalized check immediate to drive the precise code technology.<\/p>\n Instruction:<\/p>\n Based mostly on the generated check immediate \n @GGQPA-001-202603191105-[Test]-multi-plan-billing-model-aware-pricing.md, \n please generate the corresponding unit check code.<\/p>\nOutcome: all checks handed. Commit for checks<\/a>.<\/p>\n<\/section>\n<\/section>\n \nWhat this instance delivered<\/h3>\nThis marks the conclusion of a whole SPDD workflow. By means of this \n standardized course of, we efficiently delivered the next key \n outcomes:<\/p>\n\nA enterprise logic implementation with exceptionally excessive intent \n alignment (~99%).<\/li>\n Full engineering transparency, together with a transparent \n understanding of the implementation path, technical choices, and \n accepted trade-offs.<\/li>\n A structured immediate asset tightly synchronized with the present \n codebase, laying a strong basis for future iterations.<\/li>\nCompounding human experience, fostering a steady accumulation \n of developer expertise and psychological fashions as we iterate \n collaboratively with the AI.<\/li>\n<\/ol>\nView the entire code diff for this enhancement<\/a> on GitHub.<\/p>\n We have additionally ready a bonus enhancement function\u2014Enterprise \n Plan Quantity-Based mostly Tiered Billing<\/a>. Should you’re interested by getting \n some hands-on apply, we extremely encourage you to sort out it utilizing the \n SPDD workflow outlined above.<\/p>\n<\/section>\n<\/section>\n \nThree core expertise<\/h2>\nSPDD is a fabric change in how builders construct software program. In our work \n now we have recognized three core expertise that they want so as to do their \n work successfully. These expertise mirror the place the worth of builders \n is shifting in an AI-assisted world.<\/p>\n\n\nAbstraction first<\/a><\/h3>\ndesign earlier than you generate<\/p>\n\nEarlier than producing any code, it is advisable to be clear about what objects \n exist, how they collaborate, and the place the boundaries are. With out that, \n AI typically sprints on implementation particulars whereas the construction falls aside. \n Unclear obligations, duplicated logic, inconsistent interfaces, and \n the fee reveals up later in evaluation and rework.<\/p>\n<\/div>\nlearn extra\u2026<\/a><\/p>\n<\/div>\n \nAlignment<\/a><\/h3>\nlock intent earlier than you write code<\/p>\n\nEarlier than implementation, it is advisable to make \u201cwhat we’ll do \/ what we can’t \n do\u201d express, and agree on the requirements and laborious constraints up entrance. \n In any other case you find yourself with quick output and gradual rework.<\/p>\n<\/div>\nlearn extra\u2026<\/a><\/p>\n<\/div>\n \nIterative Evaluation<\/a><\/h3>\nflip output right into a managed loop<\/p>\n\nYou need AI help to behave like an engineering course of, not a \n one-shot draft. With out a disciplined review-and-iterate loop, groups both \n maintain forcing the mannequin to patch issues till the answer drifts, or they \n restart repeatedly and lose management of value and time.<\/p>\n<\/div>\nlearn extra\u2026<\/a><\/p>\n<\/div>\n<\/div>\n<\/section>\n \nThe place SPDD suits<\/h2>\n\nHealth evaluation<\/h3>\nSPDD is an engineering funding. The desk under charges how nicely it pays off by state of affairs, from extremely really helpful (5 stars) to not appropriate (1 star).<\/p>\n\n\n\n\n\n\n\n\n\n\n\n\n\nRanking<\/th>\n Situation<\/th>\n Notes<\/th>\n<\/tr>\n<\/thead>\n \u2605\u2605\u2605\u2605\u2605<\/td>\n Scaled, standardized supply<\/td>\n Excessive-repeat enterprise logic that wants long-term maintainability (e.g., constructing many comparable APIs, automating core enterprise workflows).<\/td>\n<\/tr>\n \u2605\u2605\u2605\u2605\u2605<\/td>\n Excessive compliance and laborious constraints<\/td>\n Environments the place you need to observe laws, safety requirements, or strict architectural guidelines (e.g., monetary core techniques, multi-channel \/ multi-client deployments).<\/td>\n<\/tr>\n \u2605\u2605\u2605\u2605\u2606<\/td>\n Workforce collaboration and auditability<\/td>\n Multi-person supply the place adjustments have to be absolutely traceable and reviewable end-to-end.<\/td>\n<\/tr>\n \u2605\u2605\u2605\u2605\u2606<\/td>\n Cross-cutting consistency work<\/td>\n Complicated refactors the place logic should keep tightly synchronized throughout a number of microservices or totally different languages.<\/td>\n<\/tr>\n \u2605\u2605\u2606\u2606\u2606<\/td>\n Firefighting hotfixes<\/td>\n \u201cCease the bleeding\u201d manufacturing fixes the place velocity issues greater than architectural self-discipline.<\/td>\n<\/tr>\n \u2605\u2605\u2606\u2606\u2606<\/td>\n Exploratory spikes<\/td>\n When the purpose is to validate an concept rapidly quite than ship production-quality software program, SPDD’s governance overhead will not pay again.<\/td>\n<\/tr>\n \u2605\u2605\u2606\u2606\u2606<\/td>\n One-off scripts<\/td>\n Disposable knowledge cleanup or non permanent scripts the place SPDD’s upfront value is just too excessive relative to the worth.<\/td>\n<\/tr>\n \u2605\u2606\u2606\u2606\u2606<\/td>\n Context black holes<\/td>\n When the area is poorly outlined and enterprise guidelines are unclear, you possibly can’t set significant boundaries for the mannequin.<\/td>\n<\/tr>\n \u2605\u2606\u2606\u2606\u2606<\/td>\n Pure inventive \/ visible work<\/td>\n Duties pushed by style and aesthetics quite than logic (e.g., UI visible exploration, advertising copy).<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/section>\n\nCommerce-offs to contemplate<\/h3>\nReturn on funding<\/b><\/p>\n\n\n\n\n\n\n\n\n\nProfit<\/th>\n Affect<\/th>\n Pace<\/th>\n What you get<\/th>\n<\/tr>\n<\/thead>\n Determinism<\/td>\n Excessive<\/td>\n Quick<\/td>\n Encode logic in a exact spec, which considerably reduces hallucination and \u201cinventive\u201d interpretation.<\/td>\n<\/tr>\n Traceability<\/td>\n Excessive<\/td>\n Quick<\/td>\n Each significant change might be traced again to the structured immediate, closing the audit loop.<\/td>\n<\/tr>\n Quicker opinions<\/td>\n Excessive<\/td>\n Quick-term<\/td>\n Code \u201carrives\u201d nearer to crew requirements, so opinions deal with logic and design, not formatting and cleanup.<\/td>\n<\/tr>\n Explainability<\/td>\n Medium-Excessive<\/td>\n Gradual<\/td>\n Intent and habits are seen on the natural-language degree, decreasing the cognitive load for understanding and upkeep.<\/td>\n<\/tr>\n Safer evolution<\/td>\n Excessive<\/td>\n Lengthy-term<\/td>\n Effectively-defined boundaries and stepwise implementation make focused adjustments lower-risk and simpler to iterate.<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\nUpfront funding<\/b><\/p>\n\n\n\n\n\n\n\nSpace<\/th>\n Barrier<\/th>\n Nature<\/th>\n What it takes<\/th>\n<\/tr>\n<\/thead>\n Mindset shift<\/td>\n Excessive<\/td>\n Ongoing coaching<\/td>\n Groups need to adapt to \u201cdesign first\u201d quite than \u201ccode first.\u201d<\/td>\n<\/tr>\n Senior experience up entrance<\/td>\n Medium-Excessive<\/td>\n Per-feature<\/td>\n Engineers who can translate enterprise guidelines into clear abstractions and design constraints.<\/td>\n<\/tr>\n Automation tooling<\/td>\n Medium<\/td>\n Infrastructure setup<\/td>\n With out automation, SPDD hits a throughput ceiling and struggles to maintain prompts constant. openspdd<\/a> runs the workflow on this article\u2014from evaluation and structured REASONS prompts via code and non-obligatory check help\u2014as repeatable CLI steps, so artifacts keep versioned and reviewable as an alternative of trapped in chat. Bigger organizations should layer a information platform on prime to handle and reuse property at scale.<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/section>\n<\/section>\n\nClosing<\/h2>\nThrough the use of the REASONS Canvas, clarifying intent, establishing the correct \n abstractions, breaking work into concrete duties, and locking in boundaries, \n we give AI a well-defined area to function. Inside that area, SPDD could not \n be the shortest path to \u201cgenerate code rapidly,\u201d but it surely is without doubt one of the most \n dependable methods to ship the correct change with confidence.<\/p>\n It is also honest to say that SPDD shines most in logic-heavy domains. \n In areas pushed by aesthetic judgment, frontend styling, for instance, we’re \n nonetheless exploring engineering patterns that may be as secure as purely logical \n building.<\/p>\n The framework on this article is just the \u201cstrikes.\u201d The true benefit comes \n from sharpening the meta-skills behind it: abstraction and modelling, systematic \n evaluation, and a deep understanding of the enterprise as an entire. These are the \n human strengths that in the end decide how a lot worth we will get from AI.<\/p>\n Within the AI period, software program growth is not a contest of mannequin IQ. It is a \n contest of engineer cognitive bandwidth \u2013 how clearly we will suppose, \n body issues, and make choices.<\/p>\n We’ll shut with a quote that captures the spirit \n of SPDD:<\/p>\n\n\u201cIn science, if you realize what you might be doing, you should not be doing it. \n In engineering, if you do not know what you might be doing, you should not be doing it.\u201d<\/p>\n— Richard W. Hamming<\/a><\/p>\n<\/blockquote>\n<\/section>\n \n<\/div>\n\n\nAcknowledgements<\/h2>\nWe might like to specific our honest due to Martin Fowler. Regardless of a \n busy schedule, he invested deeply on this article \u2014 from sharpening the \n narrative construction and clarifying key ideas, to elevating the visible \n storytelling with improved and new diagrams. His eager eye for element and \n dedication to precision profoundly formed the ultimate outcome.<\/p>\n We’re additionally deeply grateful to Eric (Ke) Zhou, Wei Solar, Sara Michelazzo, \n Rebecca Parsons, Matteo Vaccari, Might (Ping) Xu, Zhi Wang, Feng Chen and Da Cheng for his or her considerate critique and insights. \n Your enter helped us make clear a number of key ideas that underpin the methodology.<\/p>\n We additionally wish to acknowledge early practitioners: Jie Wang, Jian Gao, \n Yixuan Feng, Siyuan Li, Yixuan Li, Biao Tian, Wei Cheng, Qi Huang, and Yulong \n Li. Thanks for validating SPDD in actual initiatives, and to your endurance \n because the method matured. Your frontline suggestions has been foundational to \n making SPDD sensible and strong.<\/p>\n Lastly, within the spirit of working towards what we preach, this text itself \n was formed with the help of giant language fashions \u2014 Claude 4.5 Sonnet, \n Claude 4.6 Opus, Gemini 3.1 Professional, and ChatGPT 5.4. We relied on them for \n prose refinement, structural evaluation, synthesizing strategies, and as \n thought companions for steady studying all through the writing course of. \n Their contributions are a becoming testomony to the very method this \n article describes.<\/p>\n<\/section>\n<\/div>\n\n","protected":false},"excerpt":{"rendered":" As soon as a crew adopts AI coding assistants, the primary beneficial properties present up on the particular person degree: one developer can draft, modify, and refactor code a lot sooner than earlier than. However supply velocity isn’t restricted by typing. Whenever you have a look at the complete supply lifecycle, from necessities via launch, […]<\/p>\n","protected":false},"author":2,"featured_media":14276,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[56],"tags":[237,8862,8861],"class_list":["post-14274","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-software","tag-development","tag-spdd","tag-structuredpromptdriven"],"_links":{"self":[{"href":"https:\/\/techtrendfeed.com\/index.php?rest_route=\/wp\/v2\/posts\/14274","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/techtrendfeed.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/techtrendfeed.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/techtrendfeed.com\/index.php?rest_route=\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/techtrendfeed.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=14274"}],"version-history":[{"count":1,"href":"https:\/\/techtrendfeed.com\/index.php?rest_route=\/wp\/v2\/posts\/14274\/revisions"}],"predecessor-version":[{"id":14275,"href":"https:\/\/techtrendfeed.com\/index.php?rest_route=\/wp\/v2\/posts\/14274\/revisions\/14275"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/techtrendfeed.com\/index.php?rest_route=\/wp\/v2\/media\/14276"}],"wp:attachment":[{"href":"https:\/\/techtrendfeed.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=14274"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/techtrendfeed.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=14274"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/techtrendfeed.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=14274"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}

Enhancing a billing engine with SPDD<\/h2>\n
A sophisticated workflow is obscure within the summary, so
\n now we have ready an instance workflow of enhancing an present software program
\n system. This technique, and its enhancement, are neccessarily small so as
\n to be understandable inside a tutorial article. That mentioned the enhancement
\n instance is a full end-to-end instance: from creating preliminary
\n necessities, to analyzing enterprise necessities, to producing and reviewing a structured immediate,
\n to producing and verifying code, to ultimate cleanup and testing.<\/p>\n
You possibly can observe together with this instance by putting in
\n openspdd<\/a> in your individual surroundings.<\/p>\n
\n
The present system<\/h3>\n
The present system is an easy billing engine that calculates payments
\n for utilizing a large-language mannequin. It accepts a file that captures how
\n many tokens are utilized in a session and calculates a invoice.<\/p>\n
The whole codebase for this preliminary model is
\n accessible on GitHub<\/a>.
\n The repository contains the preliminary necessities story<\/a>
\n and all of the SPDD artifacts used to generate it<\/a>.
\n For brevity, we do not describe that preliminary technology right here, but it surely follows primarily the identical steps as that for the enhancement.
\n We deal with describing the enhancement as a result of most work on a system are enhancements.<\/p>\n<\/section>\n
\n
The enhancement<\/h3>\n
Pushed by evolving enterprise necessities and direct consumer suggestions, we’re enhancing the billing engine to transition from a static pricing mannequin to a extra refined,
\n versatile infrastructure. This replace goals to help various subscription methods and variable, model-specific pricing via the next key adjustments:<\/p>\n
\n
API enhancement: replace the present `POST \/api\/utilization<\/code> \n endpoint to just accept a brand new, required modelId<\/code> parameter \n (e.g., \u201cfast-model\u201d, \u201creasoning-model\u201d).<\/li>\n`
`Mannequin-aware pricing: shift from a single world fee to dynamic \n pricing, the place prices fluctuate relying on the precise AI mannequin \n invoked.<\/li>\n`
`Multi-plan billing logic: introduce distinct billing behaviors \n primarily based on the client’s subscription tier:<\/li>\n`\nCustomary plan (optimized): retains the worldwide month-to-month quota, \n however any overage utilization is now calculated utilizing model-specific \n charges.<\/li>\n Premium plan (new): operates with out a quota restrict. It \n introduces cut up billing, the place immediate tokens and completion \n tokens are charged individually at totally different charges relying \n on the mannequin used.<\/li>\n<\/ul>\nArchitectural scalability: implement an extensible design sample \n (similar to Technique or Manufacturing facility) to cleanly isolate the calculation \n formulation for various plans, making certain the system can simply \n accommodate future pricing fashions.<\/li>\n<\/ul>\nSince this new part encompasses each enterprise necessities and technical particulars, \n it’s sometimes accomplished collaboratively via a pairing session between a PO (or BA) and a developer.<\/i><\/p>\n<\/section>\n\nStep 1: Create preliminary necessities<\/h3>\nTo kick off the method rapidly, we will use the \/spdd-story<\/code> \n command to generate a consumer story instantly primarily based on the enhancement. Usually, consumer tales are offered by the PO or BA. \n Nevertheless, in our workflow, we will remodel tales of any type right into a constant format and dimension. \n So long as there’s shared alignment on the ultimate acceptance standards, this step might be carried out by a PO, BA, or developer, relying on the crew’s versatile division of labor.<\/p>\n Instruction:<\/p>\n\nHow spdd-story<\/a> works<\/h3>\nThis command breaks a big requirement into unbiased, \n deliverable consumer tales following the INVEST precept (1\u20135 days \n of labor every). Every story contains acceptance standards written in \n enterprise language, able to function enter for \n \/spdd-analysis<\/code>.<\/p>\n Its goal is to make giant necessities manageable and to make sure a standardized, \n predictable format for the subsequent steps.<\/p>\n<\/div>\n The AI analyzed the enhancement description and cut up it into two \n consumer tales:<\/p>\n The auto-generated tales are detailed sufficient to function a baseline \n for a proper venture. For this walkthrough we consolidate them right into a \n single simplified story so the instance stays self-contained.<\/p>\n Instruction:<\/p>\n\nConsolidate the next two consumer tales right into a single, simplified \n story: @[User-story-1-1-initial]Multi-Plan-Billing-Basis-&-Customary-Plan-Mannequin-Conscious-Pricing.md @[User-story-1-2-initial]Premium-Plan-Break up-Price-Billing.md<\/p>\n Necessities: 1. Merge each plans (Customary and Premium) into one coherent story. 2. Hold solely the sections: Background, Enterprise Worth, Scope In, Scope Out, and Acceptance Standards. 3. Strip implementation-level element \u2014 deal with what the system ought to do, not how. 4. Acceptance Standards should use Given\/When\/Then format with concrete numeric examples. 5. Hold the outcome concise \u2014 no multiple web page. 6. Solely maintain three high-level ACs.\n <\/p>\n<\/div>\n Directions of this sort hardly ever produce similar textual content on each run \u2014 \n fashions and sampling introduce small variations \u2014 so we nonetheless count on to \n evaluation and tweak the output earlier than treating it as ultimate. The mixed \n story under is the model we refined for this walkthrough: a intentionally \n simplified consolidation of the 2 preliminary tales.<\/p>\n<\/section>\n\nStep 2: Make clear evaluation<\/h3>\nEarlier than leaping into implementation, the developer opinions the consumer \n story to construct a shared understanding of what it means in apply. If \n there are apparent business-level points, that is the purpose to align with \n the BA or PO. On this case the story is obvious sufficient, so we transfer \n straight to breaking it down alongside three dimensions: core logic, scope \n boundaries, and definition of finished.<\/p>\n Core logic<\/b><\/p>\n One new required subject on the API: modelId<\/code>. The client \n now tells us which AI mannequin they used \u2014 that is the important thing that unlocks the \n proper value.<\/p>\n\nCustomary Plan:<\/i> Buyer has a month-to-month token quota. Utilization \n inside quota is free. Overage is charged at a model-specific fee \n (e.g., fast-model $0.01\/1K vs. reasoning-model $0.03\/1K). Current \n quota logic stays; solely the speed lookup adjustments.<\/li>\n Premium Plan:<\/i> No quota. Each token is billed from the \n first one. Immediate tokens and completion tokens are charged individually, \n every at a model-specific fee. Invoice = immediate cost + completion \n cost. This plan is totally new.<\/li>\nRouting:<\/i> The system determines the client’s plan and \n dispatches to the matching billing components. The design have to be simple to \n lengthen \u2014 Enterprise plans (Story 2) are subsequent.<\/li>\n<\/ul>\nScope boundaries<\/b><\/p>\n We’re solely calculating the present invoice. We’re NOT constructing buyer \n CRUD, NOT querying historic payments, NOT managing subscriptions, and NOT \n including\/eradicating fashions.<\/p>\n Definition of finished<\/b><\/p>\n The next situations restate the story’s acceptance standards \n with the implementation element the crew must confirm. The fourth \n merchandise (Response format) shouldn’t be a brand new enterprise AC \u2014 it captures the \n non-functional contract the developer provides to make the factors \n testable end-to-end.<\/p>\n\nValidation:<\/i> Lacking modelId<\/code> \u2192 HTTP 400. \n Unknown buyer \u2192 HTTP 404. Damaging tokens \u2192 HTTP 400. All \n present validations stay intact.<\/li>\n Customary Plan billing:<\/i> A buyer with a 100K quota and \n 90K already used submits 30K tokens for fast-model ($0.01\/1K). \n Anticipated outcome: 10K coated by quota, 20K overage, cost $0.20. \n The identical request with reasoning-model ($0.03\/1K) yields $0.60 \u2014 \n similar quota logic, totally different fee.<\/li>\n Premium Plan billing:<\/i> A buyer submits 10K immediate \n tokens + 20K completion tokens for reasoning-model (immediate $0.03\/1K, \n completion $0.06\/1K). Anticipated outcome: $0.30 + $1.20 = $1.50. No \n quota, no overage \u2014 immediate and completion are billed individually.<\/li>\nResponse format:<\/i> HTTP 201 returning invoice ID, buyer ID, \n token counts, timestamp, modelId<\/code>, and a plan-appropriate \n cost breakdown.<\/li>\n<\/ul>\nIf all these situations go, we have conquered this story.<\/p>\n<\/section>\n\nStep 3: Generate evaluation context<\/h3>\nWith the necessities and scope clarified, we use the \n \/spdd-analysis<\/code> command. By feeding it the enterprise \n necessities, we instruct the AI to generate a complete evaluation \n context.<\/p>\n \nHow spdd-analysis<\/a> works<\/h3>\nThis command extracts area key phrases from the enterprise \n necessities (e.g. \u201cbilling\u201d, \u201cquota\u201d, \u201cplan\u201d) and makes use of them to \n scan solely the related elements of the codebase \u2014 not all of it. It \n identifies present ideas, new ideas, key enterprise guidelines, \n and technical dangers.<\/p>\n The output is a context-rich doc protecting area idea \n recognition, strategic course, and danger evaluation. It serves as \n enter for the subsequent step: producing the REASONS Canvas.<\/p>\n<\/div>\n Instruction:<\/p>\n \/spdd-analysis \n @[User-story-1]Multi-Plan-Billing-Basis-&-Mannequin-Conscious-Pricing.md<\/p>\nGenerated artifact: the preliminary evaluation context doc<\/a>.<\/p>\n This command produces a strategic-level evaluation grounded in precise \n codebase exploration. The output focuses totally on the \u201cwhat\u201d and \n \u201cwhy,\u201d intentionally avoiding granular implementation particulars at this \n stage. It sometimes covers:<\/p>\n \nArea ideas: present vs. new, relationships, enterprise \n guidelines<\/li>\n Strategic method: answer course, design choices, \n trade-offs<\/li>\nDangers & gaps: ambiguities, edge circumstances, technical dangers, \n acceptance-criteria protection<\/li>\n<\/ul>\n\nEvaluation and refine the evaluation context<\/h4>\nWith our personal understanding of the enterprise necessities in thoughts, we \n evaluation the generated evaluation doc\u2014specializing in the areas \n highlighted within the alignment<\/a> talent. This evaluation serves \n two functions: confirming that our understanding aligns with the AI’s \n interpretation, and discovering edge circumstances or boundary situations the AI \n may floor that we hadn’t thought of.<\/p>\n On this particular occasion, the evaluation targeted on a number of vital \n areas:<\/p>\n \nWhether or not the Technique Sample was appropriately thought of.<\/li>\n Adherence to the OOP ideas established within the present system, \n particularly ISP and SRP.<\/li>\n The validity of the proposed technique for including new fields.<\/li>\n Figuring out edge circumstances not beforehand anticipated.<\/li>\nUncovering potential technical dangers.<\/li>\n<\/ul>\nUpon finishing the evaluation, the AI’s evaluation largely aligned with \n our architectural intent; in reality, its issues have been much more \n complete than ours in sure areas.<\/p>\n<\/p>\nEdge circumstances and dangers from the \n evaluation doc<\/a><\/p>\n<\/div>\n To be clear, at this stage we solely possess a high-level \n conceptual alignment. Whereas we will rapidly envision the implementation \n for areas the place now we have prior expertise, we can’t utterly map out \n all of the granular technical particulars for the unfamiliar elements proper \n now.<\/p>\n Nevertheless, that’s completely wonderful. The overarching course is aligned. \n We will proceed to the subsequent step: observing how the AI \u201csimulates\u201d the \n concrete implementation particulars inside our established framework and \n context. As soon as now we have these tangible particulars, we will uncover deeper, \n hidden points and make knowledgeable trade-offs primarily based on the precise \n state of affairs\u2014adopting the approaches the place the advantages outweigh the \n drawbacks, and discarding the remainder.<\/p>\n Determination: settle for the evaluation as-is and proceed.<\/p>\n<\/section>\n<\/section>\n \nStep 4: Generate structured immediate<\/h3>\n\nHow spdd-reasons-canvas<\/a> works<\/h3>\nThis command reads enterprise context (the output of \n \/spdd-analysis<\/code>, or a direct necessities description) \n and combines it with the present state of the codebase. It then \n generates a design specification throughout all seven REASONS \n dimensions \u2014 from \u201cwhy are we doing this\u201d to \u201cwhat should we not \n do.\u201d<\/p>\n The output is an executable blueprint. The Operations \n part is exact all the way down to technique signatures, parameter sorts, and \n execution steps.<\/p>\n<\/div>\n Instruction:<\/p>\n \/spdd-reasons-canvas \n @GGQPA-001-202603191100-[Analysis]-multi-plan-billing-model-aware-pricing.md<\/p>\nGenerated artifact: the preliminary structured immediate<\/a>.<\/p>\n By this level, we have already gone via high-level technique throughout \n the evaluation part\u2014so when reviewing the structured immediate, we’re not \n ranging from scratch. As a substitute, we’re inspecting how the AI has \n translated our shared understanding into the REASONS Canvas construction: \n from technique to abstraction to concrete particulars.<\/p>\n Consider it as a development: the evaluation part gave us strategic \n readability; now we’re checking whether or not that readability has been faithfully \n carried via into the architectural abstractions and implementation \n specifics. That is intent alignment at a deeper degree\u2014making certain that \n earlier than any code is generated, the AI has successfully \u201csimulated\u201d the \n whole answer inside our outlined framework. We get to evaluation from a \n world perspective quite than getting misplaced in particulars from the \n begin.<\/p>\n Focus the evaluation on the areas highlighted within the abstraction-first<\/a> talent. On this case, \n this foundational context is already embedded within the codebase and the earlier \n structured immediate<\/a>. Consequently, when producing the structured \n immediate for this iteration, the AI naturally elements in these \n architectural tips and OO ideas. Consequently, although the \n generated content material is extremely advanced, there are remarkably few main \n points. We will choose to proceed with producing the code utilizing this \n structured immediate first, after which conduct a deeper evaluation to determine \n any potential code-level anomalies later.<\/p>\n To date, now we have reached a robust consensus on the intent degree, \n clarifying each the core downside and the decision path. Whereas there \n could also be slight omissions within the particulars, this isn’t a priority; having \n aligned on the general scope with the AI makes native optimizations \n extremely controllable. Now, we transition into the code technology \n part.<\/p>\n<\/section>\n \nStep 5: Generate code<\/h3>\nThis step is extra concerned as we’re producing the product code, \n checks, and our opinions have two different outcomes.<\/p>\n\nGenerate product code<\/h4>\nAs soon as our structured immediate is locked in, use it to generate the \n product code.<\/p>\n\nHow spdd-generate<\/a> works<\/h3>\nThis command reads the REASONS Canvas and generates code job \n by job, following the order outlined in Operations. It strictly \n adheres to the coding requirements in Norms and the constraints in \n Safeguards \u2014 no improvisation, no options past what the spec \n defines.<\/p>\n The core precept: the immediate captures the intent, and the code \n is the implementation of that intent. Generated code should correspond \n one-to-one with this specification.<\/p>\n<\/div>\n Instruction:<\/p>\n \/spdd-generate \n @GGQPA-001-202603191105-[Feat]-multi-plan-billing-model-aware-pricing.md<\/p>\nGenerated artifact: code generated primarily based on the structured \n immediate<\/a>.<\/p>\n Due to the a number of rounds of logical deduction we did earlier \n utilizing structured prompts, we method the code evaluation with a transparent \n focus and set of priorities:<\/p>\n \nStructure: does the code strictly observe our anticipated 3-tier \n structure?<\/li>\n Enterprise logic: does the Service layer implementation completely \n align with our preliminary intent?<\/li>\nScope of change: are the modifications strictly confined to the \n boundaries outlined by the structured immediate, avoiding unrelated \n adjustments or scope creep?<\/li>\n<\/ol>\nOn this particular case, due to the extremely exact context, the \n generated code largely met our expectations, except for just a few \n potential \u201cmagic numbers.\u201d We are going to optimize these out as soon as the \n purposeful verification is full.<\/p>\n The important thing takeaway right here is: don’t be concerned about making errors, and \n do not stress over not catching each single element completely on the \n first strive. So long as we maintain iterating and advancing via the SPDD \n workflow, there are many alternatives to course-correct. Minor \n code smells are wonderful for now\u2014we confirm the core performance first, \n then circle again to optimize.<\/p>\n<\/section>\n\nFunction verification<\/h4>\nThroughout function validation, the SPDD workflow gives the \n \/spdd-api-test<\/code> command to generate purposeful testing \n scripts. <\/p>\n\nHow spdd-api-test<\/a> works<\/h3>\nThis command extracts API endpoint data from the code \n implementation or acceptance standards and generates a cURL-based \n check script. The script features a structured test-case desk \n protecting regular situations, boundary situations, and error \n situations. When executed, it outputs expected-vs-actual comparability \n outcomes.<\/p>\n<\/div>\n Instruction:<\/p>\n \/spdd-api-test<\/p>\nGenerated artifact: the API check script<\/a>.<\/p>\n Guided by the outlined guidelines within the command, the AI generates a \n script that formulates the required check situations utilizing curl instructions. \n We will evaluation these AI-generated situations within the \u201cTEST CASE OVERVIEW\u201d \n part of the script.<\/p>\n <\/p>\nGenerated API Check Script<\/p>\n<\/div>\n Execution: as soon as the script is generated, run it:<\/p>\n sh scripts\/test-api.sh<\/code><\/p>\n Outcome: all purposeful checks handed efficiently.<\/p>\n<\/p>\nAPI Check Outcomes<\/p>\n<\/div>\n<\/section>\n\nCode evaluation & ultimate changes<\/h4>\nDue to the rigorous intent alignment within the first a number of steps, the \n heavy lifting is already finished. At this stage, the remaining points are \n often minor logic discrepancies or surface-level code smells.<\/p>\n To keep up precision in our engineering practices, we categorize \n these ultimate changes into two distinct sorts\u2014primarily based on whether or not they \n change the system’s observable habits\u2014and deal with them utilizing \n totally different methods throughout the SPDD workflow:<\/p>\n<\/p>\nTwo responses to code evaluation adjustments<\/p>\n<\/div>\n<\/section>\n\nLogic corrections (habits adjustments)<\/h4>\nTechnique: replace the immediate first, then generate code. For points \n associated to enterprise guidelines or logic mismatches (which inherently change \n the observable habits of the software program), at all times replace the structured \n immediate to lock within the appropriate intent earlier than touching the code. That is \n an replace or bug repair, not a refactoring.<\/p>\n For example, when persisting modelId<\/code> within the invoice, we \n at the moment permit this subject to be nullable. The underlying cause is the \n want to take care of backward compatibility with historic knowledge, making \n this workaround an inexpensive architectural choice.<\/p>\n<\/p>\nImmediate wants replace<\/p>\n<\/div>\n Nevertheless, there’s an alternate. If the enterprise stakeholders can \n verify what the modelId<\/code> worth needs to be previous to this \n change, we will unify the system’s habits and remove this potential \n technical debt. Let’s assume that, after confirming with the enterprise, \n the modelId<\/code> for all historic payments needs to be set to \n fast-model<\/code>.<\/p>\n With this clear intent, we work together with the AI:<\/p>\n\nHow spdd-prompt-update<\/a> works<\/h3>\nThis command incrementally updates the present Canvas. It \n modifies solely the sections affected by the change and preserves \n all the pieces else. Based mostly on the kind of change \u2014 new requirement, \n architectural adjustment, or constraint change \u2014 it mechanically \n determines which REASONS dimensions want updating.<\/p>\n This differs from \/spdd-sync<\/code>: sync flows from \n code to spec when code has modified; prompt-update flows from \n necessities to spec when necessities have modified.<\/p>\n<\/div>\n Instruction:<\/p>\n\n\/spdd-prompt-update @GGQPA-001-202603191105-[Feat]-multi-plan-billing-model-aware-pricing.md<\/p>\n model_id is a required subject, and its default worth is fast-model. \n Based mostly on this choice, replace the corresponding elements of the structured immediate.<\/p>\n<\/div>\n The AI updates the structured immediate primarily based on this instruction.<\/p>\nUp to date artifact: the up to date structured immediate<\/a>.<\/p>\n As soon as confirmed, use the \/spdd-generate<\/code> command to \n replace the corresponding code primarily based on the newly up to date structured \n immediate:<\/p>\n \/spdd-generate \n @GGQPA-001-202603191105-[Feat]-multi-plan-billing-model-aware-pricing.md<\/p>\n The AI, guided by the principles outlined throughout the \n \/spdd-generate<\/code> command, comprehends the required adjustments \n and performs focused updates completely on the affected \n codebase.<\/p>\n Up to date artifact: the up to date code<\/a>.<\/p>\n It is very important observe that we don’t regenerate your entire \n codebase. We proceed utilizing the present structured immediate and the AI \n handles focused diffs:<\/p>\n \nDetermine the mismatch: discover that the habits of \n modelId<\/code> throughout persistence is inconsistent with the brand new \n enterprise requirement (it have to be obligatory with a default).<\/li>\n Goal the immediate snippet: copy the precise part from the \n structured immediate that defines the outdated logic.<\/li>\n Replace the immediate: paste the extracted snippet into the chat \n alongside the revised enterprise rule, instructing the AI to replace \n the structured immediate first.<\/li>\nGenerate focused code updates: as soon as the immediate displays the brand new \n fact, run \/spdd-generate<\/code> pointing to the up to date file. \n The AI mechanically performs focused diffs completely on the \n affected codebase, quite than regenerating all the pieces from \n scratch.<\/li>\n<\/ol>\n<\/section>\n\nRefactoring (clear code & fashion)<\/h4>\n\n\u201cA change made to the inner construction of software program to make it simpler \n to know and cheaper to switch with out altering its observable \n habits.\u201d<\/p>\n — Martin Fowler<\/p>\n<\/blockquote>\n Technique: refactor the code first, then sync again to the immediate. \n For structural or stylistic points that don’t change observable \n habits, instruct the AI to refactor the code instantly, after which use \n a sync command to replace the immediate documentation.<\/p>\n For instance, the AI-generated BillingServiceImpl<\/code> class \n accommodates some hardcoded magic numbers that have to be extracted into \n significant constants.<\/p>\npersonal int calculateRemainingQuota(String customerId, PricingPlan plan) {\n if (plan.getMonthlyQuota() == null || plan.getMonthlyQuota() == 0<\/span>) {\n return 0<\/span>;\n }\n\n LocalDate currentDate = LocalDate.now(ZoneOffset.UTC);\n LocalDateTime monthStart = currentDate.withDayOfMonth(1<\/span>).atStartOfDay();\n LocalDateTime monthEnd = currentDate.plusMonths(1<\/span>).withDayOfMonth(1<\/span>).atStartOfDay();\n\n Integer currentMonthUsage = billRepository.sumIncludedTokensUsedForMonth(customerId, monthStart, monthEnd);\n return plan.getMonthlyQuota() - currentMonthUsage;\n }\n<\/pre>\nInstruction 1:<\/p>\n @BillingServiceImpl.java Within the calculateRemainingQuota technique, \n there are some magic numbers that have to be processed as \n constants<\/p>\n The AI executes the code refactoring primarily based on this instruction \n (bear in mind the golden rule: at all times refactor in small, incremental \n steps). If the output meets our expectations, we use the \n \/spdd-sync<\/code> command to synchronize these newly up to date \n code particulars again to their corresponding areas throughout the \n structured immediate.<\/p>\n Instruction 2:<\/p>\n\nHow spdd-sync<\/a> works<\/h3>\nThis command compares the present code in opposition to the Canvas \n specification, then synchronizes code-side adjustments (refactoring, \n bug fixes, new parts) again into the Canvas.<\/p>\n The purpose is to maintain the Canvas as an correct design doc \n for the present code, quite than an outdated historic \n file.<\/p>\n<\/div>\n \/spdd-sync<\/p>\n The AI summarizes the adjustments primarily based on the principles outlined within the \n \/spdd-sync<\/code> command. It then follows the structural \n necessities of the REASONS Canvas to write down the detailed code \n description updates again into the corresponding sections of the \n structured immediate.<\/p>\nAs soon as each instructions are executed, we will see all of the immediate and code \n adjustments right here<\/a>.<\/p>\n For any deeper or hidden code smells, merely repeat these steps. \n The golden rule is to at all times maintain the structured immediate synchronized \n along with your newest codebase.<\/p>\n<\/section>\n \nRegression check<\/h4>\nAs soon as all optimizing is full, restart the service and run the \n API check script another time to make sure no core performance was \n damaged in the course of the cleanup.<\/p>\n Outcome: all handed.<\/p>\n<\/p>\nRegression Check Outcomes<\/p>\n<\/div>\n<\/section>\n<\/section>\n\nStep 6: generate unit checks<\/h3>\nUseful testing alone is inadequate for strong validation; it \n acts primarily as an auxiliary test and isn’t factored into code \n protection metrics. The ultimate sign-off on core logic requires \n complete unit checks. At the moment, the SPDD workflow doesn’t have \n devoted testing instructions finalized (these can be launched in \n future iterations). As an interim answer, we make the most of a \n template-driven method to generate structured prompts for unit \n testing.<\/p>\n\nGenerate the preliminary check immediate<\/h4>\nWe start by combining the implementation particulars with our \n standardized testing template to generate a baseline check immediate.<\/p>\n Instruction:<\/p>\nBased mostly on the implementation particulars immediate \n @GGQPA-001-202603191105-[Feat]-multi-plan-billing-model-aware-pricing.md, \n mixed with the template @TEST-SCENARIOS-TEMPLATE.md<\/a>, please \n generate a check immediate file.<\/div>\n<\/section>\n\nDeduplicate and refine situations<\/h4>\nAfter producing the preliminary structured check immediate, a number of the \n proposed check situations have been duplicates of what we already had. To \n deal with this, we continued the dialogue, instructing the AI to \n cross-reference the generated immediate with the present check suite, \n determine the genuinely new situations, and take away any \n redundancies.<\/p>\n Instruction:<\/p>\n @GGQPA-001-202603191105-[Test]-multi-plan-billing-model-aware-pricing.md \n There are checks which can be duplicated with present ones, evaluate the \n related checks that exist, after which solely add checks for brand spanking new \n situations<\/p>\nUp to date artifact: the check structured immediate<\/a>.<\/p>\n<\/section>\n \nGenerate the unit check code<\/h4>\nAs soon as the refined check situations are reviewed and confirmed, use the \n finalized check immediate to drive the precise code technology.<\/p>\n Instruction:<\/p>\n Based mostly on the generated check immediate \n @GGQPA-001-202603191105-[Test]-multi-plan-billing-model-aware-pricing.md, \n please generate the corresponding unit check code.<\/p>\nOutcome: all checks handed. Commit for checks<\/a>.<\/p>\n<\/section>\n<\/section>\n \nWhat this instance delivered<\/h3>\nThis marks the conclusion of a whole SPDD workflow. By means of this \n standardized course of, we efficiently delivered the next key \n outcomes:<\/p>\n\nA enterprise logic implementation with exceptionally excessive intent \n alignment (~99%).<\/li>\n Full engineering transparency, together with a transparent \n understanding of the implementation path, technical choices, and \n accepted trade-offs.<\/li>\n A structured immediate asset tightly synchronized with the present \n codebase, laying a strong basis for future iterations.<\/li>\nCompounding human experience, fostering a steady accumulation \n of developer expertise and psychological fashions as we iterate \n collaboratively with the AI.<\/li>\n<\/ol>\nView the entire code diff for this enhancement<\/a> on GitHub.<\/p>\n We have additionally ready a bonus enhancement function\u2014Enterprise \n Plan Quantity-Based mostly Tiered Billing<\/a>. Should you’re interested by getting \n some hands-on apply, we extremely encourage you to sort out it utilizing the \n SPDD workflow outlined above.<\/p>\n<\/section>\n<\/section>\n \nThree core expertise<\/h2>\nSPDD is a fabric change in how builders construct software program. In our work \n now we have recognized three core expertise that they want so as to do their \n work successfully. These expertise mirror the place the worth of builders \n is shifting in an AI-assisted world.<\/p>\n\n\nAbstraction first<\/a><\/h3>\ndesign earlier than you generate<\/p>\n\nEarlier than producing any code, it is advisable to be clear about what objects \n exist, how they collaborate, and the place the boundaries are. With out that, \n AI typically sprints on implementation particulars whereas the construction falls aside. \n Unclear obligations, duplicated logic, inconsistent interfaces, and \n the fee reveals up later in evaluation and rework.<\/p>\n<\/div>\nlearn extra\u2026<\/a><\/p>\n<\/div>\n \nAlignment<\/a><\/h3>\nlock intent earlier than you write code<\/p>\n\nEarlier than implementation, it is advisable to make \u201cwhat we’ll do \/ what we can’t \n do\u201d express, and agree on the requirements and laborious constraints up entrance. \n In any other case you find yourself with quick output and gradual rework.<\/p>\n<\/div>\nlearn extra\u2026<\/a><\/p>\n<\/div>\n \nIterative Evaluation<\/a><\/h3>\nflip output right into a managed loop<\/p>\n\nYou need AI help to behave like an engineering course of, not a \n one-shot draft. With out a disciplined review-and-iterate loop, groups both \n maintain forcing the mannequin to patch issues till the answer drifts, or they \n restart repeatedly and lose management of value and time.<\/p>\n<\/div>\nlearn extra\u2026<\/a><\/p>\n<\/div>\n<\/div>\n<\/section>\n \nThe place SPDD suits<\/h2>\n\nHealth evaluation<\/h3>\nSPDD is an engineering funding. The desk under charges how nicely it pays off by state of affairs, from extremely really helpful (5 stars) to not appropriate (1 star).<\/p>\n\n\n\n\n\n\n\n\n\n\n\n\n\nRanking<\/th>\n Situation<\/th>\n Notes<\/th>\n<\/tr>\n<\/thead>\n \u2605\u2605\u2605\u2605\u2605<\/td>\n Scaled, standardized supply<\/td>\n Excessive-repeat enterprise logic that wants long-term maintainability (e.g., constructing many comparable APIs, automating core enterprise workflows).<\/td>\n<\/tr>\n \u2605\u2605\u2605\u2605\u2605<\/td>\n Excessive compliance and laborious constraints<\/td>\n Environments the place you need to observe laws, safety requirements, or strict architectural guidelines (e.g., monetary core techniques, multi-channel \/ multi-client deployments).<\/td>\n<\/tr>\n \u2605\u2605\u2605\u2605\u2606<\/td>\n Workforce collaboration and auditability<\/td>\n Multi-person supply the place adjustments have to be absolutely traceable and reviewable end-to-end.<\/td>\n<\/tr>\n \u2605\u2605\u2605\u2605\u2606<\/td>\n Cross-cutting consistency work<\/td>\n Complicated refactors the place logic should keep tightly synchronized throughout a number of microservices or totally different languages.<\/td>\n<\/tr>\n \u2605\u2605\u2606\u2606\u2606<\/td>\n Firefighting hotfixes<\/td>\n \u201cCease the bleeding\u201d manufacturing fixes the place velocity issues greater than architectural self-discipline.<\/td>\n<\/tr>\n \u2605\u2605\u2606\u2606\u2606<\/td>\n Exploratory spikes<\/td>\n When the purpose is to validate an concept rapidly quite than ship production-quality software program, SPDD’s governance overhead will not pay again.<\/td>\n<\/tr>\n \u2605\u2605\u2606\u2606\u2606<\/td>\n One-off scripts<\/td>\n Disposable knowledge cleanup or non permanent scripts the place SPDD’s upfront value is just too excessive relative to the worth.<\/td>\n<\/tr>\n \u2605\u2606\u2606\u2606\u2606<\/td>\n Context black holes<\/td>\n When the area is poorly outlined and enterprise guidelines are unclear, you possibly can’t set significant boundaries for the mannequin.<\/td>\n<\/tr>\n \u2605\u2606\u2606\u2606\u2606<\/td>\n Pure inventive \/ visible work<\/td>\n Duties pushed by style and aesthetics quite than logic (e.g., UI visible exploration, advertising copy).<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/section>\n\nCommerce-offs to contemplate<\/h3>\nReturn on funding<\/b><\/p>\n\n\n\n\n\n\n\n\n\nProfit<\/th>\n Affect<\/th>\n Pace<\/th>\n What you get<\/th>\n<\/tr>\n<\/thead>\n Determinism<\/td>\n Excessive<\/td>\n Quick<\/td>\n Encode logic in a exact spec, which considerably reduces hallucination and \u201cinventive\u201d interpretation.<\/td>\n<\/tr>\n Traceability<\/td>\n Excessive<\/td>\n Quick<\/td>\n Each significant change might be traced again to the structured immediate, closing the audit loop.<\/td>\n<\/tr>\n Quicker opinions<\/td>\n Excessive<\/td>\n Quick-term<\/td>\n Code \u201carrives\u201d nearer to crew requirements, so opinions deal with logic and design, not formatting and cleanup.<\/td>\n<\/tr>\n Explainability<\/td>\n Medium-Excessive<\/td>\n Gradual<\/td>\n Intent and habits are seen on the natural-language degree, decreasing the cognitive load for understanding and upkeep.<\/td>\n<\/tr>\n Safer evolution<\/td>\n Excessive<\/td>\n Lengthy-term<\/td>\n Effectively-defined boundaries and stepwise implementation make focused adjustments lower-risk and simpler to iterate.<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\nUpfront funding<\/b><\/p>\n\n\n\n\n\n\n\nSpace<\/th>\n Barrier<\/th>\n Nature<\/th>\n What it takes<\/th>\n<\/tr>\n<\/thead>\n Mindset shift<\/td>\n Excessive<\/td>\n Ongoing coaching<\/td>\n Groups need to adapt to \u201cdesign first\u201d quite than \u201ccode first.\u201d<\/td>\n<\/tr>\n Senior experience up entrance<\/td>\n Medium-Excessive<\/td>\n Per-feature<\/td>\n Engineers who can translate enterprise guidelines into clear abstractions and design constraints.<\/td>\n<\/tr>\n Automation tooling<\/td>\n Medium<\/td>\n Infrastructure setup<\/td>\n With out automation, SPDD hits a throughput ceiling and struggles to maintain prompts constant. openspdd<\/a> runs the workflow on this article\u2014from evaluation and structured REASONS prompts via code and non-obligatory check help\u2014as repeatable CLI steps, so artifacts keep versioned and reviewable as an alternative of trapped in chat. Bigger organizations should layer a information platform on prime to handle and reuse property at scale.<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/section>\n<\/section>\n\nClosing<\/h2>\nThrough the use of the REASONS Canvas, clarifying intent, establishing the correct \n abstractions, breaking work into concrete duties, and locking in boundaries, \n we give AI a well-defined area to function. Inside that area, SPDD could not \n be the shortest path to \u201cgenerate code rapidly,\u201d but it surely is without doubt one of the most \n dependable methods to ship the correct change with confidence.<\/p>\n It is also honest to say that SPDD shines most in logic-heavy domains. \n In areas pushed by aesthetic judgment, frontend styling, for instance, we’re \n nonetheless exploring engineering patterns that may be as secure as purely logical \n building.<\/p>\n The framework on this article is just the \u201cstrikes.\u201d The true benefit comes \n from sharpening the meta-skills behind it: abstraction and modelling, systematic \n evaluation, and a deep understanding of the enterprise as an entire. These are the \n human strengths that in the end decide how a lot worth we will get from AI.<\/p>\n Within the AI period, software program growth is not a contest of mannequin IQ. It is a \n contest of engineer cognitive bandwidth \u2013 how clearly we will suppose, \n body issues, and make choices.<\/p>\n We’ll shut with a quote that captures the spirit \n of SPDD:<\/p>\n\n\u201cIn science, if you realize what you might be doing, you should not be doing it. \n In engineering, if you do not know what you might be doing, you should not be doing it.\u201d<\/p>\n— Richard W. Hamming<\/a><\/p>\n<\/blockquote>\n<\/section>\n \n<\/div>\n\n\nAcknowledgements<\/h2>\nWe might like to specific our honest due to Martin Fowler. Regardless of a \n busy schedule, he invested deeply on this article \u2014 from sharpening the \n narrative construction and clarifying key ideas, to elevating the visible \n storytelling with improved and new diagrams. His eager eye for element and \n dedication to precision profoundly formed the ultimate outcome.<\/p>\n We’re additionally deeply grateful to Eric (Ke) Zhou, Wei Solar, Sara Michelazzo, \n Rebecca Parsons, Matteo Vaccari, Might (Ping) Xu, Zhi Wang, Feng Chen and Da Cheng for his or her considerate critique and insights. \n Your enter helped us make clear a number of key ideas that underpin the methodology.<\/p>\n We additionally wish to acknowledge early practitioners: Jie Wang, Jian Gao, \n Yixuan Feng, Siyuan Li, Yixuan Li, Biao Tian, Wei Cheng, Qi Huang, and Yulong \n Li. Thanks for validating SPDD in actual initiatives, and to your endurance \n because the method matured. Your frontline suggestions has been foundational to \n making SPDD sensible and strong.<\/p>\n Lastly, within the spirit of working towards what we preach, this text itself \n was formed with the help of giant language fashions \u2014 Claude 4.5 Sonnet, \n Claude 4.6 Opus, Gemini 3.1 Professional, and ChatGPT 5.4. We relied on them for \n prose refinement, structural evaluation, synthesizing strategies, and as \n thought companions for steady studying all through the writing course of. \n Their contributions are a becoming testomony to the very method this \n article describes.<\/p>\n<\/section>\n<\/div>\n\n","protected":false},"excerpt":{"rendered":" As soon as a crew adopts AI coding assistants, the primary beneficial properties present up on the particular person degree: one developer can draft, modify, and refactor code a lot sooner than earlier than. However supply velocity isn’t restricted by typing. Whenever you have a look at the complete supply lifecycle, from necessities via launch, […]<\/p>\n","protected":false},"author":2,"featured_media":14276,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[56],"tags":[237,8862,8861],"class_list":["post-14274","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-software","tag-development","tag-spdd","tag-structuredpromptdriven"],"_links":{"self":[{"href":"https:\/\/techtrendfeed.com\/index.php?rest_route=\/wp\/v2\/posts\/14274","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/techtrendfeed.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/techtrendfeed.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/techtrendfeed.com\/index.php?rest_route=\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/techtrendfeed.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=14274"}],"version-history":[{"count":1,"href":"https:\/\/techtrendfeed.com\/index.php?rest_route=\/wp\/v2\/posts\/14274\/revisions"}],"predecessor-version":[{"id":14275,"href":"https:\/\/techtrendfeed.com\/index.php?rest_route=\/wp\/v2\/posts\/14274\/revisions\/14275"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/techtrendfeed.com\/index.php?rest_route=\/wp\/v2\/media\/14276"}],"wp:attachment":[{"href":"https:\/\/techtrendfeed.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=14274"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/techtrendfeed.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=14274"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/techtrendfeed.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=14274"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}