A prompt is not a requirement

Teams paste acceptance criteria into a prompt and call the story done. The model may produce something that looks right once. Production needs behaviour that is testable, versioned, and owned when it fails at 2 a.m. A prompt is an implementation artefact. A requirement is a commitment about outcomes and constraints.

Confusing the two is understandable: both are text, both influence behaviour. Only one survives a model upgrade, a vendor switch, or a new hire who never saw the original chat.

Separate the layers

Requirements: outcomes, constraints, SLAs, data rules, and who decides exceptions. Design: interfaces, failure modes, fallbacks, and the human loop. Prompts, tools, and model choices: implementation details under that design.

When the model changes, requirements should still hold. If your only specification is a prompt file, you have coupled the business to a vendor’s next release notes. That is the software equivalent of storing business rules only in a stored procedure nobody documents—except the procedure now rewrites itself when the provider ships a new snapshot.

Write what you can verify

Prefer examples, golden cases, and measurable thresholds over vibes. “Be concise and accurate” is not a test. “For these fifty claim summaries, extract these fields with at least X precision, and never invent a policy number” is a test.

AI can help draft tests; humans still decide which cases represent money, risk, and reputation. Involve domain experts in building the case set. Engineers alone will under-sample the weirdness that operations already knows by heart.

Traceability

Keep a thin chain from requirement ID to evaluation case to prompt version to release. You do not need a heavyweight ALM theatre. You do need to answer, under audit or incident review: what did we intend, what did we ship, and what evidence said it was good enough?

Store prompts in the same change-control world as code. Review them. Diff them. Tag them with the requirement they serve. A prompt edited live in a portal with no history is an uncontrolled production change.

Conversation design is still product design

For assistants, the dialogue is part of the product. Requirements should cover escalation, refusal, and what the assistant must not do—not only the happy path tone. “Never provide regulated advice; offer to connect to an agent” is a requirement. The wording that implements it is a prompt.

Product managers who only write brand voice guidelines leave the hard behavioural rules to engineers improvising under deadline. That improvisation becomes the de facto policy.

Practical working agreement

Agree in the team: no story is ready for AI implementation until outcomes and constraints are written outside the prompt. No release without evaluation against those constraints. No prompt change in production without the same review bar as a code change.

That agreement restores an old truth in a new medium: software still needs requirements. Models need instructions. Do not let the second pretend to be the first.

How this plays out in delivery

In practice, the difference between a slide and an operable change is whether teams can point to owners, controls, and evidence under pressure. That pressure arrives as an incident, an audit question, a vendor outage, or a steering committee that wants to scale a demo. If those answers are improvisations, the feature was never production-ready—regardless of how polished the interface looked in a pilot.

Delivery leaders should therefore reserve explicit capacity for the boring work: contracts, runbooks, evaluation packs, fallback paths, and decision records. Boring work is what makes ambitious AI and platform change survivable. Skipping it to protect a velocity chart is how organizations repurchase the same programme every three years under a new name.

Questions worth asking in design review

Who owns the outcome after launch? What fails first when the dependency is slow or wrong? Which data classes are in motion, and under which policy? What would make us pause or roll back within an hour? What evidence will we examine weekly to know quality is holding?

If a proposal cannot answer those questions without hand-waving, keep the scope in a controlled experiment. Experiments are useful. Unowned production traffic is not an experiment—it is a risk acceptance you forgot to record.

Working habits that keep quality compounding

Write short decision records when you cross a boundary. Keep a living list of invariants for each critical capability. Review with a checklist that targets blast radius rather than style. Instrument silent failure modes, not only HTTP errors. Revisit model, prompt, and vendor changes with the same seriousness as database migrations.

None of these habits require a new framework brand. They require leadership attention and a refusal to confuse demos with operable systems. Teams that practice them can adopt assistants, agents, and new platforms without losing the enterprise plot.

A note on language and accountability

Replace slogans with operational nouns: capability, owner, control, fallback, evidence, exit. Language shapes governance. Teams that speak only in broad transformation slogans struggle to assign accountability. Teams that speak in capabilities can fund, staff, measure, and stop work when the evidence says stop.

Hold that vocabulary in architecture forums and programme boards. Over time it becomes the shared spine that lets software development absorb AI tools without fragmenting into disconnected experiments.

Closing

The through-line is simple: treat AI-related change as architecture and operations work, not as magic. Make ownership explicit, put uncertainty where blast radius is acceptable, measure what can silently fail, and refuse to scale what you cannot run. That discipline is how software organizations get durable value from new techniques instead of a temporary theatre of progress.

A worked example

Requirement: “For registered customers, summarize the last three claim notices into bullet points covering date, type, and status. Never invent a claim number. If sources are missing, say what is missing and offer an agent. P95 latency under four seconds. Log for audit with case id.”

Prompt: the wording and few-shot examples that steer the model toward that behaviour. Eval: fifty real anonymized cases with expected fields and forbidden hallucinations. Design: retrieval source, timeout, fallback, retention. If you only keep the prompt, you cannot tell whether a regression broke the requirement or merely changed style.

Ownership across roles

Product owns the requirement. Architecture owns fit and fallback. Engineering owns the prompt implementation and tests. Risk owns acceptance of residual uncertainty for the use case. When one role writes everything into a prompt file, accountability dissolves the first time a customer complains.

Make the hand-offs visible in your tracker. A story is not ready for development without requirements outside the prompt. A release is not ready without eval evidence against those requirements. Simple rules. High leverage.

Versioning without drama

Store prompts next to code or in a versioned store with the same review rules. Tag releases with prompt and model versions. When customer outcomes change, you need to know what changed in the instruction stack. Hot-editing production prompts through a vendor UI without history is an uncontrolled change—treat it that way.

If business stakeholders want to tweak tone weekly, give them a governed path for tone fragments, not raw access to safety and policy instructions. Separate what marketing may adjust from what risk has frozen.

Putting it into the operating rhythm

None of this sticks if it appears only in a one-off workshop. Put the checkpoints into existing forums: design authority, risk intake, sprint reviews, and operations reviews. Assign named owners. Review the same metrics until the behaviour becomes muscle memory. Tools change. Operating rhythm is how architecture survives tool change.

When you expand scope—new channel, new market, new model provider—re-run the same questions rather than assuming last quarter’s controls still fit. Scope expansion is where quiet regressions hide. A short re-certification beats a long incident report.

What good looks like after six months

You can name owners for each AI-touched capability. You can show evaluation trends and cost trends. You can pause a feature without heroics. Engineers can explain the boundaries without opening a chat history. Sponsors hear outcome language and evidence, not only model brand names. That is the standard. Aim for it deliberately.

Field notes from programmes that stuck

The programmes that keep their gains share a few unglamorous traits. They name a single accountable owner for each capability touched by AI or platform change. They keep a thin evidence pack current: what the feature does, which data it uses, how quality is measured, and how to pause it. They review cost and quality on a fixed weekly or biweekly cadence instead of waiting for a quarterly surprise.

They also refuse to expand scope while basic controls are missing. That refusal feels slow in the week it happens and fast in the year it saves. Sponsors accept it more readily when you offer a dated path: we can widen to segment B when override rates stay under threshold and fallback drills pass. Conditional speed is still speed—just adult speed.

Common failure modes to watch for

Eternal pilots that serve real customers without on-call. Prompt edits in production with no version history. Retrieval corpora that mix clearance levels. Success metrics that count messages sent instead of outcomes achieved. Architecture reviews that only admire diagrams after contracts are signed. Cleanup work that is always scheduled after this critical launch.

Each failure mode is preventable with a small control. The danger is not that teams cannot invent controls. The danger is that delivery pressure makes skipping them look rational until the incident report is written.

A ninety-day improvement plan

Days 1 to 30: inventory AI-touched journeys, owners, and gaps in logging, evaluation, and fallback. Publish the list without blame. Days 31 to 60: close the top three blast-radius gaps; stand up a short design-authority slot for new use cases; put cost alerts in place. Days 61 to 90: run one controlled promotion from pilot to production using a written exit report; kill or contain at least one unmanaged experiment.

At day ninety, present evidence trends rather than tool logos. Leaders can fund what they can see. Visibility is a prerequisite for sustained investment in quality.

How this article should change Monday morning

Pick one capability you touch. Write the owner, the invariant that must not break, the fallback if the AI dependency fails, and the metric you will check next week. Share it with the people who can correct you. Then make one concrete backlog item that turns a gap into a control.

Long articles do not improve estates. Changed operating habits do. Use this piece as a mirror, not as literature. If nothing in your plan of record moves after reading it, the reading did not count.

Further depth for practitioners

If you lead architecture or engineering in this space, keep a personal register of decisions you refuse to leave implicit: where models may advise, where humans must commit, which data products are canonical, and which vendor exits you still believe are feasible. Revisit the register when a programme asks for exceptions. Exceptions are fine. Untracked exceptions become the real architecture.

Pair that register with two drills a year: a model-provider outage drill and a quality-regression drill. Prove you can fail closed or fail to human without inventing the process during the incident. Drills are cheaper than press releases.

Stakeholder conversations that unblock work

With executives, speak outcomes, cost to serve, and residual risk—not model parameter counts. With risk, bring evidence packs and known failure modes early. With delivery, bring paved roads and review checklists that match blast radius. With vendors, bring exit and data-use questions before demos set the emotional agenda.

Most stalled AI and platform work is not stalled for lack of tools. It is stalled for lack of a shared decision. Your job is often to force that decision into the open, record it, and help the organization live with it.

Keep the bar honest

Shipping is not the same as absorbing. A feature that requires permanent heroics is still a prototype, even with thousands of users. Hold the bar at operable, explainable, and pausable. When those three are true, growth is earned. When they are not, growth is borrowed from future incidents.

Long articles do not improve estates. Changed operating habits do. Use this piece as a mirror, not as literature. If nothing in your plan of record moves after reading it, the reading did not count.