← Back to Resources

The Delivery Workflow

How one feature moves through the six BVDLC phases — from first signal to measured value, with four sign-off points where human judgment decides the next move.

Purpose: A repeatable flow every feature moves through, so intent is never lost between rooms.
Pairs with: The Delivery Model, The Delivery Squad.
Key principle: Humans decide. Agents accelerate. Sign-off points make decisions explicit.

How to read this workflow

The fourteen steps below follow the six BVDLC phases. Each step has an owner, a typical agent contribution, and a primary artifact it produces or updates. Four steps are sign-off points — the work cannot move past them without an explicit human decision.

Human-led Agent-assisted Co-authored Sign-off point

Each orange diamond is a sign-off point: 1 intent approved · 2 idea validated · 3 spec locked · 4 verified for release.

The fourteen steps, phase by phase

Phase 0 establishes why the feature matters. Phases 1–2 turn the idea into a locked spec. Phases 3–4 turn the spec into shipped software. Phase 5 closes the loop with outcomes and learning — which feeds the next trip through Phase 0.

Phase 0 · Business Context & Objectives
01

Signal arrives

A customer complaint, competitive move, regulatory change, internal hypothesis, or metric drift. Something is asking for a feature.

Owner: PM Agent: surfaces related prior signals intake log
02

Intent capture

Human-led. The PM interviews stakeholders, reads the signal in full, and writes a first-cut statement of what the feature should do for whom — with the KPI it is expected to move.

Owner: PM Agent: drafts a rough feature-brief skeleton features/<name>/feature-brief.md
03

Brainstorm session

PM and agent expand the idea. The agent offers alternative framings, prior-art references, and a rough sizing. The PM edits ruthlessly.

Co-author: PM + Agent feature-brief.md (draft)
04

Sign-off: Intent approved

The PM signs off: the feature is worth pursuing further. This is the exit gate of Phase 0 — below this line the work moves from discovery to discovery-plus-design.

Decision: PM confirms the feature is on the roadmap, even provisionally. If not → park or decline, with a note in the intake log.
Sign-off 1 of 4 Decision: PM
Phase 1 · Ideation & Rapid Prototyping
05

Domain research

The BA reads the relevant rules, regulations, and edge cases. The agent pulls referenced sources and drafts edits to the business-context documentation. The BA decides which updates are accepted into the shared context.

Owner: BA Agent: reads sources, drafts edits business-context updates
06

Rapid prototype

If the feature is novel or its value is uncertain, a small prototype is built to test the idea before committing to full design. This is where weak bets get killed cheaply.

Owner: PM + Eng Agent: builds the prototype from a minimal spec prototype + validation notes
07

Sign-off: Idea validated

The PM reviews the validation signal. If the idea holds, continue. If not, pivot or kill — cheaply, before design and build costs accumulate.

Decision: PM confirms the value hypothesis is supported. If not → pivot the feature or kill it, with the learning captured in the intake log.
Sign-off 2 of 4 Decision: PM
Phase 2 · Architecture & Design
08

Requirements & acceptance

BA, PM, Quality Lead, and an agent co-author requirements.md and acceptance-criteria.md. Criteria are Given/When/Then, machine-parseable.

Co-author: BA + PM + Quality Lead + Agent requirements.md, acceptance-criteria.md
09

Design & data contracts

The Squad Lead and an agent draft design.md and data-contracts.md. Alternatives are considered and recorded. The architect reviews.

Owner: Squad Lead Agent: drafts approach and data schemas design.md, data-contracts.md
10

Sign-off: Spec locked

PM, BA, Squad Lead, and Quality Lead all sign off. After this, the spec is the contract: agents implement against it, humans verify against it.

Decision: All four roles present and aligned. If not → resolve the disagreement before implementation. No “we’ll figure it out as we go” passes here.
Sign-off 3 of 4 Decision: PM + BA + Squad Lead + Quality Lead
Phase 3 · Planning & Work Breakdown
11

Work breakdown

An agent turns the locked spec into an epic with child stories in the work-tracking system. Each story references the spec in Git. The Squad Lead reviews the breakdown before anything starts.

Owner: Squad Lead reviews Agent: generates tasks.md and tracker items tasks.md + tracker items
Phase 4 · Build, Test & Deploy
12

Build with agent assistance

The squad runs. Squad Engineers feed agents context and review drafts. The Squad Lead reviews PRs against spec intent, not line-by-line diffs.

Owner: Squad Lead Agent: drafts code, tests, docs PRs linked to spec files
13

Sign-off: Verified for release

The Quality Lead validates acceptance criteria. Agents run the regression suite. A human signs off before merge to the main line.

Decision: Every acceptance criterion is demonstrated as passing. Every one of the team's non-negotiable ground rules is re-checked. If not → back to step 12 with a specific defect list.
Sign-off 4 of 4 Decision: Quality Lead + Squad Lead
Phase 5 · Monitoring & Value Tracking
14

Ship, measure, learn

The feature ships. The success metrics defined in the feature brief are measured. PM, BA, and agent capture the learning — what to keep, what to change, what surprised the team — into the business-context documentation, the architecture documentation, or the audit log as appropriate. That learning is the input to the next trip through Phase 0.

Owner: PM Agent: analyzes KPI data, drafts learnings feature-brief.md updates, audit-log entry

Not everything starts at step 1

The full fourteen steps describe a brand-new feature. Most of a team’s work is enhancements and fixes — and they enter the same workflow at later steps, because the earlier artifacts already exist. Same workflow, three doors in.

New feature

Enters at step 1

Nothing exists yet. Run the full path: intent, validation, spec, build, measure. All four sign-offs apply.

Enhancement to an existing feature

Enters at step 8

The feature brief and validated idea already exist — update the brief, then re-run requirements through release. Dip back to step 5 only if the change touches a domain rule. Sign-offs 3 and 4 always apply.

Production fix

Enters at step 12

The spec exists; the bug is the signal. Start by writing a failing test that reproduces it, fix, verify, ship. Step 14 is mandatory for fixes — a bug usually means the shared context was missing something, and that gap gets closed.

The four sign-off points

If a feature skips any of these, it is no longer moving through the BVDLC delivery model. Everything else can flex. These cannot.

1. Intent approved. PM sign-off that the feature is real and on the roadmap. Stops work on vague or duplicate ideas.
2. Idea validated. PM sign-off that value is plausible. Stops work on features that would have failed a prototype test.
3. Spec locked. PM + BA + Squad Lead + Quality Lead alignment before any agent implements. Stops the telephone-game drift that makes AI-built features wrong in expensive ways.
4. Verified for release. Quality Lead + Squad Lead sign-off before merge. Stops the “tests pass but the feature doesn’t solve the problem” failure mode.

How human involvement changes as trust builds

The four sign-offs are permanent — they are business judgment, and they never move to an agent. But the smaller human reviews between the sign-offs should shrink as the team earns confidence. Tie each reduction to a measured threshold, not a feeling.

Human review Why a human is there at first When it can relax
Reviewing the agent’s task breakdown (step 11) Early on, agents misjudge slice size and dependencies. When the breakdown needs no edits for several consecutive features, switch from full review to spot-check.
Reviewing every agent PR line-by-line (step 12) Trust in agent output against the pattern library isn’t established yet. When reviewer-requested changes stay low across a quarter, review against spec intent only and sample the rest.
Manually re-running acceptance scenarios (step 13) The automated verification suite hasn’t proven it catches what humans catch. When the suite has caught every defect humans also caught over a sustained period, humans review the report instead of re-running the checks.
The four sign-offs themselves What to bet on, what the contract says, and what ships are business decisions. Never. These are the speed bumps you keep on purpose.

What “verified” actually means: three different checks

At sign-off 4, three different kinds of checking happen, and they are not interchangeable. The first two exist in traditional software. The third is new — and teams that skip it find out the hard way.

1. Acceptance checks. Do the user-facing paths work end-to-end against the acceptance criteria? Pass/fail per scenario. Owned by the Quality Lead.
2. Regression coverage. Does the automated test suite now cover the new surface, so the next change can’t silently break this one? Output: new tests merged into the suite.
3. Quality evaluations — the AI-native addition. When a feature itself contains AI behavior, “did it pass?” isn’t a meaningful question — output varies run to run. Instead, run a large set of realistic scenarios, score each within an agreed tolerance, and track the pass rate over time. A falling pass rate is drift, and it is the earliest warning you will get. Feed the eval results into Phase 5 monitoring alongside the business KPIs.

How this goes wrong

Three recurring failure patterns. If you see one, treat it as a signal that the delivery model is degrading.

The skipped lock. Implementation starts while the spec is still in draft. By the time disagreements surface, agents have already written the wrong code. Fix: enforce step 10 as a real gate, with a named meeting and an artifact.
The rubber-stamp verify. Verification becomes a checkbox. The Quality Lead signs off on the fact that tests exist, not on the fact that the feature solves the problem. Fix: sign off on acceptance criteria being demonstrated, not just written.
The orphaned learning. The feature ships; metrics move or don’t; nothing goes back into the shared context. Next time, agents make the same mistakes. Fix: step 14 is not optional; the Context Manager owns making sure learnings land in files.
← Back to Resources