> ## Documentation Index
> Fetch the complete documentation index at: https://unko.design/llms.txt
> Use this file to discover all available pages before exploring further.

# Sequences: Cross-Company Workflows

> How strands and genes work together as repeatable, accountable flows.

# Sequences: Cross-Company Workflows

Sequences are the repeatable, cross-functional workflows that run your company. They connect DNA (intent) → Strands (systems) → Genes (rules) so work moves predictably between teams and tools.

<Callout type="info">
  A sequence is not a document. It's an executable flow: triggers, ordered steps, gates, owners, and signals. Well-designed sequences let you change behavior by changing a few genes and steps — not by endless meetings.
</Callout>

## Why sequences matter

* They translate strategy into repeatable action.
* They encode responsibility and handoffs between strands.
* They make testing, automation, and AI safe by constraining operations to known steps and checks.

## Sequence anatomy

* Trigger: What starts the flow (bug report, customer request, roadmap item).
* Inputs: Required artifacts (ticket, spec, mock, contract).
* Steps: Ordered actions mapped to strand owners; each step references one or more genes.
* Gates: Tests, reviews, or approvals that must pass before proceeding.
* Outputs: Deliverables and signals (release, invoice, updated docs).
* Rollback: Clear undo steps and metrics that indicate rollback conditions.
* Owner: Single sequence owner responsible for health and cadence.
* Signals: 3–5 metrics that show flow health (lead time, cycle time, error rate, customer satisfaction).

## How strands and genes interact inside a sequence

* Each step maps to a strand and applies the relevant genes (e.g., UI gene for component changes, Security gene for auth changes).
* Gates enforce genes: CI contract tests (Tech gene), copy sign-off (Brand gene), accessibility checks (UX gene).
* When a gene changes, update affected sequences and run a lightweight impact review across strand owners.

## Canonical example: Feature Development Sequence

Trigger: Product roadmap ticket approved
Inputs: PRD, high-fidelity mocks, acceptance criteria, data schema
Steps:

1. Product: Confirm acceptance criteria (Gene: Ownership — Human)
2. Design (UI/UX): Create accessible mocks (Gene: Accessibility) → deliver components
3. Tech: Implement backend + API (Gene: API Response Format) → run contract tests
4. Data: Add/evolve schema, update metrics (Gene: PII Access Control)
5. QA: Run automated tests + exploratory checks (Gate: CI & QA signoff)
6. Ops: Release with feature flags (Gate: Rollout plan & rollback ready)
7. Conversation/Sales: Publish release notes and enable enablement content (Gene: Brand Voice)
   Gates:

* CI green, contract tests pass
* Security quick-scan for auth/data changes
* Product sign-off on acceptance tests
  Outputs:
* Release toggled to 100% or staged rollout
* Changelog, ticket closed, metric dashboard updated
  Signals:
* Lead time, customer-reported bugs, deployment success rate, activation metric

<Callout type="tip">
  Map every sequence step to specific genes and a single executor (Human | AI | Hybrid). If AI executes a step, require tests, logging, and a human gate for irreversible outcomes.
</Callout>

## Governance: change flow for sequences

* Propose sequence change as a PR linking affected strands and genes.
* Run impact checklist: owners, gates, tests, rollback.
* Add CI/agent policies where Executor=AI/Hybrid.
* Run a 1–2 cycle post-rollout review and bake learnings into genes.

## Quickstart: your first sequence (10–30 minutes)

1. Pick a high-friction workflow (e.g., “bug → fix → release”).
2. Document trigger, 4–6 steps, owner, one gate, and 2 signals.
3. Map each step to one gene.
4. Run the flow once manually, note failures, then automate linting/tests.
5. Iterate: tighten genes or adjust steps based on signals.

## Common anti-patterns

* Monster sequences with too many owners (no single owner).
* Gates that are just meetings (make gates testable).
* Sequences that ignore rollback or observability.
* Letting AI make irreversible decisions without human gates.

## Short checklist: healthy sequence

* Clear trigger and owner
* 4–8 discrete steps
* Gates with automated checks where possible
* Mapped genes on every step
* 3 signals tracked and reviewed weekly

<Callout type="success">
  Sequences are the living glue of DNA: change the sequence or its genes, and you change behavior company-wide — predictably and safely.
</Callout>