Nº 01
Discover
Inputs
- Customer Feedback
- Goals & Initiatives
- Corporate Strategy
Output
Product Brief
The Cadence framework
A delivery system
built for AI.
Cadence is how Sigao moves an engineering org toward AI-native delivery across the SDLC. It ships with the roles, events, and rollout playbook to get there. We coach your team to run it, then step out, and your team keeps running it long after we leave.
What Cadence is
The engine that turns AI capability into engineering velocity.
You don’t install Cadence or read it once and move on. It runs your delivery org day to day. We built it for teams where engineers and AI agents now ship side by side, and designed out the handoffs that slow delivery down.
The three pillars below hold the system up. The full thing runs eight chapters deep: lanes, pods, roles, signals, artifacts, events, the AI Guild, and the reasoning behind each one.
Pillar · Nº 01
Lanes
Discovery, delivery, and enablement run as three separate streams, so new product work doesn't stall ongoing improvement and neither one starves the push to build AI capability.
Read more →
Pillar · Nº 02
Pods & roles
Small pods that stay together pair your engineers with AI agents on one delivery system. Everyone knows who owns what, every agent has a defined boundary on what it's allowed to do, and the trail is there when an auditor asks.
Read more →
Pillar · Nº 03
Events
Four coordination events keep teams aligned without filling the calendar. When something blocks a pod, the same events give it a path to escalate.
Read more →
The value funnel
Ideas come in on the left. Customer value comes out on the right.
Each stage sharpens the idea, checks that it fits, and drops the work that doesn’t hold up. Cadence is built to ship value, not output. What can’t earn its place never reaches your customers.
Ideas inCustomer value out
Nº 02
Understand
Inputs
- Epic Document
- Research & Design
- UX Walkthroughs
- Product Research
Output
Specification
Nº 03
Focus
Inputs
- Testing Patterns
- UX Patterns
Output
Plan
Nº 04
Deliver
Inputs
- Patterns & Practices
- Codebase Context
Output
Product Increment
Nº 05
Measure
Inputs
- Observability
- Feedback
- Input
Loops back to Discover
Spec-driven development
The spec is the control plane.
Every stage of the funnel reads and writes one artifact: a living, machine-readable spec. Engineers and AI agents build from it, quality gates verify against it, and what ships updates it — that's the loop Cadence runs on.
- Nº 01
Intent
The value the product must create: the brief, the goals, the constraints.
- Nº 02
The spec
Living, machine-readable, in source control. Behavior, constraints, and definition of done become the system of record.
- Nº 03
Build
EngineersAI agentsBoth build from the same spec, with no retelling and no drift.
- Nº 04
Quality gates
Tests, evals, and review verify the work against the spec, not against memory.
- Nº 05
Ship
A production increment traceable back to the intent it came from.
What ships feeds back: learnings update the spec, not just the code, so the system of record never goes stale.
The handbook
Read it chapter by chapter.
Eight chapters: lanes, pods, roles, signals, artifacts, events, the AI Guild, and the reasoning behind how Cadence is put together. Each one takes a single part of the system and goes deep.
Chapter · Nº 01
Why Cadence
Modern software delivery has two failure modes. AI without process redesign burns people out. Process without AI gets outpaced. Cadence addresses both, in practice rather than as a manifesto.
Read the chapter →
Chapter · Nº 02
Three lanes
Three lanes group every role in the org by what it owns: direction, delivery, or what delivery runs on. Together they form one delivery system.
Read the chapter →
Chapter · Nº 03
Pods
How a Deliver pod is composed: 3–4 humans, two AI agents, a public and private library, and the temporary multi-pod alignment used only when work depends on it.
Read the chapter →
Chapter · Nº 04
Roles
Chief Product Owner, Engineering Manager, Engineers, Product Owners, Flow Leads. Plus the process maturity model the Engineering Manager tracks across five SDLC dimensions.
Read the chapter →
Chapter · Nº 05
Signals
Signals are the inbound data points that drive prioritization. We name them explicitly — strategic, customer, operational, financial, AI-capability — and describe how they get curated into the product briefs that feed Signal Inspection.
Read the chapter →
Chapter · Nº 06
Artifacts
Backlog, Product Briefs, Specifications, Plans, Implementation, Instrumented Product Increment. Each artifact is named explicitly so the org can inspect, validate, and improve what flows between events.
Read the chapter →
Chapter · Nº 07
Scaled events
Signal Inspection, Scaled Planning, Flow, and Scaled Review & Retro — four deliberate pauses in continuous flow, plus the red button either lane can pull at any time.
Read the chapter →
Chapter · Nº 08
AI Guild & Maturity
The Guild sets the standard. The Community of Practice carries it. Five maturity dimensions — Strategy, Value, People & Culture, Governance, Tools & flows — get tracked continuously.
Read the chapter →
For engineering leaders
What you can take to the board.
Rolling out a system across the org is a call you have to make and answer for. Here is what Cadence buys you, stated in the terms you’ll defend it in.
Roadmap predictability
Prioritization runs off one backlog, intake is driven by signal, and PIs land on time. Your delivery org commits to dates it can actually hit, so you can put them on a slide and stand behind them.
Defensible ROI
We measure cycle time, defect-escape rate, and how often you deploy, and we move them together. You take DORA-style numbers to a CFO instead of a story about velocity going up.
A bet that gets safer
The rollout is staged, the trail is there for an audit, and the capability stays with your team. No reorg overnight, and no vendor you can't function without when it's over.
Answering to a board or PE sponsor
If a board or PE sponsor is asking you about AI, you’ll walk in with a sequenced plan backed by evidence and DORA-style metrics you can defend. You’ll also get a straight answer on what AI will and won’t do on your timeline.
Adopting Cadence
Your team runs it without us.
We roll Cadence out in stages and coach your people while the real work is happening. By the end, your team runs it without us and keeps the capability for good.
Phase 01
Pilot one pod
We adapt Cadence to how you actually work, then run it inside a single team. Nothing changes org-wide until the model has proven out on your stack and your constraints.
Phase 02
Partner and coach
Your engineers, POs, and flow leads learn the system by running it next to us. Specs, signals, events, and quality gates become how they work, not something we do on their behalf.
Phase 03
Expand by lane
We add teams one at a time instead of forcing a reorg. An AI Guild and per-team Communities of Practice carry the standard so each new team starts from what already works.
Phase 04
Hand off
You're left with the handbook, the people who can teach it, and a team that runs Cadence on its own. We step out and delivery doesn't skip a beat.
The lift outlasts the engagement. You keep the system, and the people who know how to run it.
Why a framework matters
Alone, AI and process both hit a ceiling. Together they compound.
Bolt AI onto a process you never changed and you stay capped. Rebuild the process but give humans and agents no shared way to work, and you cap out just the same. Cadence changes both at once.
AI alone
A coding assistant inside a broken intake. Faster generation, more rework. Quality usually trends down.
More rework
Outcome
Process alone
Scrum, lean, DevOps without AI. Real gains, but capped at the velocity humans alone can sustain.
Plateaus
Outcome
Process + AI, aligned
Specs that humans and agents both read. Quality gates between every funnel stage. One artifact moves the work forward on both sides.
Compounds
Outcome
What you operate with
Humans and AI agents on the same delivery system — safely, with everything traceable.
- One source of truth for prioritization across the whole org
- Specs that humans and agents can both consume
- Clear authorization boundaries for AI actions
- Cross-team coordination and impediment escalation built in
- Audit-friendly delivery: every decision traceable
- A staged rollout path that doesn’t require a big-bang reorg
Want a walkthrough?
We’ll walk Cadence through your setup.
Bring your team shape, your stack, and where you are with AI. We’ll point out the parts of Cadence that fit you, the parts that don’t, and what the first 90 days of rollout would actually involve.
Schedule a walkthrough