Sigao

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.

The AI Guild.

The AI Guild is the standards body for AI within the organization. It lives inside the Enable lane, and its job is to keep AI usage maturing across the company in a way you can measure, without becoming a gate that pods have to wait on.

What the Guild owns

  • Standards. Shared prompt libraries, approved patterns for agent architecture, retrieval, evaluation. The reference shape pods inherit and then specialize for their own context.
  • Governance. AI risk policy, audit cadence, vendor evaluation, the escalation path when something goes wrong. These exist so individual pods don’t each have to invent their own answers and so the org can defend its decisions later.
  • Maturity tracking. The Guild measures the org’s AI capability across the five dimensions described later in this chapter. The point of tracking is to know where to invest next, not to grade pods.

The principle: consult, don’t gatekeep

The Guild raises the floor. Pods raise the ceiling.

The Guild publishes patterns, exemplars, and policy. Pods adopt, adapt, and report back what they learned. When the Guild becomes a checkpoint pods route around, the model breaks. When it becomes a peer pods come to for help, the model works.

The AI Community of Practice.

The Guild’s standards aren’t worth much on paper. The AI CoP is where they actually become live practice across the org.

What it is

The AI CoP is a regular gathering of pods from across the organization, not only the ones aligned on a single product or value stream. It’s open by default. Anyone using AI in their pod is welcome.

What it’s for

  • Guild-to-pod communication. The CoP is the Guild’s primary channel. New standards, deprecated patterns, evaluation updates — they spread here first.
  • Pod-to-pod learning. Patterns one pod proved out get shared with others. Mistakes one pod made get prevented elsewhere. AI maturity moves at the speed of the org’s fastest learners.
  • Realized value. AI capability that stays inside one pod is theoretical from the org’s point of view. The CoP is where it spreads and starts to add up across teams.

Without the CoP, pods reinvent the same patterns in parallel and the standards never reach them. With it, the gap between “the policy says X” and “every pod actually does X” closes.

Five maturity dimensions.

The AI-Native Software Engineering Maturity Model splits AI capability into five dimensions. The Guild tracks all five continuously. The goal isn’t a single score; it’s knowing which dimension is the constraint right now.

Dimension · Nº 01

Strategy
  • Goals for software delivery
  • AI strategy
  • AI use-case portfolio
  • Cross-functional partnerships

Dimension · Nº 02

Value
  • Measure AI impact
  • Manage AI costs
  • Communicate value to stakeholders

Dimension · Nº 03

People & Culture
  • Upskill developers
  • Manage change
  • Developer experience
  • Evolve roles & staffing

Dimension · Nº 04

Governance
  • AI risk and governance policy
  • Adherence to policy

Dimension · Nº 05

Tools & flows
  • AI tool & vendor selection
  • AI implementation
  • Workflow redesign
  • Context engineering

Dimension Nº 01

Strategy

Strategy is the dimension that asks: does the org know what it’s trying to do with AI? Most companies have an enthusiasm strategy and call it a strategy. This dimension forces the question to be answered concretely.

  • Defined goals for software delivery. Specific outcomes the org expects software to produce, with and without AI.
  • A refined AI strategy. Not a statement of intent, but a working strategy that gets revised as evidence accumulates.
  • A managed AI use-case portfolio. The org knows where AI is being used, where it’s being piloted, where it’s being shut down, and why.
  • Cross-functional partnerships. Strategy isn’t a CTO solo project. Product, ops, legal, security, and engineering all sit at the table.

Dimension Nº 02

Value

Value is where AI claims meet AI evidence. This dimension is about measurement: what AI is actually delivering, what it’s costing, and how that gets communicated outside engineering.

  • Measure AI impact. On delivery speed, on quality, on customer outcomes — concretely, with numbers, not anecdotes.
  • Manage AI costs. Compute, vendor, license, and human time. AI bills compound quietly without explicit management.
  • Communicate value to stakeholders. The board, the customers, the regulators, the team. Each audience needs a different story; each story needs to be true.

Dimension Nº 03

People & Culture

The dimension that takes the longest and gets shortchanged the most. People & Culture is about whether the human side of the org is moving forward at the same rate as the tooling side.

  • Upskill developers. Practical training that lands in real workflows, not one-off workshops.
  • Manage change. Transparent communication about what’s shifting, who owns what, and what hasn’t been decided yet.
  • Manage developer experience. AI that helps, doesn’t condescend, and doesn’t create new toil that makes developers regret their day.
  • Evolve roles and staffing. What a senior engineer does in 2026 is not what they did in 2022. Recognize that, then staff and structure teams around it.

Dimension Nº 04

Governance

Governance is what keeps AI usage defensible. It’s a short list of items with outsized consequence: the difference between an AI program that scales and one that gets paused after an incident.

  • Establish AI risk and governance policies. Written policy on data handling, model use, agent authorization, and incident response.
  • Ensure adherence. The policy in production has to match the policy on paper. Audit cadence and traceability close the loop.

Dimension Nº 05

Tools & flows

Tools & flows is the most visible dimension and the easiest to mistake for the whole picture. It covers the AI tooling stack and the workflows the tools live inside.

  • Select AI tools and vendors. Deliberate selection criteria, periodic re-evaluation, exit plans for tools that don’t earn their seat.
  • Implement AI tools. Onboarding pods to new tools is its own discipline. Done poorly, expensive licenses sit unused.
  • Redesign workflows. Bolting AI onto an unchanged workflow is the headline failure mode. Reworking the workflow itself is where the gains come from.
  • Enable context engineering. Getting an agent the context it needs, when it needs it. Increasingly one of the highest-leverage skills on the team.