series · 7 essays
The Agent-Ready Platform
Platform engineering spent a decade making the developer the customer. These seven essays argue the customer just changed: developers now arrive with an agent beside them, and the agent cannot work around the ambiguity humans absorb every day. The thread is not adding AI to the platform. It is what the platform owes a consumer that only knows what we make available to it.
The essays build one argument. The new user gets named first, along with a test most platforms would fail: could an agent figure out how to deploy correctly without asking anyone? Context comes next, because the scarce resource is trusted context and providing it is platform work. Then the interface moves up a level: developers state intent, agents compile it into approved primitives, and the golden path becomes the compiler target. The fourth essay lands the destination: AI developer experience that stops being a destination and becomes how the platform behaves. The last three turn to what the platform owes once agents act through it. A shared agent workflow needs a release process before another team can inherit it. An agent acting under delegated authority needs an identity of its own and a stop the platform can prove. The closing essay keeps the engineering system durable enough that the agent itself stays replaceable.
Read them in order; each essay opens where the one before it stopped. This thread runs on platform ground where The Agentic SDLC runs on delivery ground, and they meet at the gate: standards live in systems, and verification does not care who produced the change. The supporting reads carry the prior arguments the series leans on.
reading order
The essays
- 01Your Platform Has a New User: The Agent11 min read
Names the new user: the platform now serves the developer and the developer's agent, and agent readiness becomes a maturity test.
- 02Context Is a Platform Capability Now9 min read
The first capability: trusted context, canonical, versioned, and fresh, provided by the platform instead of gathered by hand.
- 03The Golden Path Needs an Intent Interface7 min read
The interface moves up: developers state intent, agents compile it into approved primitives, and the golden path becomes the compiler target.
- 04The Best AI Developer Experience Might Be Invisible13 min read
The destination: AI stops being a chatbot to visit and becomes how the platform behaves, healing what it owns and proposing what it does not.
- 05Your Agent Workflow Needs a Release Process10 min read
The first obligation: a shared agent workflow ships as a versioned release with an owner, so the next team inherits a supported capability.
- 06The Agent Is Wearing Your Badge14 min read
The second obligation: an agent acting on someone's authority needs its own identity, evidence linked across systems, and a stop the platform can prove.
- 07The Agent Is Not the Platform17 min read
The boundary: the agent is a user of the platform, not the platform, so intent, verification, and evidence must survive a change of agent.
supporting reads
Around the spine
- Complexity Is Inevitable. Drag Is Optional.
The pre-agent thesis: good platforms convert unavoidable complexity into defaults and automation, the muscle this series extends.
- Trust the Gate, Not the Actor
The governance rule the series leans on: the gate checks the result against the constraints, no matter who produced it.
- Nobody Pages the Agent
The open question answered: when intent compiles to implementation, operability has to enter the spec.
- Your Platform Is Teaching Your Engineers
The teaching problem: defaults, errors, and exceptions show engineers and agents what normal work looks like, whether or not anyone chose the lesson.