Platform EC

Your AI team.
Persistent. Coordinated.
Actually yours.

Most AI setups are a chat window that forgets you every session. Platform EC is a framework for running Claude as a real team — agents with memory, defined roles, and a coordination layer that survives any context reset. I build it, configure it to your context, and hand you something that works from day one.


AI that resets every session isn't a team. It's a notepad.

The tools are genuinely good. The missing piece is continuity — agents that know your context, your standards, and what was decided last week. Building that architecture from scratch takes longer than most operators want to spend.

Stateless

Every session starts from zero. You re-explain the same context endlessly.

Uncoordinated

Multiple AI sessions with no shared memory or handoff protocol between them.

Fragile

Context lives in chat history. One crash or cleared tab and it's gone.

Generic

No understanding of your clients, voice, operating decisions, or team structure.


How it works

Platform EC is a portable, Git-backed framework. The architecture is being built toward an open release — a foundation others can build on. Getting it running and calibrated to your context is where the engagement comes in. I've already found the sharp edges so you don't have to.

Scoping

A quick call to understand your team, tools, and objectives. No pitch deck — just enough to confirm the fit and define what the engagement involves before anything starts.

01
Discovery

Mapping your operating context — agents, roles, workflows, decision patterns, and integration points. This becomes the foundation the framework is configured against.

02
Production

The build and deployment. Agent configuration, memory architecture, coordination layer, and session protocols — calibrated to your context and delivered at a pace that meets your objectives.

03
Support

As your team grows or shifts, the framework adapts. New agents, new projects, new machines, new integrations — handled as retainer or fixed-scope work as needed.

04

Built to be open. Not there yet.

The framework is being developed with a public release in mind — a documented foundation that teams and operators can build on, extend, and adapt. The intent is layers: a core others can fork, with bespoke configuration on top.

The engagement is for operators who want this running and calibrated to their context. Initial setups are fixed-scope. Ongoing customisation runs as retainer.

Engagements start with a scoping call. We move at a pace that fits your objectives.

The open release will be announced here. If you want early access or want to follow the build, get in touch.

Building an AI team?

Drop your email. I'll follow up directly to understand your setup and whether Platform EC is the right fit. No pitch deck, no automated sequence — just a direct conversation about what you're trying to build.

contact@erikchristensen.ca

No newsletter. Just a direct reply from me.


The stack

Platform EC is platform-agnostic by design — plain Markdown, standard Git, no proprietary tooling. It works with any AI layer and survives any tool change. Everything is readable by humans and machines alike.

Memory

Git repos per agent. Every session commits. Nothing lives only in chat history.

Live state

PocketBase as the operational layer — real-time agent status, task state, and session identity. Lives on your own server.

Identity

GitHub OAuth device flow. One approval provisions both your PocketBase session and repo access — no credentials in chat, ever.

Tasks

Asana for human accountability. PocketBase for agent-level task state. Both layers stay in sync.

Portability

Fetch one URL and you're connected. Any session — Cowork, claude.ai, or other — boots from a single endpoint. No files to load, no credentials to paste.

Open layer

A documented public release is planned. The foundation, protocols, and architecture — built to be shared.