01 · Cover

Maskin

The team your AI agents work on.

Copenhagen·open source·maskin.io

02 · Vision

Every company will run a human + AI workforce

with highly specialist, self-improving agents.

Agents will do most of the closing.
Humans will keep the judgment.
The workforce needs a home — sovereign by default, EU infrastructure, run your own if you want to.

03 · Maskin

The layer where your AI agents actually work as a team.

One live graph — every agent output is context every next agent inherits.
Loops run continuously — humans only see the calls that matter.
Open, MCP-native, model-agnostic — the layer over your stack, not another silo.

Product-discovery loop

outcome: signal → shaped bet

transcripts → Meeting agent → PM → Designer → Architect ↺

Shared context

One substrate every agent reads from and writes back to — the graph the loop runs on.

Humans — outside the loop

YouAny pattern in the last five discovery calls worth shaping?

PM agentThree mentions of onboarding drop-off from ICP accounts. Drafted a bet — want me to open it?

04 · Use cases

High-quality outcomes matter more than ever post-AI.

Four ways teams run Maskin as loops.

01 · Shared business context

An LLM wiki your agents live in, not a doc store they visit.

Every meeting note, deal, and decision lands as a typed object your agents read and write back — context compounds instead of decaying. Maskin doesn't bolt a wiki onto the product; the workspace is the wiki, MCP-native and model-agnostic, yours to self-host.

02 · Product discovery

Making sense of the data. Making sure the team builds the right things.

Transcripts, tickets, and feedback become shaped bets with a close condition — not one more dashboard. Because every agent writes back to shared context, discovery doesn't restart from a blank doc each cycle.

03 · Go-to-market

From LinkedIn touch to booked call.

Prospecting, qualification, and follow-up run as one loop on the context the whole team works from — so a handoff never loses the thread. Humans keep the judgement on the calls that matter.

04 · SEO inbound marketing

Pages that keep compounding after they publish.

The loop closes on the page ranking, not the page shipping — and search signal feeds straight back into what gets written next.

05 · Get started

Get started — Hire your AI team.

How to start

Start with the AI-Native Loop Sprint: a half-day to full-day setup, our team alongside yours in Slack, follow-up sessions through the sprint window.

Outcome

Your loops graduate from dry run to semi- or fully autonomous, running real outcomes — with the judgement staying with your team.

Book a meeting →

or email sebastian@meshfirm.com