Multiplayer work. Multiplayer AI.

Them AI brings people and AI agents into one shared workspace. It captures the judgment behind real work. Turns it into reusable skills and makes the whole team smarter every time you use it.

Join the waitlist See how Them AI learns

Mac-only for now. Free during the beta (invite-only). Works with Claude, ChatGPT, Cursor, and any MCP agents.

Them AI product

Humans and agents in the same room, with a human approval gate.

Multiplayer by design.
Smarter every time you use it.

Who is Them?

You already know Them.

Them is the person who...

sees problems differently from everyone else
has years of judgment your company depends on
knows how things work beyond the process documents
remembers why important decisions were made
has expertise that would take years to replace
knows what to do when there is no playbook
sees problems differently from everyone else
has years of judgment your company depends on
knows how things work beyond the process documents

They are the person everyone turns to when the obvious answer is not good enough.
People like Them help build companies. Them AI helps their expertise keep building.

The problem

Transferring memory to AI is not enough. Expertise should also follow.

Your company saves the work, but not the thinking behind it.

The presentation survives, but not the reasoning behind it.

The process document explains what usually happens, but not what practically happens.

When a customer asks for an exception or when a complex problem shows up, a decision has to be made without enough information. That is when experience matters.

But experience lives in people. In corrections. In conversations. In hundreds of decisions nobody thought to document. When those people leave, change roles, or simply forget, the company keeps the files but loses judgment.

A real answer with its reasoning and three cited memories

A real answer with its reasoning and three cited memories.

NovanowAgent

12% loyalty discount, and the reasoning is attached: Priya traded it against the onboarding tier Meridian never used, and planned to earn it back with a 50% expansion inside two quarters. The number was a trade, not a concession. That trade logic is the reusable part.

Meridian pricing history v3 · owner PriyaThe Meridian trade v1 · owner PriyaDiscount policy v3 · owner Marcus

A real answer with its reasoning and three cited memories.

Files can tell an AI what happened, but Them AI helps it learn how your best people handled it.

The multiplayer workspace

Humans and AI in the same room. One workspace. Many minds.

Create a room for a project and bring in the people who understand it along with AI agents that can help. Then let them work together with the same context.

One workspace. Many minds.

Chats

Humans and agents in the same room, with a human approval gate.

A shared room drafting the Meridian renewal

Collective

The whole organisation thinking in real time, in one governed brain.

The live pulse of the collective

Memory

Facts, decisions, and context: owned, versioned, and cited.

Living memories with owners and versions

Skills

Repeatable ways of working, taught once and credited forever.

A taught skill with its owner and usage

Harness

A private room where you train your agents on how you work.

Training agents in the harness room

Review

Agent work waits at a human gate before it becomes canon.

A deliverable waiting for human review

One agent researches. One analyses. One drafts.
Another checks the work.

Your people guide the thinking. Correct the mistakes. Make the final call.

When one person improves the way a problem is solved, the whole team can benefit from it.

How Them AI learns

The learning loop. Every project trains the next one.

Them AI does not get better by collecting more random information. It gets better by learning from real work.

The learning loop in the product

Every correction should fix more than one answer.
With Them AI, it can improve every answer that comes after it.

Memory and skills

Knowing is not the same as knowing how.

Memory remembers

Memory keeps the facts, decisions, sources, owners, corrections, and context behind the work. But memory alone does not recreate expertise.

Living memories, owned and versioned

Skills know what to do next

That's where skill comes in. Skills know what to do next. Together, they give it a better way to think about the problem.

A taught skill with its owner and usage

Do not just teach your AI what your company knows.
Teach it how your company works.

AI teammates

Your AI team

Create specialised AI teammates for the work your company repeats. Each agent has a role and a specific set of skills it needs. They work with your people and the other agents in the room.

Six specialised AI teammates at one desk
A researcher

that finds the evidence.

An analyst

that challenges the assumptions.

A proposal writer

that knows how your best proposals are structured.

An account manager

that remembers every important customer decision.

A project manager

that knows how your team plans, reviews, and delivers.

A reviewer

that checks the work before it reaches you.

Human control

You decide what it learns.

Nothing should become company practice simply because an AI said it confidently. Them AI keeps learning visible, reviewable, and accountable.

An agent's work waiting for one human tap

Nova wants to send v2 to Meridian under Priya's name.

Approvals also arrive as Slack DMs when the bridge is on. One tap either place.

Approve and send

The gate, in the product: an agent's work waits for one human tap.

Your company already has the intelligence, so now let's build a team that keeps learning from it.

Turn real work into reusable skills and make every project improve the next one.

Mac-only for now. Free during the beta (invite-only). Or become a design partner.

People move on. Expertise should stay and evolve.