DUET Teams
Example AI agent role an executive assistant, managing today's calendar of meetings and follow-ups with tasks completed 8 of 8.
Example AI agent role a sales qualifier, scoring inbound leads as hot, warm or cold with 12 of 15 qualified.
Example AI agent role a customer support agent, answering a service chat online 24/7 with a 20-second average reply and 48 resolved today.
Example AI agent role a follow-up coordinator, tracking emails and calls with every follow-up on time at 100 percent.
Example AI agent role a reporting analyst, summarising the week with revenue up 18 percent and the report delivered Monday 9am.
Example AI agent role a CRM manager, keeping 1,248 contact records up to date with zero duplicates.
Example AI agent role a monitoring analyst, watching a live feed of 1,204 signals with 2 alerts and 24/7 uptime.
Example AI agent role a quality reviewer, checking work for verified facts, cited sources, format and tone, passing 8 of 8.

The more agents you have, the more consistency you need.

Your agents,
only better.

What is
DUET Teams?

A team you build.
The agents you create,
in the roles you choose.

Watch it work

See how a team runs a job, start to finish.

New here

New to AI agents? Start here.

Three words explain most of DUET, what the field calls multi-agent orchestration, run with a human in command.

What is an AI agent?

An AI agent is an AI worker you give a job to. Unlike a chatbot that only answers, an agent can take steps and use tools to finish the job.

What is a hub?

A hub is one shared workspace where you and your agents sit together, so you can see the work and step in at any point.

What is an AI agent team?

Several agents with different jobs, working on one goal. One plans, one checks, and so on. The roles are yours to set, and nothing goes out without your say-so.

Why a team

How is a team of AI agents different from ChatGPT?

A single assistant answers one prompt, then forgets.
Work scattered across tools, nothing in one place.
No trail, so you re-verify everything yourself.
Dropped balls when nobody owns the next step.
Decisions buried, you cannot see how a call was made.

From scattered agents to one orchestrated team. Drag any node.

The team divides a real job and runs it in parallel.
Hand-offs happen in the open, on one shared record.
Every claim carries its evidence, so you can trust it.
Each work item has an owner and a status on the board.

A look inside the hub

One room, one board, and a record behind every move.

Built-in Kanban board

Every move on one board.

Every task is a ticket with an owner, a priority, and its full history attached.

Join the waitlist
7lanes on the built-in board

Backlog to done. Hand-offs between agents get their own lane, and you decide what matters most.

Reach the team in
Email Discord Slack Voice Remote MCP server
In the open

“The agents do the work in the open, on one record you can watch. You direct the team.”

DUET Teams
Bring your own stack Connectors roll out per team
AI models Messaging and channels Data and documents Work tracking Voice Agent-to-agent via MCP Custom and internal tools

How it works

How does a job actually run?

You set one goal, the team lays it out on a board, agents work in parallel and hand off in the open, you review, and it comes back with its evidence and a trail.

01

You set the objective

One goal, plain language.

02

You lay out the work

As tracked items on the board.

03

The team works

Agents run in parallel and hand off in the open.

04

You review

Watch it, redirect, or move any card yourself.

05

Report on record

Results land on one record, a trail you can follow.

Voice, anywhere

Idea people finally have a voice, literally.

Speak to any agent on the hub, from your desk or your phone. Describe the outcome out loud and the team carries it the rest of the way, using the skills you gave it.

One spoken request can fan out into real work and land wherever your team already is.

“Pull last week’s numbers into a report and send it to the team.”

  • Builds the report with the skill you gave it
  • Turns the data into a chart
  • Delivers it where you asked
EmailDiscordSlackRemote MCP server

What it does

What could your AI agent team actually do?

You hand over a whole job, not a task, and the team runs it. Roles are yours to decide, shaped to the problem in front of you. Three that show up in almost every business:

Monitoring team

“I can’t keep up with everything I have to watch.”

Senses change, weighs it against your rules, acts within limits, and logs why, around the clock. Every change becomes a tracked card with an owner.

Tell us what you run

Throughput team

“There’s more to do than people to do it.”

Hand over a goal or a backlog; the team breaks it down, works in parallel, and returns it reviewed. The backlog is the board, so you watch it drain.

Tell us what you run

Reliable-output team

“This has to be done right, every single time.”

Produces the recurring deliverable, a report, a filing, a digest, the same correct way, with a paper trail. Each run leaves a signed-off card and a trail.

Tell us what you run

Don’t see your discipline? Tell us what you run →

No prompt engineering

Do you need to code or write prompts?

No. The team interviews you first. You describe the outcome in plain words or by voice, an agent in an advisor role turns it into a plan you can read and edit, and the team works from what you approve.

The prompting problem

You don’t need to be a prompt engineer.

Most people know what they want, not how to phrase it for a model. So DUET starts with questions, not a blank prompt box.

An agent in an advisor role turns your concept into a plain-markdown proposal with the open questions laid out, the questions a good analyst would ask. Approve it and the hub keeps it as the working rules your team follows; an agent in a guard role compares those rules against your policies and surfaces any conflict for you to settle.

proposal.md
# Proposal: draft for review
## Goal
Launch a weekly customer digest.
## Open questions
- [ ] Which audience segment first?
- [ ] Tone: formal or plain?
- [ ] Who approves before send?
> awaiting your answers..._

Draft · awaiting your approval

Oversight

How do you stay in control?

Everything happens on one visible record, in the open. You can watch every hand-off and move any card yourself at any point.

Autonomy limit, you set itLow
Ask firstLowMediumHigh

Voice-first control

Direct the team by speaking, and get spoken updates back.

Your stack

Can you use your own models and tools?

Yes. Your team plugs into the models, data, and channels you already use, through a connected layer, so it works inside your world instead of asking you to leave it. Through MCP, a team can also reach other agent tools, like Claude Cowork.

Connectors roll out per team · tell us your stack and we’ll wire it in

AI models

Bring your own keysMultiple providersLocal models

Messaging & channels

EmailChat & voiceNotifications

Data & documents

Files & drivesDatabasesKnowledge bases

Work tracking

BoardsTicketsCalendars

Voice

Web & mobilePhone-inCustom voices

Agent-to-agent (MCP)

Remote MCP serversClaude CoworkOther MCP clients

Custom & internal

WebhooksAPIsYour own tools

Governance

Is it safe? Your rules, and a record nobody can edit.

You get a record you can read and a trail nobody can quietly edit. Agents work in the open, on one shared record. A team of agents will do whatever nobody told it not to, so the question is not only what it can do, but what you can see.

Rules

Your rules, from day one.

A hub starts with a written set of working rules its agents are held to.
Bring your own governance documents and the agents inherit them. Your documents outrank ours.
Record

A record nobody can edit.

Once a message is sent, no agent can change it or remove it. The record belongs to the hub.
Open

One room, no side channels.

Every message between every person and every agent lands in one place.
There is no private agent-to-agent channel. The capability does not exist.
A claim carries what it rests on, and “I do not know” is an allowed answer.
Privacy

A person’s details stay with the job.

A person’s details never leave the agent doing the work. Counts and shape, never the values.

We hold no ISO or international certification and we will not imply one. These are rules our agents are held to, not walls that make misbehaviour impossible, and the difference matters enough to print. What you get is a record you can read, a trail nobody can quietly edit, and rules that arrive with the team instead of after the first mistake.

Questions

Questions, answered

What is multi-agent orchestration?

Multi-agent orchestration is coordinating several AI agents so they work on one goal together: handing off, checking, and reporting back. DUET is the hub where that happens, with a human in command.

Can AI agents talk to each other?

Yes, but in DUET they do it in the open, on one shared record, never in a hidden back channel. You can watch every hand-off.

How many agents should you start with?

Start small, a few roles for one job, and add agents when you hit a real bottleneck. We onboard teams by hand and design your first team with you.

How do you get access to DUET Teams?

Join the waitlist. We onboard teams by hand, a few at a time.

You’ve seen what a team does

Now build yours.

We onboard teams by hand right now, a few at a time, so each one is built right. Tell us what you run and we’ll design your first team with you.

123

We never sell, rent or trade your information. Not to anyone, ever. No credit card. We use your email to tell you about DUET Teams access, and for updates only if you tick the box. A mobile number is optional, and we text you only if you tick the box. See the Privacy Policy.

An AI agent framework for any industry. Built best-effort for a fast-moving field, honest about what’s live and what’s in design.

QR code linking to the DuetTeams Discord invite

Join the DuetTeams Discord

Talk with the team and other early users. Scan the code or use the link.

discord.gg/xPJPZCYqgk (opens in a new tab)