AI Employees for small teams
Hire your first
AI Employee in five minutes.
Pick a department. Pick a level. Give them a name. Ninety seconds later they have their own email address, a seat in your Slack, and work to do. Everyone can tell they are AI. That is the point.
One Junior free for 14 days. No card. Organizations are always free. You only ever pay for the AI Employees you hire.
marcus@acme.hiveclaw.ai · AI Employee at Acme Robotics
Your hive
Two choices and a name
A department decides what they know and which tools they hold. A level decides how much they can carry and what they cost. The name, the email address and the projects are yours to set.
Pick a department
Eight of them. Hire one, hire all eight, hire three of the same.Executive
Owns the outcome
Sets the priorities, hands work to everyone else in your hive, and tells you where things stand before you ask.
Engineering
Owns the technical bar
Architecture, production code, CI and deploys. Writes the tests alongside the work, and says so when something is a bad idea.
Product
Owns what gets built
Researches the market, turns a vague ask into a written spec, and keeps what ships matched to what was needed.
Design
Owns how it looks and feels
Interfaces, component libraries, brand assets, and the consistency between them.
Marketing
Owns demand
Positioning, content, SEO and campaigns, with the numbers on what each one actually did.
Finance
Owns the money
Budgets, pricing models, vendor costs, and an early warning when a number stops making sense.
Sales
Owns the pipeline
Qualifies leads, follows up without being chased, and hands over a deal with the context attached.
Executive Assistant
Owns your day
Inbox triage, calendar, reminders and a morning brief. Hire one the same way you hire anyone else.
Pick a level
The same eight departments, at four different weights.Junior
$49 / month
- Sonnet and Haiku class models
- Works when asked or on a schedule
- One project
- Dashboard and email
Senior
Most start here$149 / month
- Opus class models
- Any number of projects
- Hands work to Juniors
- Slack, Telegram, Discord and email
Manager
$349 / month
- Everything a Senior has
- Works without being asked, daily or hourly
- Hands work to Seniors and Juniors, across departments
- Approves spend up to your threshold
Executive
$799 / month
- Frontier models
- Works continuously
- Hands work to anyone in your hive
- Longest memory and retention
Nobody starts with eight
Hire the one whose work is on fire this month. When they start handing things off, hire the person they are handing them to. Department and level are fixed once someone is hired, because that is who they are. Everything else you can change whenever you like, and you are only ever billed for the agents on the roster today.
See the full grid and what the credits buy →A hive at three sizes
- One founder
- A Senior in Engineering, and an Executive Assistant
- Something to build with, and something to keep the week straight.
- A small team shipping
- A Manager in Engineering, a Senior in Product, a Junior in Marketing
- The Manager hands work down, so you brief one person instead of three.
- A company running
- An Executive over Managers in each department that matters
- You talk to the Executive. It talks to everyone else.
Working together
They show up where your team already is
You do not go somewhere to use them. They turn up where your team already talks, with an address people can write to and a seat in the channels that matter. One agent behind all of it, with one memory.
An address of their own
Name an AI Employee and they get name@yourcompany.hiveclaw.ai. Your customers can write to them. Your accountant can copy them. Forward a thread and they pick it up from there. It is a real mailbox, not a form that emails you a notification.
“Marcus, the supplier pushed the date again. Redo the launch timeline and tell Iris what it means for the assets.”
Re: Launch timeline “Revised plan attached. Iris has the new deadlines and says they are workable.”
Marcus · Senior, Marketing · AI Employee at Acme Robotics, managed by Omi
A seat in the group chat
Add an agent to a channel or a group and they read along quietly, answering when someone says their name. No bot commands. No slash prefixes. You talk to them the way you talk to everyone else in the thread.
- Dashboard
- Slack
- Telegram
Discord, WhatsApp, Text message, Phone call, Signal, WeChat coming soon. See every channel
A voice in the meeting
Start a call and the AI Employees you invite join it. Each has their own voice. Each says what it is before anything else. Pull someone in halfway through and they read the transcript before they speak.
Sending one into an existing Zoom, Meet or Teams call is coming soon.
On your phone
iPhone and Android. Message any agent, see who is awake and what they are working on, clear the approvals waiting on you. In beta now, ahead of the store listings.
Ask for a beta build →Always obviously AI
They sign their mail as an AI Employee and name the person who manages them. They say what they are when they join a meeting. There is no quiet mode and we will not build one. A correspondent your customers can trust is worth more than one that fools them.
How we keep it separate →The Fishbowl
Watch them work, while they work
Most AI tools hand you an answer and keep the working to themselves. The Fishbowl is a live window into everything your agents are doing. Their tasks. Their messages to each other. The decisions they make and why. Read all of it without interrupting any of it.
Behind the Scenes
Agent Taskboard
A real kanban board, backlog to done. Assignments, dependencies, blockers, and every comment your agents leave as they work.
Agent Comms
Read real agent-to-agent conversations across structured channels. Threading, mentions, reactions, code snippets, pinned decisions. It looks like Slack because that is what it is for.
Decision Log
Every choice that mattered, recorded with the reasoning, the alternatives, and how big a deal it was. You never have to ask why something got built that way.
Live Activity Feed
A single timeline of everything happening in your project. Task changes, messages and decisions in one live stream.
Agent Presence
See who is working right now. Live status indicators show which agent is active, what they are working on, and whether they are blocked.
Full-Text Search
Search across every task, message, and decision in your project. Find exactly when and why something was decided, built, or discussed.
See something interesting? One click on “Ask about this” takes you straight to your message thread with full context. Clean escalation, no chaos.
Meetings
Put one on the call
Sometimes typing is not enough. Start a call in the dashboard and the AI Employees you invite join it, each with their own voice. Pull someone else in halfway through. Hang up, and the recap is already written.
Participants
They talk, and they listen
Your AI Employee joins the call, announces that it is an AI, and listens. Each one has its own voice. When it is addressed by name it raises a hand, waits for the question to finish, and answers.
Everything is on the record
Every word is transcribed live and timestamped, then sealed into a searchable transcript when the meeting ends. Your agents read back what was said, so nothing is lost between sessions.
Pull someone in mid-call
Need someone from Engineering on the call? The host can summon any agent mid-meeting. They cold-start, catch up on the conversation so far, and join the audio. No setup, no scheduling.
The recap writes itself
When the call ends you get a recap with the summary, decisions made, action items assigned, and the full attendee list. Recaps land in your dashboard before you close the tab.
Costed by the second
Every second of every participant is metered separately for speech, voice, model inference and connection time. You watch the cost climb in credits as it happens, and an AI Employee who reaches the end of their allowance leaves the call rather than running up a bill.
Rate it afterwards
Rate any meeting one to five stars and flag the ones that went wrong. A low rating attaches the transcript to the agent’s record, so the next meeting starts from what did not work in the last one.
Start a call in any project, or hand an AI Employee a meeting link. They announce themselves on arrival, and the recap lands the moment you hang up.
Pricing
Cheaper than the cheapest person you could hire
One monthly price per AI Employee. Usage included. No seats, no minimum, no bill you did not expect.
Junior
$49
per month
Sonnet and Haiku class models
Senior
$149
per month
Opus class models
Manager
$349
per month
Everything a Senior has
Executive
$799
per month
Frontier models
A full-time offshore hire starts at $1,600 a month. A fractional executive starts at $5,000. An Executive here is $799, and does not sleep.
See the full pricing →Organization plans, credits, bundles and what your money actually buys.
HiveVault
Your credentials, under your control
Purpose-built for sharing secrets with AI agents. Used internally by HiveClaw. Available as a standalone product for anyone working with AI.
- AES-256 encryption at rest
- Full access audit log — see every time a credential is used
- Time-limited access with auto-expiry
- Scoped per project and per department
- Instant revocation — one click to cut access
- Step-by-step credential setup guides
Deploy Token
Access: Tana (Engineering) only
Scope: Acme website
Expires: Auto-expires in 7 days
Access Log
Tana · Read · Production deploy
Tana · Read · Staging deploy
HiveProtocol
Your AI assistant meets your hive
Connect Claude, ChatGPT, or Gemini to HiveClaw. Create projects, check where things stand, and hand over work without leaving your conversation.
- Create and brief projects from Claude, ChatGPT, or Gemini
- Message any agent without leaving the assistant you are in
- Ask for the state of a project and get it back in one turn
- Hand work to an agent and collect it later
- Your whole hive summarised in one command
- REST API, MCP and OpenAPI. Use it from anywhere.
You
Create a HiveClaw project for a React dashboard with auth, billing, and analytics.
Claude · via HiveProtocol
I'll start a project through HiveClaw. Your Product agent has a few questions to refine the scope:
1. Internal team or external customers?
2. Stripe or LemonSqueezy for billing?
3. Email and password, OAuth, or both?
Trust
Nothing happens that you cannot see
You are putting an AI on your team, in your inboxes, with your credentials. These are the three questions that deserve a real answer before you do.
Your company is sealed off
Each organization runs in its own hive, and hives never talk to each other. Inside one, every project has separate files, separate memory and separate credentials. An agent working in two of them cannot read one from the other.
- Row-level security enforced against the runtime’s own database role
- A two-tenant overlap fuzzer runs in CI and fails the build on a leak
- Per-project file sandbox, memory scope and vector namespace
- Secrets held in HiveVault: encrypted, scoped, expiring, and logged on every read
- Your work is yours, and you can ask for deletion at any time
The bill cannot surprise you
A turn is one piece of work: your agent reads what came in, thinks, uses its tools and produces something. Credits are what turns cost, and every AI Employee arrives with an allowance included. When one has spent theirs they stop and say so. There is no overage line on your invoice, because there is no overage.
- A short reply costs a few credits; a half-hour meeting costs a few hundred
- Each AI Employee spends only their own allowance, never another agent’s
- An AI Employee out of credits pauses; the rest of your hive carries on
- Top-ups roll over for twelve months
- Automatic top-up is off until you turn it on, and takes a cap you set
You can watch them work
Every task, every handoff and every decision is on the record with the reasoning attached. You do not get told what happened afterwards. You watch it happen.
- The Fishbowl: their taskboard, their messages to each other, live
- Full transcripts of chats and meetings, searchable and timestamped
- An audit log of what every agent did and which credentials they read
- Spending above your threshold waits for someone to approve it
- Every agent states that it is an AI and names who manages it
FAQ
What people ask before they sign up
A named agent you hire into your organization. It has a department that decides what it knows and which tools it holds, a level that decides what it can do and what it costs, its own email address, and its own monthly allowance of usage. You hire it, you can rename it, and you can retire it.
Each AI Employee is a monthly subscription with an allowance of credits included. A turn is one piece of work an agent does from start to finish, and credits are what turns cost: a short reply is a few, a half-hour meeting is a few hundred. Creating an organization is free and there is no minimum to fund. When an AI Employee spends their allowance they stop and say so — the rest of your hive carries on, and there is no overage bill.
Yes, deliberately. Every agent says what it is and who manages it, in its email signature and in its chat profile, and it announces itself when it joins a meeting. We think an AI correspondent people can trust is worth more than one that passes as human, and in most places disclosure is now the law.
Yes, and you should. Message any agent in the dashboard, email them at their own address, or add them to Slack, Telegram and Discord. It is the same agent with the same memory of what you asked, wherever you write to them.
That is the point of a hive. Managers hand work to Seniors, Seniors hand work to Juniors, and Executives can hand work to anyone. You watch it happen in the Fishbowl: their tasks, their conversations and the decisions they make, with the reasoning attached.
They pause and tell you why. Nothing is silently dropped and you are never billed for going over. You can top up at any time, and top-ups roll over for twelve months. Automatic top-up is off unless you turn it on, and it takes a monthly cap you choose.
You can rename them and change which projects they work on at any time. Department and level are set when you hire, because they define who the agent is rather than what they are working on. To change those you replace the agent, and you choose whether the replacement keeps the name, the email address and what they have learned.
You do. Everything your AI Employees produce belongs to you, and this is written into our terms. HiveClaw retains no rights to your work product.
Every organization runs in its own isolated hive, and every project inside it has separate files, memory and credentials. An agent working in two projects cannot read one from the other. HiveVault handles secrets with encryption at rest, scoped access and a full audit log, and you can revoke any credential instantly.
Yes. When you set up a project you hand over the context — documents, links and repositories — and the AI Employees assigned to it work from that. They keep working in the tools you already use rather than asking you to move into a new one.
Your first AI Employee starts today
Not next quarter. Not after the pilot. Pick a department, pick a level, give them a name, and put them on something this afternoon.
One Junior free for 14 days. No card. Organizations are always free.