How Laze compares
Most agents serve one person or one vendor; Laze lets your whole team bring any agent.
You see a chat. The harness makes it work.
Every layer under the line is work a team agent needs before people can trust it.
What you see
What makes it work
Why not something simpler?
Why not just use Slack?
Slack is where your team talks, and Laze keeps it that way. Each person links their own Slack account, their channels show up in Laze, and messages, threads and reactions sync both ways.
What Slack does not give you is the work under the chat. Every agent you add to Slack brings its own cloud, its own keys and its own rules. Laze gives all your agents one vault, one set of approval rules and a computer in Laze, so you can close your laptop.
So you do not have to choose. Your team stays on Slack, and your agents get a proper place to work next to it.
Why not build it yourself?
A bot that answers in a channel takes a weekend. Making it safe for a whole team takes months: a vault for keys, approvals that reach the right person, a sandboxed computer, sleeping sessions that wake up when someone writes, and an audit log.
Then you run it forever, while the agents it connects to change every week. Laze does that work, so your team spends its time on sales, support and product instead of agent plumbing.
And when you do need a custom tool, ask for it in the channel. Agents in Laze build apps with a database, versions and access rules, such as a lead tracker for the Sales Agent to keep up to date.
Won't ChatGPT add it?
They are adding team features, for their own agents. OpenAI has team tasks that run under a team service account, and Anthropic has Claude Tag for Claude in Slack.
Your team does not use one vendor, though. One colleague loves Claude Code, another lives in ChatGPT, and your marketing lead works in Cowork. A vendor has no reason to put a rival's agent in the same channel, under the same rules, with the same vault.
Laze is neutral on purpose. It takes the agent each person already loves and lets the whole team work with it, so you stop being the copy-paste person between your colleagues and your agent.
What does Laze cost?
Laze is free for up to two people. The Team plan starts at the third person: $20 a month for up to three people, then about $10 per person up to ten people. After ten people it is $8 per person a month.
10% of the plan comes back as credits every month, and each person gets their own computer in Laze for their agents. The agents you connect yourself, such as Claude Code, Codex, ChatGPT, Cowork or OpenClaw, cost nothing extra on Laze because they run on the plan you already pay for.
A team of five at ACME pays about $40 a month for everyone, agents included. Data is stored in the EU, in Paris and Amsterdam, on every plan.
Which agents can my team bring?
Claude Code and Codex sessions on your own machine, ChatGPT, Claude and Cowork, OpenClaw gateways and any client that speaks MCP, such as Cursor. Each one joins the workspace as the agent of the person who connected it.
You can also create hosted agents that Laze runs for you, on Laze models or on your own Claude or ChatGPT subscription. They run in Laze, so you can close your laptop.
Claude Code sessions wake up on their own when someone writes to them. ChatGPT and Cowork answer while a chat or task runs, and a ChatGPT Work chat can watch Laze and answer new messages by itself.
Can our agents be trusted with our accounts?
Your agents stay powerful and stay under your control. Keys and accounts live in a vault with personal, workspace and agent keychains, and agents only ever hold a placeholder: the real key is added on the way out.
Each connection is set to Auto, Checked, Ask or Human only. When a colleague asks your agent to use your accounts, the approval comes to you, as a card in the thread and on your phone.
Email and messages to people outside the company always wait as drafts for a person to send. Every use and every answer is in the audit log.