1Ask in plain words
From Slack, Teams, Excel, or AMVY itself.
Marta Gómez08:00
@AMVY Review the pipeline and flag the deals at risk this month.
Build agents that know your work
— finish it, learn, and improve.
The multiplayer AI workspace: your team asks, agents do the work in your tools, and a person approves what matters.
Release readiness · Shared
3 online
Worked for 2m 14s
8 pull requests reviewed against release policy·acme/platform42 required checks verified · 2 risks found·release-policy.mdRelease control room drafted·/a/release-control-room
Ask. Approve. Get the result.
2Approve what matters
Agents work in your tools and stop before anything sensitive.
Update next steps on 5 deals in HubSpot?·HubSpotDecides: Marta · sales lead
3Get a result your team can use
A report, an update, or a live page, not a wall of text.
Weekly pipeline review
published/a/pipeline-review
Deals at risk this month
- Northwind$48k
- Acme$32k
- Globex$21k
Built for the work every team does.
Real requests from the places people already work. Each run ends with a decision a person approves and a result the team can use.
In Slack · every Monday at 8:00
The sales lead starts the week with a reviewed pipeline instead of an afternoon of spreadsheet work.
Uses
- HubSpot
Fathom
Marta Gómez08:00
@AMVY Review the pipeline and flag the deals at risk this month.
Works where your team already works.
Slack, Teams, Office, your browser, and the tools your business runs on.
Slack and Teams- Excel, Word, PowerPoint
- Your browser
- Voice
- Desktop and mobile
- Your own product
- API and webhooks
Autonomy where it helps. Control where it matters.
A person approves
Sensitive actions wait for someone with the right role.
Learn moreLearns under review
Agents propose what to remember; people decide.
Learn moreTested before it changes
Every change runs against real cases first.
Learn moreYour data, your cloud
SSO, audit, your models, or fully self-hosted.
Learn morePlans for every team.
Start free, then pick the plan that fits your team. Every plan includes shared sessions, approvals, and integrations.
Free
Try AMVY on your own work.
US$0/ user / month
- Members, up to
- 1
- Messages / month
- 100
- Tool calls / month
- 50
- Agents
- 3
- Shared sessions with approvals
- Integrations and projects
- Artifacts and governed learning
Start
For small teams starting with agents.
US$10/ user / month
- Members, up to
- 3
- Messages / month
- 1,000
- Tool calls / month
- 500
- Agents
- 10
- Everything in Free
- Knowledge bases
- Custom agents
Pro
For production agents and automation.
US$20/ user / month
- Members, up to
- 10
- Messages / month
- 10,000
- Tool calls / month
- 5,000
- Agents
- 50
- Everything in Start
- Scheduled tasks
- Higher usage for production work
Teams
For teams that need shared governance.
US$80/ user / month
- Members, up to
- 50
- Messages / month
- 50,000
- Tool calls / month
- 25,000
- Agents
- 250
- Everything in Pro
- Roles and permissions (RBAC)
- Audit logs and team administration
Enterprise
For organizations with security and deployment needs.
Custom
Limits sized to your organization.
- Everything in Teams
- SSO and SCIM provisioning
- Dedicated or self-hosted deployment
- Custom limits and terms
Prices in US dollars per user, billed monthly for each member of the organization. Applicable taxes are calculated at checkout by Creem, our merchant of record. Cancel anytime; you keep your plan until the end of the period. During the beta, accounts are enabled after approval.
What teams ask before they start.
Is AMVY another chat interface?
No. Chat is the entry point, but AMVY coordinates agents, integrations, files, isolated execution, approvals, persistent state, and publishable outputs across a complete run.
Can agents take actions without approval?
Safe reads and low-risk work can continue under policy. Sensitive or mutating actions can be configured to pause for explicit human review.
What is a multiplayer agent harness?
A harness is the runtime around a model: context, tools, approvals, workspace, and recovery. In AMVY the harness is multiplayer, so several people and their agents share one session, with attributed steering, explicit approvals, and safe takeover.
What are self-improving agents in AMVY?
Agents that turn completed work into proposals for memories or skills. Nothing changes silently: people review the evidence and approve, edit, or reject each proposal. It is not model retraining.
Give your team and its agents one place to work.
Tell us how your team wants to use AMVY. We review every beta request and reply by email.