Weavz
Verwalteter App-Zugriff für KI-Agenten — eine gehostete Laufzeitumgebung, die über 1.000 Apps (12.000+ Tools) durch Code Mode MCP bereitstellt, mit Human Gates-Genehmigungen, benutzerspezifischen Anmeldeinformationen, Sandbox, Dateien und dauerhaftem Zustand.
Dokumentation
Connect your AI to the apps you use.
Connect your AI to 1,000+ apps. Bring your own accounts and reusable instructions, or build the same app access into your product.
Start free. Unlimited connected accounts and MCP Workspaces.
See how a task comes togetherExplore tasks
Your request
“Get me ready for my next customer call.”
Gmail Recent conversation
HubSpot Account context
Customer call brief
The context, together.
- The customer’s latest questions
- Open commitments and next steps
- Links back to the source records
Your connected AI uses the apps and permissions you configure. Add approval rules when you want a person to review an action.
[
For your AI
Connect Claude, ChatGPT, Cursor or another compatible client. Authorize your apps and start with a task.
Connect your client
For your product
Let customers connect their accounts. Give your agent scoped tools through MCP, the API or SDKs.
Explore the developer platform
Your apps give AI access.Your Skills give it direction.
Save how you prepare a brief, review a reply or organize research. Ask your AI to read the skill, then use the apps you have connected.
SkillsExample library
Task → Read instructions → Use connected tools → Result
For me
Reusable instructions
Customer call brief
Walk into the call with context.
- Read recent conversations and account notes.
- Separate commitments from open questions.
- Link to the records behind the brief.
Give your AI reusable instructions for the work you do most.
When the work needs a screen, give your AI a computer.
A cloud desktop for browser, file and terminal work. Watch the screen, take over with separate authorization, then return control when you are ready.
Cloud computerExample session
Research workspace
Research, ready to review.
Open a source. Check the details.
Save a note for the next step.
research-notes.md
Agent has control
An authorized person can watch, then take over when needed.
Watch the work, take control when needed, and return control to your AI.
Pick a first task.
See the apps, example request and setup steps before creating a workspace. Connect the accounts your task needs, then continue in your agent.
Prepare an account follow-up
An account follow-up draft and proposed CRM changes.
- HubSpot
- Salesforce
- Pipedrive
- Apollo
Sales CRM Copilot · Preview setup
Explore a citation trail
A literature shortlist with citation context and links to the original papers.
- Semantic Scholar
- PubMed
- Zotero
- Google Drive
Research Workspace · Preview setup
Draft a support response
A sourced support draft ready for a support agent to review.
- Zendesk
- Intercom
- Gmail
- Slack
Support Ops Center · Preview setup
Triage a GitHub issue
An issue triage note with links to related engineering work.
- GitHub
- Linear
- Slack
- Notion
Engineering Agent · Preview setup
Review a campaign
A campaign review for the marketing team to assess.
- Mailchimp
- HubSpot
- Google Sheets
- Gmail
Marketing Growth Studio · Preview setup
Prepare a daily brief
A daily brief that brings meetings and important messages together.
- Filesystem
- State KV
- Agent Memory
- Agent Scratchpad
Everyday Agent Workspace · Preview setup
The apps your work runs on.
Email, CRM, chat, tickets, databases and documents. Explore the actions, triggers and authentication each integration supports.
- Slack
- Google Sheets
- Gmail
- Google Calendar
- Discord
- GitHub
- Notion
- OpenAI
- Linear
- HubSpot
- Airtable
- Asana
- Figma
- Twilio
- Telegram
- Shopify
- Salesforce
- Dropbox
- Google Contacts
- Jira
- Trello
- ClickUp
- Todoist
- Monday.com
12,000+ agent tools, 1,400+ triggers and 14 built-ins across the catalog. Each MCP Workspace exposes the apps and actions you configure.
Need a different app? Tell us what you need.
Start free. Pay for the capacity you need.
Unlimited connected accounts and workspaces on every plan. Choose your action capacity and support level.
Free
$0/month
Connect your apps and try a first task.
- 20,000 actions/month
- 2 team members
- Community support
Pro
$29/month
More capacity for the work you do every day.
- 200,000 actions/month
- 5 team members
- Email support
New-account trial · no credit card needed
Team
More capacity
$129/month
Shared capacity and support for your team.
- 1,000,000 actions/month
- 25 team members
- Priority support
New-account trial · no credit card needed
Compare plans and included capacity →
A few questions before you connect.
Start with your own AI, or build app access into your product. Here is how the pieces fit together.
What is Weavz?
Weavz connects AI clients and agent-based products to 1,000+ app integrations. It provides account connections, MCP tools, reusable Skills and cloud Computer Use, with optional approval rules, files, state and activity history. Use it with your own AI or embed app access into the product you build.
How do I connect ChatGPT or Claude to my apps?
Add the Weavz remote MCP connector, sign in, choose a workspace, and connect the accounts the agent should use. Enable the approved tools in your AI client. Connector availability and setup depend on your client and account.
How do developers embed customer integrations?
Use the REST API or TypeScript and Python SDKs to create a workspace, add integration aliases, register end users, and generate Hosted Connect links. Each customer connects their own accounts. Your agent can then use scoped MCP access or execute approved actions through the API.
What can I set up from a template?
Templates prepare a workspace for a real agent job, such as customer follow-up, issue triage or a research brief. Preview the apps, example requests and setup steps, then connect your accounts and review permissions. A template does not install an autonomous agent or a running schedule.
What does Code Mode MCP do?
Code Mode exposes three primary tools: weavz_search, weavz_read_api, and weavz_execute. Agents discover available actions, read their typed inputs, and run them without loading every action schema at the start. The workspace controls which apps and actions are available.
How can people review sensitive agent actions?
Tools are available by default. Add an approval rule when a particular action needs review or should be blocked. A configured reviewer approves or rejects each matching request. Saved inputs are optional reusable values; they do not grant access.
What are Skills in Weavz?
Skills are optional reusable task instructions. Keep them personal or share a separate copy with people allowed to use your workspace. An agent searches names and descriptions, then explicitly reads a selected skill. Reading a skill does not run actions, connect accounts or grant permissions.
Can I start with a free plan?
The Free plan includes 20,000 actions per month, unlimited connected accounts and MCP Workspaces. Paid plans add capacity and features. Finite limits are prepaid capacity caps with no automatic overage charges.
Put your agent to work. Keep control.
Start with your own apps, or give your customers governed access from your product. Free to start. Upgrade for more action capacity.