Uses real data
The agent searches products and variants from your Wantly catalog instead of guessing from a prompt.
MCP for AI agents
Let your AI agent work inside Wantly.
Connect ChatGPT, Claude, Cursor or your own agent to Wantly through MCP. Your agent can work with your real product catalog, pricing, contacts and proposals — not just generate text.
Sign in with your Wantly account
Scoped permissions
Dry-run before changes
ChatGPTYour AI assistant
Connected
ClaudeYour AI assistant
ConnectedCursor / Claude CodeAgentic workspaces
ConnectedMCP
W
WantlyProducts · Catalog · Pricing · Proposals
Create & import productsOrganize catalogCreate & update proposalsUpdate pricing
More than another AI chat
The agent doesn’t just answer. It can do the work in Wantly.
MCP standardizes how AI applications connect to tools and business systems. Wantly exposes controlled tools for your catalog, contacts and proposals so your agent can take action with real business data.
Works with your pricing
Proposals are prepared from your Wantly catalog, templates and existing data, so the process stays inside your system.
Can take actions
Create or update a proposal, contact or catalog data according to the permissions you grant.
Try the agent
See what working with Wantly through ChatGPT or Claude could actually feel like.
Choose an AI assistant and a sample request. The demo shows the conversation, Wantly MCP actions and the result — similar to how an agent would actually execute the task.
What can I help you do in Wantly?
Choose an example or type your own request.
W Wantly MCP
Interactive demonstration. No data is sent to Wantly.
What your agent can do
From your existing product data to a clean catalog and a ready proposal.
Your agent can do more than find a product or create a proposal. Through Wantly MCP it can create and maintain catalog structure, products, variants, pricing, images, contacts and proposals — according to the permissions you grant.
Import your existing products
Give the agent an Excel, CSV, supplier price list or other structured source. It can prepare the data and create products, SKUs, prices, categories and attributes in Wantly through MCP.
Clean and structure your catalog
Find uncategorized products, create or rename categories, merge duplicates, clean up tags and propose a clearer catalog structure.
Normalize variants and attributes
Turn separate models, colors or sizes into one variable product and normalize shared attributes such as Color, Size and Material.
Work with pricing
Update sale and cost prices, quantity tiers, supplier pricing and variation-level pricing — with dry-run previews before applying changes.
Use proposal templates
Find the right Wantly template and use it as the basis for a new proposal while keeping your structure and design.
Create proposals
Select the contact, products and template, add quantities, variants and custom lines, and create a proposal draft in Wantly.
Update proposals
Change products, quantities, variants, lines, pricing or other proposal details without rebuilding the proposal from scratch.
Work with contacts
Search, create and update client contacts and use them in the proposal workflow without duplicate manual entry.
Connect your agent
One MCP endpoint. Different AI clients.
Use Wantly sign-in where OAuth is supported, or a scoped API key for developer tools and your own agents.
ChatGPT
Add the Wantly MCP server as a supported custom MCP / connector connection and sign in with your Wantly account.
No API keyClaude
Claude.ai, Desktop or mobile can connect to Wantly MCP through your account.
No API keyClaude Code / Cursor
Use the Streamable HTTP MCP URL with an API key, or OAuth when your client supports it.
For developersYour own agent
The same actions are available through MCP and JSON API v2, so you can choose your own agent architecture.
MCP + APIYour MCP endpoint
https://YOUR-TENANT.wantly.eu/api/mcpControl & safety
Give the agent only the permissions it needs.
Agents can take real actions, so the important part is not only connectivity but control. Wantly MCP supports user-based sign-in, scoped API keys and dry-run previews before changes.
View API v2 + MCP documentation↗OAuth sign-in
In supported ChatGPT and Claude setups, the user signs in with their Wantly account and grants permissions.
Scoped permissions
Separate catalog read/write, proposal creation, sending and other actions.
dry_run
The agent can preview what would change while the operation is rolled back before a real execution.
First-party MCP
The agent connects directly to your Wantly tenant MCP endpoint rather than through a third-party aggregator.
✓
Best practice: keep sensitive actions such as sending or deleting proposals behind an explicit human approval in your AI client.
MCP and API
One action layer — for agents and your integrations.
Wantly API v2 actions are available both as a JSON API and as MCP tools. Your agent and your backend can use the same underlying business actions instead of maintaining two separate integration models.
Claude Code
claude mcp add --transport http wantly \
https://YOUR-TENANT.wantly.eu/api/mcp \
--header "Authorization: Bearer YOUR_API_KEY"Use a scoped API key. Never paste your API key into a public prompt.
FAQ
Common questions about Wantly MCP
What is MCP?
Model Context Protocol is an open standard that lets AI applications connect to external tools and data sources through a standardized interface.
Can ChatGPT and Claude create Wantly proposals?
Yes. Through a supported MCP connection, the agent can search contacts and products, use proposal templates and create or update proposals according to the permissions granted.
Do I need an API key?
For supported ChatGPT and Claude OAuth setups, no API key is required. Claude Code, Cursor or custom agents can use a scoped API key.
Can an agent change data without control?
Permissions depend on the user or API key scopes. Write access can be limited, and changes can be previewed first with dry_run.
AI agents + Wantly
Want your agent to prepare proposals from your real business data?
We’ll show how MCP can work with your catalog, pricing, contacts and current proposal workflow.
