Grow and run
Use your CRM from your AI assistant
Gandiv offers an MCP server so assistants like Claude or ChatGPT can look up leads, summarise a pipeline or draft a follow-up, with access you control and an audit trail.
Read-only default
Write permissions are optional and switched on by you
Scoped tokens
Choose which modules and actions the assistant can reach
Audit log
Every assistant request recorded with user and time
What is mcp and ai assistants?
MCP, the Model Context Protocol, is an open way for AI assistants to connect to business tools. With the Gandiv MCP server, you can ask your assistant, for example, which Hot leads have no follow-up this week or to summarise yesterday's calls for the Pune team. The assistant reads from Gandiv through a token you create and limit, so it only sees what that token permits, and every request is logged.
Benefits
Ask in plain language
Question your pipeline, calls and tasks without opening reports or building filters.
Read-only to begin
New connections can read only. Write scopes such as creating a task are optional.
Limited by token
Each token is limited to chosen modules and actions, and can be revoked in one click.
Respects your roles
The assistant acts as a user or role you choose, so field restrictions and visibility still apply.
Every request logged
The audit log shows which token asked what, with date and time, for review.
Keys stay server-side
Model keys and CRM tokens are kept on the server and never put into the browser page.
In detail
What is MCP
A standard door for AI assistants
MCP lets an assistant use tools in a consistent way. Instead of copying data into a chat, you connect once and the assistant asks Gandiv for exactly the records it needs, then answers using fresh data.
- Open protocol
- Connect once
- Fresh data each time
Open protocol
Connect once
Fresh data each time
Assistants
Works with MCP-capable assistants
Any assistant that supports MCP can connect, including Claude and ChatGPT where MCP connectors are available to your account. Setup is a server address and your token. Availability depends on the assistant and its plan.
- Claude and ChatGPT
- Server address and token
- Depends on assistant plan
Claude and ChatGPT
Server address and token
Depends on assistant plan
Scopes
You decide how far it can reach
Choose modules, fields and actions per token. Keep it to reading leads and calls, or allow creating tasks and notes. Anything not granted is simply not available to the assistant, however it is asked.
- Per-module scope
- Optional write scopes
- Revoke in one click
Per-module scope
Optional write scopes
Revoke in one click
Audit
A record of everything it did
Each assistant request is written to the audit log with the token, action and time. Admins can review what was asked and whether anything was changed, which makes it safe to let managers try it.
- Who and when
- Writes flagged
- Exportable log
Who and when
Writes flagged
Exportable log
How it works
- 1
Create an AI token
Pick the user role it acts as and the modules and actions it may use.
- 2
Add it to your assistant
Enter the Gandiv MCP server address and token in your assistant's connector settings.
- 3
Ask and review
Ask your questions and check the audit log for what the assistant accessed.
Works with
Used in these industries
Frequently asked questions
Can the assistant change my data?
Not unless you allow it. Connections are read-only by default, and write scopes are optional per token.
Does my data go to the assistant provider?
Records the assistant requests are sent into your conversation with that provider, under its own terms. Limit scopes to what you are comfortable sharing.
Which assistants work?
Assistants that support MCP connectors. Whether your plan with that assistant includes connectors is decided by the provider.
Can I stop it quickly?
Yes. Revoke the token in settings and access ends immediately.
Is this the same as the AI suite?
No. The AI suite runs inside Gandiv on your calls. MCP lets an assistant outside Gandiv read your CRM on request.
See Gandiv on your own leads
Start free, or book a short demo and we will set it up with your pipeline.