Skip to content

Ask your AI about your partners.

Agent Connect is the Unifyr MCP server, so Claude, ChatGPT, Gemini or Copilot can answer from live partner data with no report to build.

Works in the AI assistant your team already uses

Add Unifyr from your assistant's connector marketplace, or paste one endpoint under Add custom connector. Any other MCP client connects the same way, with nothing to build.

ChatGPT

Gemini

Your partners will love working with you even more.

Registering a deal, finding a pitch deck or creating a campaign takes just a quick chat with Agent Connect.

Today

Deal registration: Every deal starts with another portal, another login and another form, so registration gets pushed to later or skipped altogether.
Collateral: Partners know the right asset probably exists. Finding the right version, for the right deal, is the work.
Program questions: A simple question turns into an email, a ticket or the next status call. The answer arrives when the deal has already moved on.

With Agent Connect

Deal registration: Partners register the deal in a sentence, from the AI tool they already use, and Unifyr gets the record without another login or form.
Collateral: Partners describe what they need and get the right, current collateral back, ready for the deal.
Program questions: Partners ask the program directly and get the answer when they need it, without waiting on your team.

Your channel team will have even more time to grow the program.

No more chasing deals, resending assets or following up on approvals. Agent Connect handles the busywork.

Today

Channel and RevOps: Answering a pipeline question means pulling reports, setting filters and piecing the numbers together.
Channel management: Deals stall unnoticed until someone finds them, often when there is little time left to change the outcome.
Partner enablement: Onboarding moves at the speed of your team's follow-up. The more partners you add, the more there is to chase.

With Agent Connect

Channel and RevOps: Ask the pipeline question directly and get the answer from live program data, shaped around what you need.
Channel management: Ask what stopped moving and why, then take the next action in the same conversation.
Partner enablement: Partners get what they need through the AI tool they already use, without your team guiding every step.

The assistant and the portal work on one partner record

A deal updated from a thread is the same deal your channel manager opens in Leads & Deals, with the change in its history.

Partner & vendorAI client
  • ChatGPT
  • Claude
  • Gemini
  • Copilot
Agent Connect
Partner PortalWhere your partners already log in
UnifyrAI-native PRM & TCMA
  • IQ agents
  • Memory
  • Analytics
  • Deals
  • Onboarding
  • Leads
  • Collaterals
  • Campaigns and MDF
  • Training
  • Incentives
  • Contests

Ask, update and reply from one thread

Reads run freely. Anything that changes a record or reaches a person waits for your approval in the client.

Ask with your own definitions

Agent Connect queries live program data using the definition you give it, such as dormant meaning no deal in 90 days. There is no view to build or filter to set, a follow-up keeps the thread, and the answer can go straight into a deck.

  • Live program data
  • Follow-ups in the same thread
  • No preset filters to configure

The assistant writes to Unifyr once you approve

Read and write access are granted separately. A write, such as a new deal value, is shown in your client before it runs and logged against your own user, in the same history as a change made by hand.

  • Read and write granted separately
  • Every write confirmed before it runs
  • Logged against your own user

One request reads Unifyr, your CRM and your mail

Connect other MCP servers beside Unifyr and one request can use them all: the partner and deal from Unifyr, the end customer from your CRM, and a reply drafted for your own mail client.

  • Partner and deal from Unifyr
  • End customer from your CRM
  • Reply sent from your mail client

The connector sees what your account sees

What is MCP?
The Model Context Protocol, first created by Anthropic and then shared with the world in late 2024. It is an open standard that describes how an AI assistant discovers and uses the capabilities of an external system. The assistant asks the connector what it can do, and the connector responds with a set of capabilities it can then invoke on your behalf. Because capabilities are exposed through the protocol rather than a vendor specific plugin format, the integration is written once and the same connector works in every compliant client, including ones that do not exist yet.
Which AI assistants can connect?
Any client that implements the protocol. Claude Desktop, Copilot and ChatGPT connect the same way, as do agentic AI frameworks and custom agents your own team builds. Building on a standard means you are not maintaining a separate integration for each assistant.
How do we switch it on?
A few steps. An administrator can enable Agent Connect for your whole organization, or each user adds Unifyr in their assistant's connector settings: one click from the marketplace in Claude or Microsoft Copilot, or by pasting the endpoint https://mcp.unifyr.ai/mcp under “Add custom connector”. You sign in at login.unifyr.ai with your own Unifyr account and can directly ask the first question.
Do we need developers or an API key?
No. There is no middleware to deploy and no API client to write, because the protocol handles discovery of what the connector can do. Sign-in happens in the browser, so there is no token to manage.
Can it change data, or is it read only?
Both. Read access and write access are granted separately, and retrieval runs freely. Anything that creates a record, changes a status or sends something to a real person is declared as a write, so the approval prompt in your client tells you what is about to happen before it happens. Throughout, you keep control, because every action starts from your own input.
Can we see what was done through it?
Yes. The connector reads under your own account, so the controls that bound you in the console bound it too. Actions are recorded against the user who requested them, in the same history as work done by hand, and an approval made this way follows the same workflow as one made in admin.
Is this a chatbot inside Unifyr?
No, it works the other way around. A chatbot on top of a product leaves the intelligence in the chat window. Your team stays in the assistant they already use, and that assistant reaches into Unifyr, where the IQ agents analyse your pipeline, score deals and prepare partner communication.
What do we need on our plan?
Agent Connect is included on the Grow and Scale plans, which also carry API and webhook access, vendor and partner single sign-on, and a monthly allocation of AI Service Credits. Credits are the shared currency for the IQ agents and are consumed as agents execute actions. Allocations refresh monthly and do not roll over, and additional packs are available as usage grows.
What can the assistant see?
The tenant you chose when connecting, and within it the data your own account can already reach. The connector does not grant access you did not have; it changes where you can ask from.
What is the endpoint?
https://mcp.unifyr.ai/mcp. Most clients need only that; the setup instructions in the product cover the handful that want more.

Connect it in about a minute

We will add the connector during the call and you can ask it something about your own program.

Get a demo