Skip to content
Tulina
Book a demo

The collaborative platform for agentic automation

Book a demo

In production at

PartooAtlasSeeftMonte Carlo CapitalOQY Labs

Connectors, administered

A connector set up once works for the whole team. An admin decides who can use each connector. Tulina also brokers API access: when you have no key for a service, a process can use Tulina's, and the usage is billed in credits.

Claude

Mistral

ChatGPT

Gemini

Salesforce

MCP

HubSpot

REST

Linear

GraphQL

LinkedIn

Auth

Notion

MCP

Apollo

REST

Browse all 90+ connectors

Hosted agents

Runs on a schedule or a webhook, with your laptop closed. Describe the process in plain language. Tulina builds it, runs it from trigger to finish on its own infrastructure, and keeps running when you close the chat. Nothing to host, nothing to maintain.

Outreach to book calls+3Qualify dealsFollow up with leads+2Outreach to visitors+1Customer supportCollect product feedbackLeverage your network
Browse every process

Runs you can read

Every run, step by step, with what it cost. A failed step is visible, and comes back rewritten. Each run shows the tools it called and the credits it spent, so the month's cost is known before the invoice. Call recordings, Slack threads, email, and closed deals keep folding in as your team works.

A monthly lost-deal revival run stops when the signal scrape times out, the fix is written back into the process, and the next run passes; a month later the batch comes back, and the deals reopened climb after the process rewrites what each draft leads with and which contacts it flags. Every step and what it cost are on the run's page.

Business operators need to automate their work. They know what they want to achieve, but they don’t know how to build it, and they lack the infrastructure to make it happen.Tulina.ai is the automation infrastructure for teams and AI agents: hosted agents that run processes across their stack, administered from one place, with every run on the record.

Build your agents from Claude

Add the Tulina MCP to the AI account your team already uses and create a process in the chat. Hand it to Tulina and it keeps running on a schedule or a webhook, with the chat closed.

Custom connectorLive

https://mcp.tulina.ai/mcp

Claude, ChatGPT, Mistral, and the 90+ tools your company already runs. How to connect it

The three systems behind a hosted agent.

System of execution

It runs on its own

Put it on a schedule or behind a webhook. Tulina runs the process on its infrastructure and reports back when it's done, whether your laptop is open or not.

System of administration

It stays administered

Connectors, processes and skills live in the workspace. An admin decides who uses what, keys stay in the vault, and the data Tulina buys for you is billed in credits.

System of record

It keeps a record

Every run lists its steps, the tools it called and what it cost. A failed step shows on the page, with its fix.

Your whole team will benefit from

Shared context

  • Skills and memory, shared.

    Built once, in the workspace, not in files on one laptop. Everyone runs the same version, and a fix reaches the whole team.

  • Connectors, set up once.

    One connection serves the whole team. An admin decides who gets access to which one.

  • Secrets, one click.

    Credentials and API keys set up once, held in a vault. Nobody pastes a key into a chat again.

  • Processes as a graph.

    A process the team can read and a client can follow, without opening the code behind it.

  • Tables the team works in.

    Shared tables where the team and its clients review, edit and approve, before anything reaches the CRM.

Common questions

Tulina runs AI agents for your team, hosted, on a schedule or a webhook. It connects to the tools you already use, keeps every credential in a vault, and shows each run step by step with what it cost.

No. Tulina connects to Slack, Notion, your CRM, and the rest of your stack, and works across them.

Access follows roles and teams. An admin decides who can use each connector, credentials sit in an encrypted vault, and anything sensitive stops for approval.

In the European Union, on Scaleway in France: the platform, its database, its object storage and the credential vault. Your working data is processed there.

Your own API keys run without limits. When Tulina buys data on your behalf from a provider you have no key for, it is billed in credits and shown on the run, so the month's spend is visible as it happens.

Your first agent gets built on the intro call, from two of your real tools. Rollout to a team usually starts the same week.