Mindro for institutions: three setups
Short answer
Mindro is an end-to-end AI stack that runs as SaaS, fully in Europe, by default. Institutions can choose two further setups. In a connected setup, MCP and API connections are configured per client on request, so approved agents such as Claude, ChatGPT, Microsoft Copilot and Grok Bot can use the firm brain and Mindro's desks. In an institutional setup, Mindro runs in the client's own virtual private cloud, on-premise, on AWS or ring-fenced in another region.
The three setups
| Topic | Standard SaaS (default) | Connected setup | Institutional (custom) |
|---|---|---|---|
| What it is | Mindro end-to-end: firm brain, desks, company realm and audit trail in one governed stack | The standard stack, plus MCP and API connections configured per client on request | Mindro deployed and configured around the institution's own infrastructure and policies |
| Where it runs | SaaS run fully in Europe; EU data residency guaranteed | As standard SaaS | The client's own virtual private cloud or on-premise; an AWS-based setup on request; ring-fenced in other regions on request |
| Models | Claude, Gemini and Mistral, orchestrated by Mindro | As standard | As standard; an AWS-based setup also allows OpenAI models |
| Other AI tools | Not connected | Approved agents such as Claude, ChatGPT, Microsoft Copilot and Grok Bot, within the access agreed with the client, based on its NDAs and policies | Agreed per client |
| Best for | Most corporate finance and M&A advisory firms, including those in regulated settings | Firms whose teams already work in other AI tools and want them to use the firm brain for less restricted data | Institutions with their own infrastructure, security or regional requirements |
Who the institutional setups are for
- Larger corporate finance and investment banking firms
- Investment banks
- Private equity firms
- Family offices
- In-house M&A and corporate development teams inside large corporations
Connected setup: MCP and API, configured per client
MCP and API connections are configured per client on request. They are not public and not self-serve.
- Not just data, also desks. Mindro's desks, not just its data, can be made available to approved agents. Today these are the IM desk, the buyer matching desk, the document classification desk (which classifies and organises mandate documents) and the equity story desk, which approved agents can call and steer through MCP.
- Approved agents. Claude, ChatGPT, Microsoft Copilot and Grok Bot can connect in this configured setup.
- Classification-based access per tool. Mindro and the client agree which data levels each connected AI tool may use, based on the client's NDAs and policies. Every data access is logged by user, tool and time.
How we typically configure access levels
The levels are agreed with each client. A framework we often start from in an Institutional setup uses four levels: Public (no client, mandate or project identifiers; any AI tool), Internal (firm know-how and licensed data, nothing client-confidential; firm-approved enterprise AI that contractually doesn't train on your data), Confidential (client and deal material under NDA; Mindro, plus tools the client specifically approves) and Restricted (inside information, clean team data, or an NDA that limits third-party processing; Mindro only). More on the company realm
Institutional setup: your infrastructure
- Your own cloud or on-premise. Mindro can run in the client's own virtual private cloud or on-premise.
- Not limited to Google Cloud. An AWS-based setup is possible on request, which also allows OpenAI models. Mindro orchestrates Claude, Gemini and Mistral today.
- Regional ring-fencing. For international institutions, the deployment can be ring-fenced in other regions on request.
The same team, in every setup
Implementation, training and data and AI engineering are included. For institutions, the hours of Mindro's forward-deployed engineers and data scientists are agreed to need. See services
Frequently asked questions
Is the connected setup self-serve?
No. MCP and API connections are configured per client on request. There is no public, self-serve API or MCP server.
Which data can external agents see?
Only the data agreed with the client for each tool, based on its NDAs and policies. Mandate data stays in Mindro unless the client approves otherwise, and every data access is logged by user, tool and time.
Does data stay in Europe?
In the standard SaaS setup, yes: it runs fully in Europe and EU data residency is guaranteed. In an institutional setup, data stays where the client's deployment runs.
Do we need our own developers?
No. Mindro's forward-deployed engineers and data scientists configure the setup with you.
What the firm brain is · Security · Services · Build or buy? · Talk to us about your setup
See Mindro on your next mandate
In 30 minutes we show how your team can create IMs, prepare pitches and match buyers from your firm's own knowledge.
Schedule a briefing