Why Headless: No Head, All Brain

If you manage research projects, you don't just live in a research management system like turnKeyRM. You live in MS Teams, in email, documents and in meetings. The system is where the answers are, but getting them out means logging in, finding the right tab, running the right report, exporting it, and pasting the result into the conversation where the question was actually asked. The data is good. The distance between you and it is the problem.

Headless is the name for closing that gap.

The overarching platform of any business has two parts. The head is the user interface, the tabs, screens, buttons and clicks. The brain is everything underneath: the data model, the automation, the business rules, the permissions. Headless means the brain keeps working while you talk to it through whatever surface you're already in, instead of through its own screens.

What makes this practical right now is the Model Context Protocol, or MCP. MCP is an open standard that lets AI assistants like Slackbot, Claude, MS Copilot, ChatGPT connect securely to tools and data sources. A useful analogy is USB-C: one common connector, so any capable assistant can plug into any system that speaks the protocol, without a bespoke integration for every pairing.

Salesforce, which owns Slack, has been investing heavily in agentic AI and has announced MCP servers that bring AI and live Salesforce data directly into Slack. This is both novel and core to us at Foveal, because our product turnKeyRM is built natively on Salesforce. Everything a research organisation runs through turnKeyRM, projects, milestones, budgets, approvals, governance records, lives as structured Salesforce data. Which means it is exactly the kind of brain an AI can talk to.

A research manager can type a question in Slack: which projects have milestones due this quarter?

Through an MCP connection to their Salesforce instance, Slackbot queries live turnKeyRM records and answers in the channel, projects, milestone dates, responsible leaders. No login, no report builder, no export.

They follow up: summarise spend against budget on Project X. Slackbot reads the actual finance records and gives the position. Then the one that usually eats an afternoon: draft a progress update for the funder. Slackbot pulls the project's milestones, status and financials from turnKeyRM and drafts the update, grounded in the real numbers rather than someone's memory of them.

Nothing about the underlying system changed. The brain is the same. We can just stop walking over to its head.

The obvious worry with AI reading your data is control, and this is why being built on Salesforce is critical. An MCP connection into a Salesforce instance operates inside that org's existing security model. The permissions, sharing rules and audit trail that govern what a person can see in the UI govern what the AI can see on their behalf. There is still one source of truth, held on a platform research organisations already trust with funding, IP and governance data.

That's the real meaning of Headless: the AI is not a copy of your system or a workaround bolted onto it. It's a new head on the same trusted brain.

Salesforce's direction is clear: agents that don't just answer questions but take action, and open protocols like MCP as the way those agents reach data. For research management, the near-term win is already substantial. The distance between a question asked in Slack and an answer grounded in live project, finance and milestone data shrinks to a single message.

We built turnKeyRM so research centres, CRCs and universities could run projects, funding, governance, IP and reporting on one platform, with branded portals for project leaders. The headless story is the same investment paying off again: because everything is structured Salesforce data, it's ready for whatever head comes next.

If you're wondering what your research data could do with a better conversation, we'd be glad to talk.