AI-agent infrastructure for healthcare practices
Documentation support, intake planning and document research with a fully self-hosted AI agent: no patient data ever leaves the building.
A healthcare practice by definition works with information that must never end up anywhere else: patient files, intake conversations, documentation covered by medical confidentiality. The very thing that makes AI so valuable for healthcare is structuring and searching large volumes of text at speed. That's exactly why "just paste a note into ChatGPT" isn't an option in this line of work. Every prompt to an external AI provider is patient information leaving the building.
That's why we set it up the other way around: not the data going to the model, but the model coming to the data. On the BedrijfsLab, the language model itself runs entirely on hardware that stays yours. For sectors where confidentiality genuinely matters, such as healthcare or law, that's often the deciding reason to choose self-hosted: no patient conversation, no intake form and no note that has ever touched a server belonging to an American AI provider.
For the technically curious
The language model runs via Ollama, llama.cpp or vLLM, set up on hardware at your own premises.
The concrete use case
On that self-hosted foundation, we deploy an agent with its own, clearly bounded task, the same approach as our broader AI-agent infrastructure. For a healthcare practice, that means for example:
Documentation support
Consultation notes and dictation get turned into a structured, complete note following your own template, so a practitioner spends less time on admin after each consultation.
Intake planning
We've already built this pattern once, in generic form: a system where a team reports weekly how much room there is for new intakes, so a planner sees capacity at a glance. For a healthcare practice, we build this out around your own waiting list and referral flows.
Document research
Referral letters, prior files and sources get searched and summarized on the specific point a practitioner is asking about, without anyone having to manually work through a complete file.
In every case, the same principle applies as with our broader unmanned office: if something falls outside the agreed boundaries, the agent escalates to a human instead of making a medical judgment call itself. The agent takes the preparatory work off your hands, not the treatment decision.
Why this is safe with patient information
The approach we set up every agent with is particularly well suited to work covered by confidentiality obligations:
A branch and pull request per change
Nothing gets applied directly; there's always a readable trail of what changed and a clean point to roll back to.
Isolated working copies
An agent working on a file does so in its own sandboxed environment, never in a copy someone else might be working in at the same time.
An independent check
A second, independent step verifies the actual result, not just the agent's own summary of what it thinks it did.
A fixed list of things asked first
A human stays ultimately responsible for every medical judgment call; the agent takes the groundwork off your hands, not the decision.
For a healthcare practice, that's not an incidental detail. It's the core of it: the same diligence you'd expect of yourself with a patient file applies to the agent looking over it too.
These safeguards are also summarized on a single page, to pass along to whoever needs to weigh in internally, for example a quality officer or practice owner who doesn't visit this site themselves: download the approach as a PDF.