Małgorzata WachockaInfection prevention · Digital · AI

Clinical AI Lab

Prevention systems, built with the ward.

Seventeen years of hospital practice, joined with a team of software and agentic engineers. Together we design, pilot and deploy monitoring and AI tools that infection-prevention teams actually use.

Why domain first

Most hospital AI fails on the ward, not in the code.

Engineers can build a dashboard in weeks. Knowing which outlet matters, which alert should wake someone up, and what a ward will really do with it takes years inside a hospital. The Lab puts both in one team from day one.

DomainMałgorzataHospital hygiene, water, air, surfaces, devices, hands, audit, regulation
EngineeringSoftware and agentic engineersData integration, agents, apps, cloud, security
OutcomeSystems for hospitalsMonitoring, copilots, audit and reporting tools in daily use

Concept demo

A morning on the prevention console.

How the four Lab solutions could come together in one hospital. Agents watch water, air, audits and reporting, and draft actions. People approve. Try approving an action.

Prevention console Concept demo · illustrative data · not a live system 08:00
Ward 7 ICU OR 3 Haematology works zone Endoscopy Laboratory Plant room W14W15I02I03O01E01E02P01P02

Water outletAir sensorWorks zone

Agent feed awaiting human approval

    Air · Haematology works zone —

    Water · PCR screen —

    Hand hygiene · ICU audits

    Weekly audit score, illustrative

    What we build

    Four solutions, one risk system.

    Each solution pairs what I bring from the ward with what the engineering team builds. All of them sit on top of systems the hospital already runs.

    Water · Air · Rooms01

    Environmental monitoring

    Water outlets, air particles, construction dust and controlled rooms in one live view, with thresholds that mean something clinically.

    I bring
    Which outlets and rooms matter, sampling logic, alert thresholds and who acts on each result.
    We build
    Sensor and laboratory data integration, dashboards, alert routing to the right team.
    Outcome
    Risk found before patients meet it, and every alert has an owner.
    32 technologies and studies on the capability map →
    Agents · HAI02

    AI surveillance copilot

    Agents that read microbiology alerts, audit results and HAI data, spot patterns early and draft the next action for the infection-prevention team.

    I bring
    What a meaningful signal looks like, how investigations really run, what must never be automated.
    We build
    Agentic workflows over existing hospital systems, with a human approving every action.
    Outcome
    Less time collecting data, more time acting on it.
    31 technologies and studies on the capability map →
    Devices · Hands · Surfaces03

    Audit and checklist automation

    Digital ward audits, device-reprocessing traceability and hand-hygiene feedback loops, captured on the ward and fed back in days, not months.

    I bring
    Validated checklists, audit design and the feedback that changes behaviour.
    We build
    Mobile capture, traceability from patient to device to washer cycle, trend views for each ward.
    Outcome
    Audits teams own, and improvement everyone can see.
    24 technologies and studies on the capability map →
    Regulation · Committees04

    Compliance and reporting agents

    Agents that draft water-safety risk assessments, committee packs and accreditation evidence from live data, ready for expert review.

    I bring
    What regulators and committees actually need to see, and how to evidence it.
    We build
    Report generation from source data, version history, sign-off workflow.
    Outcome
    Reporting that takes hours, not weeks, and stays traceable.
    21 technologies and studies on the capability map →

    How we work

    Discover, pilot, then scale.

    Small, measurable steps. Nothing goes hospital-wide until it has worked on one ward.

    1. 01

      Discover

      2–4 weeks

      A clinical discovery sprint: map the workflow, the data sources, the risks and what success should look like on the ward.

    2. 02

      Pilot

      8–12 weeks

      Build on one ward, one water system or one process, with the people who will use it. Measure against the baseline.

    3. 03

      Scale

      Ongoing

      Roll out across the hospital, integrate with existing systems, train teams, and keep improving with real use.

    Domains covered WaterAirSurfacesDevicesHands Surveillance

    Guardrails

    Safe by design, not by promise.

    Infection prevention is patient safety. These principles apply to everything the Lab builds.

    • Human in the loop

      Agents draft and flag. People decide, and every action has a named owner.

    • Built on what you have

      Laboratory, building-management and sensor systems you already run, not a parallel world.

    • Privacy by design

      Patient data minimised and protected from the first sketch, in line with GDPR.

    • Regulation checked early

      Whether a tool falls under medical-device or AI rules is assessed before it is built, not after.

    • Measured on the ward

      Success is fewer risks and faster action, not dashboard logins.

    • Vendor-neutral

      No product endorsements. Tools are chosen for the hospital, not for us.

    Have a ward, a water system or a process in mind?

    Start with a two-to-four-week discovery sprint. We map the problem, the data and what success would look like, then decide together whether a pilot makes sense.

    Propose a pilot →Explore the capability map