Healthcare: The Triage Desk — Agentic AI for Africa Innovation Challenge

← Agentic AI for Africa Innovation Challenge

Overview

Walk into most public hospital outpatient departments on the continent and the queue doesn't start at the doctor's door — it starts at triage. One or two nurses, a blood pressure cuff, a thermometer, and a stack of case cards, working through everyone who walks in before anyone gets assigned a priority. On a busy morning that single desk is the bottleneck for the whole department: a patient with a genuinely urgent condition can sit in the same queue as a routine follow-up, because nothing at that desk is telling the nurse to look twice until they physically get to that card.

This isn't about replacing the nurse at that desk. It's about giving her something that flags who needs to be seen next, backed by the vitals and symptoms already in front of her, so the wait itself stops being where the risk is.

60–90 min

Typical wait before a patient is even seen at the triage desk in a busy public OPD

Seconds

What your agent should take to turn a set of vitals and a presenting complaint into a cited priority level

Nurse decides

Your agent stops here and hands over a sourced priority, not a diagnosis

One rule that is not negotiable: your agent scores or sequences the queue, it never diagnoses, never prescribes, and never decides who gets seen — a triage nurse or clinician confirms every priority level before a patient is sent to a bay. If a patient's condition changes while they wait, the agent can raise that for a second look; it cannot act on it. Every priority level must cite the vitals or symptom entry behind it; an unsourced urgency flag is worse than no flag at all.

Choose one theme

Symptom-based priority scoring
Presenting complaint and vitals turned into a defensible priority level, grounded in an existing triage scale rather than invented from scratch.
Queue & flow optimisation
Matching patient volume and staff availability to cut time-to-first-assessment, not just re-order the same queue.
Deterioration watch
Flagging a patient whose condition may be worsening while they wait, from a re-check or a patient-reported change — for a second look, not a verdict.
Referral & department routing
Getting a patient to the right department queue the first time, cutting the bounce-backs that quietly lengthen every other queue.

Jargon, defined up front

MCP — Model Context Protocol: An open standard for exposing tools to an AI model. You write a small server; any agent can call its tools.
Agentic: Plans a multi-step task, calls tools, checks itself, recovers from failure. Not a chatbot.
Triage scale: A structured, pre-existing system (such as the South African Triage Scale or the Manchester Triage System) for turning vitals and symptoms into an urgency level. Build on one; don't invent your own from scratch.
Vitals: Blood pressure, pulse, temperature, respiratory rate, oxygen saturation — the baseline measurements taken at triage.
Deterioration: A patient's condition worsening after their initial assessment, often while they're still waiting to be seen.
Open-weights: A model you can download and run yourself. Qwen, Llama, Gemma, Mistral, Aya.
Human in the loop: The agent pauses, and a named person approves before anything consequential.

Requirements

Projects missing any item in either list are ineligible for prizes.

What to build

Your own MCP server. Three or more different tools, and at least one that does something — scores a priority, raises a re-check flag, routes a referral. Read-only is not an agent's hands.
One MCP server you did not write. Community, official, or vendor. One line on why it beats writing it yourself.
Open-source orchestration. LangGraph, CrewAI, Pydantic AI, smolagents, Letta, Goose, n8n or equivalent. No closed no-code builders.
One full task on an open-weights model. Patient health data cannot leave the country, full stop. Show a frontier model alongside if you like, on synthetic data only.
A logged tool call for every action, and a gate on every irreversible one — before a priority level, a referral, or a deterioration flag reaches a patient's queue position. Inputs, outputs, timestamps, and a named clinician approving.
New work, built during the submission period. Existing open-source libraries are fine; an existing project resubmitted is not.

What to submit

A public code repository with an OSI-approved licence and a README that gets a stranger running in one command.
A demo video under 3 minutes, uploaded to YouTube or Vimeo and publicly visible. An unedited agent run with tool calls on screen, not slides.
A text description of what it does, which sub-theme it fits, and which facility and workflow it serves. Around 300 words.
ARCHITECTURE.md — one page. Agent shape, MCP servers built versus borrowed, and why.
EVALS.md — eight or more test tasks with pass and fail results, plus one failure you did not fix and what you would try next.

Three tools that work beat nine that half-work. In five days, scope is the skill we are watching for.

Prizes And Awards

Mentorship and build resources
Top 40 — selected from all submissions
Every shortlisted team gets mentor support and the other build resources described in the challenge brief.
All expenses paid
Top 10 — selected from the top 40
The top 10 teams are fully funded to attend the MCP Conference in Nairobi this November.
Amount TBC
1^st
Amount TBC
2^nd
Amount TBC
3^rd
Amount TBC
Best Women-led solution

How selection works: the top 40 submissions are shortlisted from all entries and receive mentorship and build resources; from those, the top 10 are selected and all expenses are paid for them to attend the MCP Conference in Nairobi this November. One prize per team. Judges may decline to award a prize if no submission meets the bar.

Judging criteria

Criterion	What we're checking	Points
Agentic depth and MCP craft	Does the agent genuinely plan, call tools, and recover — or is it one model call in a loop? Are the tool boundaries ones a stranger could reuse?	30
Open-source rigour	Can we clone it and run it? Is the README honest about what is unfinished?	20
Fit to the triage desk	Does a named nurse or clinician do a real thing faster, at the desk, under real queue pressure? Did you talk to one?	20
Evaluation and reliability	Is the test set honest rather than curated to pass? Is run-to-run variation measured or hidden?	15
Defensibility	Could this priority level survive a clinical audit question? Sourced findings, readable log, known cost per run.	10
Demo	Three clear minutes showing the agent working, including where it fails.	5
One hundred points in total.		100

Judging happens in two stages: a written review against the criteria above, shortlisting the top scores; then a 20-minute live technical review — ten minutes walking us through your code, then a new requirement added live while we watch.

Judges: panel to be announced.

Resources

There's no starter pack for this track. By this stage we're assuming you've already got MCP servers you trust and know how to stand up quickly — what data you build on is entirely your call: a published triage scale as your scoring basis, a synthetic patient stream you construct yourself, or case data you have explicit consent to use from a facility you work with. Patient health data is sensitive regardless of source — treat synthetic data as the default, not the fallback.

Questions any time in the community channel; every answer is posted publicly so no team gets a private advantage. Shortlisted teams move into a mentor-supported phase with access to better compute and, where relevant, clinical partnerships to take the prototype toward something a real facility could pilot.

Rules

Eligibility: Open to developers aged 18 and over, resident anywhere in Africa. Solo entries or teams of up to three. Organisers, judges, and their immediate families may not enter.
Submission period: Opens TBC, closes TBC. Repositories are cloned at the deadline timestamp; commits after it are visible and will disqualify a submission.
Multiple submissions: One project per team. You may not enter the same project under more than one theme.
Ownership: You keep everything you build. You grant us permission to reference and demonstrate your project in community materials. Your repository must carry an OSI-approved licence to be eligible.
Data conduct: Whatever you build on is your choice — a public triage scale, your own synthetic patient data, or real case data you have explicit consent to use. Never real patient data without consent, even de-identified.
AI assistance: Expected and not penalised — use whatever coding tools you normally use. The live review is where we separate directing a coding agent from accepting output you cannot explain.
Language: Submissions in English, wherever you are building from. Your agent may handle documents in any language.

Instant disqualification: A chatbot wrapper with no autonomous tool use · an MCP server that proxies a single endpoint · an agent that diagnoses, prescribes, or moves a patient in the queue without nurse or clinician approval · a priority level with no source · real patient health data used without consent, in any form · a repository that does not run · results presented as real runs that never happened · someone else's project, undisclosed.

Schedule

Suggested build milestones — same shape for every track.

Day 1
Kickoff, team finalisation, and deep-dive into sector-specific problem statements and available open datasets.
Days 2–3
Ideation, architecture design, and initial prototyping. Milestone: "Paper Prototype" review with domain mentors.
Days 4–5
Core development, API/MCP integrations, and UX refinement. Milestone: midpoint "Stress Test" with end-user representatives.
Day 6
Final polish, documentation drafting, and open-source repository packaging.
Day 7
Final submission via the challenge portal; selection begins.

Ready to enter healthcare?

Solo entries or teams of up to 3. You can add your repository link later.

Apply for Healthcare

The Triage Desk