See stock risks early. Keep patients connected to care.
I build practical AI and automation systems for pharma companies, distributors, health programmes and pharmacies in East Africa. Stock visibility, field-team tracking, access reporting and patient support between visits. Assistants built around your approved guidelines. Built by someone who spent 13 years doing the work.
One view of field, stock and access numbers.
An illustrative view of field activity, stock and programme reporting in one place. Each pilot starts with the data your team already collects. Messaging, voice-note capture and scheduled reports are scoped and tested before rollout.
Sample data for illustrationHealth programmes rarely lack data.
They lack a clear view of it, in time.
The numbers exist. They're in field reports, distributor spreadsheets, WhatsApp groups and someone's inbox. By the time they're compiled for the review, the stock-out has already reached a patient.
Distributor data arrives late, in six formats
- Sell-out figures by email, PDF and phone call
- No view of stock by outlet or county
- Reorders triggered after the shelf is empty
Field teams report on WhatsApp, and it stops there
- Visits, orders and objections buried in chats
- Managers retype reports every Friday
- No record of which outlets were missed
Stock-outs are found by patients first
- Low stock only noticed at the counter
- Expiry risk spotted too late to redistribute
- Access programmes lose trust they worked hard for
Reporting eats the week before every review
- Market-access and donor reports built by hand
- Numbers that don't match between regions
- Senior people doing copy and paste
Guidelines exist, but answers are slow
- Staff search 80-page PDFs for one dose rule
- General chatbots guess, and can't cite a source
- New staff take months to get confident
The last mile still runs on notebooks
- Small chemists log sales and M-Pesa codes by hand
- No monthly software bill they'll accept
- Their data never reaches anyone who could help
"The problem is rarely missing data. It's that nobody can see it in time to act."
Systems built around health supply and patient access.
Services available for scoped pilots. Start with one workflow and one region. Integrations and automation are configured and tested for each project.
Supply and distributor visibility
Bring distributor sell-out, outlet stock and orders into one view, with alerts when an outlet drops below its threshold or stock nears expiry.
- Distributor and channel reporting
- Stock and days-of-cover by outlet
- Low-stock and expiry alerts
- Redistribution shortlists
Field-force and programme tracking
A proposed workflow captures visits, orders and follow-ups from staff updates. WhatsApp and English or Kiswahili voice-note processing can be included in the pilot, with accuracy checked against real examples.
- Visit and activity capture
- Coverage against target outlets
- Follow-up queues for each rep
- Weekly summaries for managers
Market-access and donor reporting
Build reporting around patients reached, sites reporting, regional uptake and pricing commitments. Scheduled drafts use agreed data sources and reporting formats, with your team reviewing the figures before submission.
- Access programme dashboards
- Donor and grant reporting packs
- Consistent numbers across regions
- Review-ready summaries
Grounded health assistants
Assistants designed to retrieve answers from approved guidelines, protocols and product information, with source citations. The diabetes guidelines prototype is in development. Each pilot includes accuracy checks and tests for unsupported questions.
- National and internal guidelines
- Citations on every answer
- Accuracy testing before launch
- Staff onboarding and refreshers
Keep patients connected between visits.
A proposed pilot offer for pharmacies, clinics and health programmes. Build consent-based follow-up around medicine access, refills, patient education and the barriers that interrupt care.
Refills and review reminders
Help care teams follow up when a refill or review is due. Agree message timing, preferred channels and opt-out steps before starting.
- Consent and contact preferences
- Refill and appointment reminders
- Missed-visit follow-up queues
Access and affordability follow-up
Record reported barriers such as cost, stock availability or travel. Assign follow-up to the right team and track whether the issue is resolved.
- Structured access-barrier records
- Referral to available support programmes
- Named owners and follow-up dates
Education and care-team handoffs
Share education approved by the care team. Route patient questions and concerns to a named healthcare professional for review.
- Approved patient education
- Care-team escalation and referral tracking
- Engagement and follow-up summaries
Start with one facility and one patient group. Agree consent, data access, message approval and clinical handoffs before enrolment. This service supports care-team workflows; clinical decisions remain with qualified professionals.
Discuss a patient support pilot → More on patient support →Three entry points, one method.
The same thinking works for a regional supply team and a single chemist. What changes is the scale and the price.
Pharma, distributors and health programmes
Supply visibility, field-force tracking, access reporting and patient support workflows for commercial, access and programme teams.
- Scoped to one workflow and one region first
- Run remotely, with on-the-ground support
- Your data stays in your accounts
Pharmacy groups and clinics
Help choosing and setting up systems for stock records, patient follow-up and staff workflows. Consent, access controls and team training are included in the project scope.
- System selection and setup
- Staff training and adoption
- Stock records and patient follow-up
WhatsApp record keeping
A pilot for recording sales, payments and stock with less manual work. Message capture, voice notes and evening WhatsApp summaries are agreed and tested during setup.
- Designed around familiar tools
- Voice-note support tested during the pilot
- One-time setup fee; optional ongoing support
Any messaging, AI or hosting charges are quoted separately before you commit.
Same outlets. Same stock.
Only one team sees it coming.
What a supply problem looks like before the system, and after it.
"Kisumu outlet out of insulin pens again 😩"
"Noted, will check with the distributor"
Nakuru order sent??
Ask Brian Monday
Q3 numbers: 4 regions missing
Access report due Friday. Need the uptake figures from Western. Or did they send them last week?
Result: a patient finds the stock-out first. The report is built from memory.
Afya Plus Chemist, Kisumu
Sample data. This is the shape of what the system produces.
Inventory, sales log, P&L dashboard and customer intake. Where the method was first proved.
Answers from Kenya's national diabetes guidelines only, with a citation on every answer.
Sales and stock recording pilot, with optional WhatsApp summaries. Setup and any ongoing service costs are agreed upfront.
Visits, orders and follow-ups from WhatsApp, with coverage against target outlets.
Sell-out and outlet stock in one view, with low-stock and expiry alerts.
Patients reached, sites reporting and uptake by region, drafted for review.
Tested on a real business first. Same problem, smaller scale.
Before selling the method to health teams, I built it for a Nairobi retail and repair business with the same symptoms: stock in notebooks, payments across M-Pesa and banks, follow-up on memory.
Bikes & Accessories KE
Customer enquiries, inventory, sales, service records and financial tracking across several payment channels, running in one system.
Business context
A growing bike retail and repair business with conversations across WhatsApp, Instagram and walk-ins, and no system behind them.
What was broken
Inventory in notebooks. Sales across M-Pesa and several banks with no reconciliation. Follow-up running on memory.
What was built
Inventory database, sales log, purchase records, expense tracking, a P&L dashboard and a customer intake workflow.
What is live now
The intake system and financial tracker are in daily use, capturing real submissions. Full case study in progress.
Not from the tech.
Thirteen years in pharma and global health across East and Southern Africa and the Horn. Medical and key account roles at Pfizer, AstraZeneca, Eli Lilly and Novo Nordisk. Compliance Lead and then Access to Care Manager for Middle Africa at Novo Nordisk. Then Programme Manager for type 1 diabetes markets at CHAI.
I've worked with distributors in five countries, prepared tenders, chased sell-out data and built access reports the night before reviews. I've watched stock sit close to expiry in one place while patients went without in another.
Computers were my first love. In 2024 I went back to them through data science, and then AI. Mindset & Machines is where the two meet: someone who knows the work, building the systems for it.
Daisy Mutegi, founder. BSc Biochemistry (First Class), PMP.
Real demand, loyal customers, and no way to see what was happening. Not for lack of effort. For lack of visibility.
Medical and key account sales
Pfizer, AstraZeneca, Eli Lilly and Novo Nordisk. Doctors, hospitals, pharmacies and distributors across the region.
Compliance Lead, Middle Africa
Novo Nordisk. Keeping commercial and access work inside the rules across many markets.
Access to Care Manager, Middle Africa
Novo Nordisk. An access programme that reached about 21,000 patients and supplied close to 1 million insulin vials.
PMP and data science
Project management certification, then a data science certificate. The start of the AI work.
Programme Manager, T1D Markets
CHAI, ALIGN-T1D interim secretariat. Market shaping for type 1 diabetes in low- and middle-income countries.
Mindset & Machines
AI systems for the health supply, access and programme work I know from the inside.
Numbers from the work itself.
Five rules every build follows.
Visibility comes before strategy.
Before pricing, targeting or scaling, you need to see what's actually happening in the field. A clear view is a prerequisite, not a feature.
Simple systems beat complicated software.
A workflow your team uses every day is worth more than a platform they log into once. The measure of a system is adoption.
Health AI answers from trusted sources, or not at all.
Every assistant cites where its answer came from and is tested before anyone relies on it. If the source doesn't say, it says so.
Your data stays yours.
Systems run in your own accounts where possible, follow Kenya's Data Protection Act, and collect only what the job needs.
Nobody should need to be "techy" to use it.
If it needs a manual, it isn't finished. WhatsApp, voice notes and the reports people already read come first.
No giant projects. No unnecessary complexity.
One workflow, one region, one clear measure of success. Then build from there.
Map the workflow
A 30-minute call on how the work runs today. Who reports what, where it goes, and where it gets stuck.
Find the bottleneck
Usually one or two points where visibility breaks, not the whole operation. That's where we start.
Run a paid pilot
2 to 4 weeks on one region, product line or team, with what counts as success agreed in writing first.
Scale what works
Roll out to more regions or teams only once the pilot has shown results. Every extra layer earns its place.
Notes from inside the work.
Short pieces on health supply, access and the systems behind them. No generic AI content.
Why stock-outs are found by patients first
The data that would have warned you usually existed. It was sitting in a WhatsApp group or a distributor's inbox. A look at where early warnings get lost.
What a bike shop taught me about pharmacy records
Building a live stock and sales tracker for a Nairobi business. What worked, what broke, and what carries over to chemists.
What "tested" should mean for a health assistant
Why a chatbot that sounds right isn't good enough for clinical guidelines, and how to check accuracy before anyone relies on it.
Get each note by email when it's published.
Thanks. You'll get the first note when it's out.
Tell me what you're trying to track.
Tell me about your workflow, without including patient names or medical details. A few lines is enough. I'll reply within two working days, usually sooner, with either a time for a call or an honest "this isn't the right fit".
Book a 30-minute call.
Not a demo and not a pitch. We map how the work runs now and find the one place where better visibility would save the most time or stock.