When the Clinic Comes to Oja Oba: The Sire Mobile Unit Operating Model

It is 06:45 on a Tuesday morning at Oja Oba market in Akure. Mama Tola is setting up her tomato stall. A green-and-white Toyota Hiace pulls into the slot beside the market entrance and parks. The wrap on the side reads Sire — Your Doctor Where You Are. Mama Tola sees it from her stall before she has finished laying out the first basket. She does not need to walk five minutes across the square to a small lit room any more. The room came to her.

Part one of this series put a fixed kiosk at the market square — a single venue, one nurse, one rotating doctor pool, six hours a day, six days a week. The architecture worked. Then the operating reality moved. A kiosk reaches the patients within a five-minute walk of one location, and the next ten kiosks reach the patients within a five-minute walk of ten more, and Mama Tola at Oja Oba on Tuesday is the same person who is at Oja Tuntun on Friday — at which point the kiosk is no longer the right shape of clinic. The shape that follows the patient instead of waiting for her is the Sire Mobile Unit.

This piece walks through that operating model. The vehicle, the Hub at the centre of the cluster, the supervising MD on the queue, the software that runs the network, the cold-chain primitive that makes lab work possible from a moving vehicle, the weekly informal-sector billing rhythm, and the expansion sequence that prefers density to geography. The clinical architecture from Part 1 — Whisper, the local-LLM tier, the pharmacist gate, the in-country patient record — carries through unchanged. The operating model is what wraps around it.

Six Weeks After the Kiosk Worked

The kiosk in Part 1 served the patients within a five-minute walk of one corner of Oja Oba. Mama Tola was one of them. Her cousin who sells fabric at Oja Tuntun, eight kilometres across town, was not. Her neighbour whose daughter attends a secondary school in Akure South, fifteen kilometres in the other direction, was not. The kiosk reached a few hundred patients in a catchment area defined by walking distance, and the unit economics worked for that catchment.

Then the pilot proved out, and the question became how to reach the patients the kiosk did not. Three answers were on the table. The first was to build more kiosks — a fixed venue in every market, every neighbourhood, every school. The capital cost per location made this prohibitive. The second was to add a delivery layer — kiosks plus a mobile arm. This was operationally complex without a clear unit-economics improvement. The third was to invert the model. The same clinic that Mama Tola walked five minutes to find, packed into a vehicle, rotating to where she already was. This is what Sire became.

What Goes in the Hiace

The Mobile Unit is a Toyota Hiace, bought used, kitted out internally and wrapped externally. Inside there is a fold-down examination bed, a reception and triage station, a mini-pharmacy locker, a sample-collection area, and two tablets running Sire OS. The clinical kit is the same as the kiosk in Part 1 — digital BP cuff, glucometer with strips, pulse oximeter, digital thermometer, basic ECG, otoscope, weighing scale, height rod. The rapid-test stock covers malaria, HIV, pregnancy, and urinalysis. HbA1c and lipids run on point-of-care devices for results before the patient leaves the vehicle.

The pharmacy locker is restocked daily from the Hub. The dispensing log carries the same audit-chain discipline as the kiosk's — doctor's signature, pharmacist's signature, AI tier's involvement as a separate field. Connectivity is 4G primary with a Starlink Mini fallback, because the consultation cannot fail when the network does. Power runs from the vehicle battery through an inverter, with a small portable solar panel for extended-hours stops at the same rotation site.

The branding matters more than it sounds. The full vehicle wrap, recognisable in the same green-and-white livery, is recurring advertising every time the unit parks. Mama Tola does not need to remember a clinic address or a telephone number. She remembers the day of the week the green-and-white vehicle is at her market. The brand is the schedule.

The Hub Is Not the Clinic

The Sire Hub is the network's operating base, not its customer-facing clinic. One Hub per city anchors a cluster of ten to twenty Mobile Units. It is a thirty-to-fifty square metre ground-floor space, near a major arterial road for mobile-unit access, and it houses the supervising MD's telemedicine queue station, the pharmacy bulk stockroom, the cold-chain depot for samples and vaccines, the mobile-unit dispatch and maintenance bay, and a small backup consultation room for patients who prefer in-person and have walked in.

The reason the Hub matters architecturally is that it changes the meaning of the supervising MD's day. In a single-clinic model, one doctor sees one queue of patients in one place. In the cluster model, one supervising MD sees a queue of telemedicine consultations escalated by nurses across the entire fleet — Unit 1 at Oja Oba in the morning, Unit 3 at a secondary school in the afternoon, Unit 7 at a Friday mosque post-Jumat screening. The same MD covers all of them, in a queue, from the Hub. The Mobile Units' rotation schedules align so that no two units escalate at the same moment and saturate the queue.

The HEFAMAA mobile-facility licensing for the Mobile Units lists the Hub as the registered facility, with each Mobile Unit as a declared route extension. The legal opinion that grounds this — under the Nursing and Midwifery Council scope of practice for nurse-led primary care, paired with the MDCN 2024 telemedicine guidelines for the remote MD oversight — is the regulatory work that has to land in writing before the second Mobile Unit goes on the road.

One Doctor, Ten Vehicles

The economics shift is the unlock. The supervising-MD cost in a single-clinic model is paid by one clinic. The supervising-MD cost in the cluster model is paid by the cluster — one Hub, ten Mobile Units, one MD covering all of them. The same doctor whose cost would have bottlenecked a single fixed clinic now covers ten times the patient reach without ten times the cost.

This is not a software trick. It is the same architectural primitive as cmdev's tiered AI routing patterns — one expensive resource at the centre, many cheap resources at the edge, an intelligent router between them. In the Sire architecture the expensive resource is the supervising MD, the cheap resources are the nurse-led Mobile Units, and the router is Sire OS plus the rotation schedule. The MD cost per Mobile Unit drops as the cluster grows, and the floor — the cost at which the model breaks — is well above where a single-clinic operator would walk away from the deal.

This shape is what makes mass-market primary care economically viable for the patient paying three thousand naira a visit rather than fifteen. The kiosk in Part 1 worked because the AI tier compressed the per-consultation cost. The mobile network works because the cluster compresses the per-unit supervisory cost. Both compressions stack.

Sire OS — Three Apps, One Spine

Sire OS is the operating system that runs the network. It is the platform commitment that differentiates Sire from a logistics company doing telemedicine and from a telemedicine company doing logistics. Neither shape would work; the platform is the shape that does.

The operator app runs on the nurse's tablet in the Mobile Unit. It is the most important UI in the company — patient intake and vitals, FHIR Patient and Observation resources, symptom triage with red-flag escalation, the telemedicine call to the supervising MD, the drug-dispensary inventory, lab ordering and sample tracking with cold-chain logging, the patient queue, multi-stop session management, membership and billing. It runs offline-first because connectivity at a rural school stop or a market with patchy 4G is a daily reality, not an edge case.

The patient app runs on WhatsApp and on a browser PWA. Tier-two Nigerian patients will not install a native app, and the operating model accepts this without arguing. WhatsApp Business API plus a PWA covers ninety-five per cent of the use case — finding the nearest Mobile Unit, seeing the weekly route, booking a slot before arrival, tracking the unit in real time like an Uber, managing membership, requesting a prescription refill, receiving lab results, getting chronic-care nudges, paying weekly. The WhatsApp channel is the fallback for everything above when the PWA cannot reach the patient.

The network back office runs on the web for HQ — Mobile Unit onboarding, operator training tracking, the supervising MD queue, clinical quality monitoring, supply-chain management, HMO billing under NHIA accreditation, corporate retainership management, network analytics, financials. The data spine is Postgres with the FHIR R4 schema underneath everything, which keeps the door open for the V2 interoperability backbone and for HMO integrations that the regulator will eventually require.

The logistics layer is the part of Sire OS that does not exist anywhere else in the Nigerian healthcare market. Route planning, pre-arrival booking, real-time unit tracking, multi-stop session management, hub-and-spoke inventory sync, cold-chain telemetry, geofenced encounter logs, demand heatmap. Helium Health cannot build this without becoming a network operator. Uber-class logistics platforms cannot build the clinical core. The combination is the moat.

The Fridge That Knows It Is Lying

The cold chain is the architectural primitive that turns the Mobile Unit from a triage station into a clinical facility. A medical-grade fridge runs in the vehicle at two to eight degrees Celsius with continuous temperature logging, twelve-volt and twenty-four-volt operation from the vehicle battery while moving, AC operation at the Hub overnight. The samples drawn during the day — blood for HbA1c panels, blood for lipid panels, plasma for chronic-care follow-up — sit in the fridge, QR-coded, logged into Sire OS, until the end of the shift when the Mobile Unit returns to the Hub and the samples transfer to the Hub depot fridge for the next morning's lab courier.

The fridge's temperature stream is part of Sire OS. Every two minutes the fridge reports its core temperature, the door state, and the battery state. Any excursion above eight degrees or below two flags the affected samples for re-collection and escalates to the supervising MD. The fridge knows when it is lying about its temperature — the door-state telemetry catches the case where a sensor reads safe but the door was opened for too long during a hot drive. This is not a feature on top of the clinical workflow; this is the safety architecture that makes mobile sample collection auditable and defensible under the Medical Laboratory Science Council's sample-collection compliance regime.

The same telemetry stream covers cold-chain medications and vaccines. The Mobile Unit can carry a small vaccine load and run an immunisation arm at corporate stops and at school health days, which the kiosk in Part 1 could not, because the kiosk was not designed for vaccine storage in the original architecture. The mobile network inherits this capability by accident of the fridge being a hard requirement for sample work.

The Ajo Rhythm

The membership product is five thousand naira a month covering Mama Tola, her husband, and two children. She does not pay it monthly. She pays it weekly at twelve hundred and fifty naira — the same cadence as her cooperative-society ajo, the rotating-savings scheme that her informal-sector financial life already runs on. The payment is taken through Paystack with a Flutterwave fallback, and it is structured as a recurring weekly mandate rather than a monthly direct debit because the monthly cycle is hostile to the cash-flow rhythm of a tomato seller.

This is the operating-model decision that surprised the engineering side hardest. Standard SaaS billing assumes monthly. Standard healthcare insurance assumes annual or quarterly. Both fail at the informal sector. The patient persona the entire business is built around — a market woman earning four to eight thousand naira a day, net, with a weekly ajo rhythm — needs a billing surface that respects that rhythm. The product team did not arrive at this from a whiteboard. The community-society interviews during the patient willingness-to-pay survey is what surfaced it, and Sire OS now treats weekly recurrence as the default for member billing and monthly as the exception.

Corporate retainerships for SMEs, civil-service offices, and factories run on the monthly rhythm, because corporate cash flow tolerates it. The school partnerships run on annual contracts. The chronic-disease cohorts on hypertension and type-two-diabetes subscriptions run on a monthly cycle for the medication side. The product surface respects each rhythm, and the back-office reconciliation handles the heterogeneity.

Density Before Geography

The instinct is to spread the brand across cities quickly. The expansion sequence in the strategic plan does the opposite. Density wins.

A Mobile Unit in Akure with four other Mobile Units sharing the same Hub gets the shared supervising MD queue, the shared drug-supply chain from the Hub stockroom, the shared lab-pickup route from one courier, the cross-rotation cover when Unit B is in for service and Unit A picks up its Tuesday market, and the compounding brand recognition across neighbourhoods that all see the same livery on the same rotation. The same Mobile Unit operating alone in a city without a Hub is logistically isolated and economically marginal.

The sequence is to go from one Unit to ten in Akure first — months one through eighteen, the cluster proved — then start the Osogbo cluster, then Ile-Ife and Ilesha, then Ibadan and Abeokuta, then out-of-corridor expansion. The shape of the network is not a thin spread across a region. It is a sequence of dense city clusters, each of which has to clear the unit-economics bar inside the cluster before the next cluster is opened. This is the same discipline that built Moniepoint and Opay agent networks at national scale — proven density before geographic spread.

What This Operating Model Does Not Do

The Sire Mobile Unit operating model is built for primary care and chronic-care follow-up. It is not built for surgery, for emergency response, for intensive care, for cancer treatment, for tertiary specialist consultations. The operating model knows what it does and stays inside the bounds. Referrals from the Mobile Unit go to the appropriate hospital — Federal Medical Centre Owo, OAU Teaching Hospital, the relevant State Specialist Hospital, the appropriate private partner — and the patient's Sire record travels with them in FHIR R4-portable form when the receiving hospital can accept it.

The model also does not bypass any of the architectural commitments from Part 1. The AI tier still does not prescribe. The doctor still does. The pharmacist still has veto power on every dispensed medication. The patient record still stays in Nigeria. The escalation to the frontier model still requires explicit patient consent. The compression that the cluster shape adds is on the operating side; the clinical side is unchanged.

If the kiosk had to become a vehicle to reach Mama Tola, what does that say about every fixed clinic still waiting for her to find it?

Companion content

• The Doctor at the Market Square: A Hybrid Telehealth Micro-Clinic Architecture — Part 1, the kiosk concept this operating model evolved from
• IAM for AI Agents: Delegation and the Principal-of-Record Chain — the audit-chain pattern the Sire OS clinical actors reuse
• The AI Act Enforcement Cliff: The August 2026 Deadline — regulatory pattern adjacent to NDPA Section 41–43
• Hardened Containers for AI Workloads: The Tier Decision Most Teams Get Wrong — the AI workload isolation tier that the Sire OS clinical-note service deploys onto

How to engage

Sirehs is a CreativeMinds-incubated project moving toward a pilot of the operating model above in Akure, Ondo State. The Phase 0 work — NMDCN licensing scope, NAFDAC pharmacy partnership, NDPA Section 65 designation, clinical advisory board, medical co-founder, pilot site agreement, unit economics validation, local-LLM clinical benchmark, model availability SLA, and African-AI lab partnership scope — is ongoing and collaborative. If you operate in tier-two Nigerian primary care, mobile-asset healthcare logistics, HMO accreditation, NDPA advisory, or African-LLM infrastructure, the architecture above is the conversation starter. Talk to us at creativeminds.dev/contact. The Mobile Units that go on the road in the pilot will be built with the partners we meet now.