Scale of the problem
1 in 10
medical products in many low- and middle-income countries is estimated to be substandard or falsified.
Estimated. Source: WHO — Substandard & falsified medical products
AI medicine verification
Know a medicine is real before it's sold.
TrueDose uses computer vision and AI to verify medicines at the point of sale, so pharmacies and distributors catch counterfeit and substandard drugs before they reach a patient.
Built on AWS Accelerated with NVIDIA Aligned to national track-and-trace standards
The problem
Fake medicine is a money problem and a life problem.
Counterfeit and substandard medicines circulate widely across many emerging markets. They fail to treat illness, drive repeat visits, and erode trust in legitimate pharmacies.
Licensed pharmacies buy stock from layered, sometimes gray-market distributors and have no fast way to tell a genuine pack from a convincing fake at the counter.
Regulators cannot physically inspect at the scale the market moves. The cost lands on three groups: pharmacies carrying liability, regulators without reach, and patients who pay for medicine that does not work.
Who carries the liability
Pharmacies
& distributors
& distributors
Licensed pharmacies and the distributors who supply them carry the legal and reputational risk when a fake pack reaches a patient — with no fast counter-side check today.
The inspection gap
Manual
can't scale
can't scale
Regulators rely on lab testing and physical inspection that cannot keep pace with the volume and velocity of the market.
How it works
Point. Scan. Verify in seconds.
01
Capture
A pharmacist photographs the pack, bottle, or blister with any phone — no special hardware.
02
Analyze
TrueDose checks packaging artwork, print micro-features, hologram cues, and batch/expiry formatting against a manufacturer reference library.
03
Verify
In seconds, TrueDose returns Genuine, Suspect, or Recalled — with the reason and an audit record the pharmacy can keep.
Capture
Analyze
Verify
AI technology
AI is the engine, not a feature.
TrueDose is built around models that learn what genuine packaging looks like and flag what does not belong.
What the AI sees
Photos of packaging, blisters, labels, holograms, batch and lot codes — captured in real pharmacy conditions on ordinary phone cameras.
What it checks
Artwork and print consistency, micro-feature integrity, batch/expiry formatting, and lot numbers cross-referenced against recall and reported-counterfeit data.
How it decides
Image-classification and feature-matching models score authenticity; an anomaly-detection layer flags print drift that signals a new counterfeit run the system has not seen before.
Why it improves
Every verified and reported scan strengthens the reference library and the anomaly models — the system gets sharper as it sees more of the market.
Product
What pharmacies and distributors actually see.
The scan result, in the pharmacist's hand
A single tap returns a clear verdict with the reasons behind it — and a record saved to the audit log. Two real-world outcomes shown: a verified pack and a flagged one.
- Plain Genuine / Suspect / Recalled verdict
- Reasons in plain language, not a black box
- Every scan saved to the audit log
GENUINE
Artwork match 98%
Batch LOT-7741-A verified
No recall match
14:22 · 06 JunSaved to audit log
SUSPECT
Print micro-features inconsistent
Batch code format mismatch
Quarantine + report suggested
14:31 · 06 JunSaved to audit log
One screen for the whole counter
Scans today, suspect flags that need attention, and a verified-batch log — so a pharmacy manager can see exposure at a glance.
- Live metrics across the team
- Suspect flags surfaced first
Scans today
182
Suspect flags
4
Verified batches
37
| Time | Drug | Batch | Verdict |
|---|---|---|---|
| 14:31 | Co-amoxiclav 625mg | LOT-9920-C | SUSPECT |
| 14:22 | Amoxicillin 500mg | LOT-7741-A | GENUINE |
| 14:08 | Artemether/Lumefantrine | LOT-4410-B | GENUINE |
| 13:55 | Metformin 850mg | LOT-2231-D | RECALLED |
| 13:41 | Paracetamol 500mg | LOT-6650-A | GENUINE |
Suspect-flag trend · last 10 days
Every batch verified, exportable
Distributors get a filterable trail of every batch they moved and its verification status — ready to hand to a partner or a regulator.
- Filter by status, drug, or date
- One-click CSV export
AllGenuineSuspectRecalled
| Batch | Drug | Verified | Qty | Status |
|---|---|---|---|---|
| LOT-7741-A | Amoxicillin 500mg | 06 Jun | 2,400 | GENUINE |
| LOT-9920-C | Co-amoxiclav 625mg | 06 Jun | 1,150 | SUSPECT |
| LOT-4410-B | Artemether/Lumef. | 05 Jun | 3,200 | GENUINE |
| LOT-2231-D | Metformin 850mg | 05 Jun | 900 | RECALLED |
| LOT-6650-A | Paracetamol 500mg | 04 Jun | 5,000 | GENUINE |
Suspect clusters, by area
Regulators see where suspect packs are concentrating — turning thousands of scans into an early-warning map of where a new fake is spreading.
- Cluster counts by region
- Drill into the underlying scans
- Central district21
- Northgate12
- Riverside6
- Eastmarket3
- Hillview1
Suspect-cluster counts · sample data, last 30 days
Technology & infrastructure
Built on cloud and accelerated for real-time verification.
How we use AWS
We use AWS to host our verification APIs, store the manufacturer reference library and scan records securely, manage authentication, deploy scalable databases, serve the frontend, and monitor performance. We use Amazon S3 for image and reference storage, RDS for verification and reporting data, Lambda and ECS for API and inference orchestration, CloudFront for delivery, CloudWatch for monitoring, and we plan to use Amazon Bedrock for future model workflows.
S3RDSLambdaECSCloudFrontCloudWatchCognitoBedrock (planned)
How we use NVIDIA
We use NVIDIA technologies to train and accelerate our packaging-recognition and anomaly-detection models. We use CUDA and GPU training for model development, TensorRT and Triton Inference Server for fast, low-latency inference so a scan returns a verdict in seconds under load, and we have a path to NVIDIA Jetson edge devices so verification works in pharmacies with poor connectivity.
CUDAGPU trainingTensorRTTritonJetson (roadmap)
aws
NVIDIA
AWS and NVIDIA are trademarks of their respective owners. TrueDose is not affiliated with or endorsed by them.
| Capability | Stack | What it does for verification |
|---|---|---|
| Computer vision | CUDA · TensorRT · Triton | Recognizes packaging artwork, micro-features, and batch codes from a phone photo, fast enough for the counter. |
| Healthcare AI pipelines | GPU training · medical-imaging-grade pipelines | Trains and validates models with the rigor used for medical-imaging workloads, with reproducible evaluation. |
| Anomaly detection | Triton · Bedrock (planned) | Flags print drift that signals a new counterfeit run the system has not seen before. |
| Edge verification | NVIDIA Jetson (roadmap) | Runs verification in pharmacies with poor connectivity, offline-first. |
Benefits
What you get.
Catch fakes fast
Flag a counterfeit or recalled pack in under 5 seconds per scan (beta goal).
Cut your exposure
Reduce a pharmacy's exposure to fake stock by keeping unverified packs off the shelf (beta goal).
Keep an audit trail
Keep a verifiable audit trail for every batch — exportable for partners and regulators.
Early warning
Get early warning when a new fake appears in your area, from aggregated suspect clusters.
Works offline
Verify offline-first in low-connectivity locations (roadmap).
Impact
The difference we're built to make.
We're in MVP and pre-pilot — so these are the stakes we're built to cut and the outcomes we're building toward: measured and sourced, never overstated.
$30.5B
spent on substandard and falsified medicines across low- and middle-income countries every year — the loss TrueDose exists to cut.
< 5 sec
to return Genuine, Suspect, or Recalled on a pack — on any phone, at the point of sale.
Every scan
becomes a verifiable, exportable audit record for pharmacies, distributors, and regulators.
Any phone
no scanners and no special hardware — verification runs on the camera a pharmacy already has.
These are sourced estimates or stated targets — not results. We publish measured impact once pilots produce it.
Why now
Why now.
The gap between "verification is required" and "an affordable tool exists" is the opening TrueDose is built for.
- 01Cameras are good enough.
Smartphone cameras are now sharp enough for reliable packaging inspection.
- 02GPU inference is cheap enough.
Running verification at scale is now affordable, not a research-lab luxury.
- 03Regulators are mandating track-and-trace.
Markets across several regions are introducing serialization and track-and-trace requirements.
- 04Counterfeit volume is rising.
Fakes grow as supply chains digitize unevenly, widening the gap genuine pharmacies must close.
Market opportunity
Market.
Target users
- Licensed retail pharmacies
- Regional drug distributors
- Hospital procurement teams
- Regulators
Launch markets
Starting with high-counterfeit-burden markets:
NigeriaGhanaKenya
confirm before launch
Revenue model
- Per-scan verification credits
- Pharmacy subscription tiers
- Enterprise contracts for distributors & regulators
Growth plan
- Start with pharmacies in one city
- Expand by distributor network
- Then expand by market
Industry-size figures are shown only when verified with a cited source; we do not publish an invented TAM.
Traction
Where we are.
MVP in development
MVP in development, covering the 25 most-counterfeited medicines.
Pilot — in preparation
Preparing to pilot with selected pharmacies in Lagos, Nigeria.
Dataset
Building a verified-scan dataset with partner pharmacies.
Open now
Waitlist open for pharmacy chains and distributors.
Pilot partners in discussion — named only once confirmed. We do not list partners, testimonials, or user counts we cannot verify.
Pilot partner
in discussion
in discussion
Pilot partner
in discussion
in discussion
Pilot partner
in discussion
in discussion
Roadmap
Roadmap.
| Phase | Milestone |
|---|---|
| Phase 1 | MVP: scan + verify top counterfeited drugs |
| Phase 2 | Pharmacy pilot + verified-scan dataset |
| Phase 3 | Anomaly detection + recall cross-referencing |
| Phase 4 | Distributor audit trail + regulator reporting |
| Phase 5 | NVIDIA-accelerated cloud inference at scale |
| Phase 6 | Jetson edge devices + expansion to new markets |
Team
Who is building it.
Chinonso Emmanuel
Founder / CEO
With a background spanning pharmacy supply chain and product, he turns frontline realities into tools pharmacies and patients can trust at the point of sale.
LinkedIn
Christina Okafor
Technical Lead
Leads TrueDose's engineering and applied AI — designing the computer-vision models and ML infrastructure that verify medicines in seconds.
X (Twitter)Our team includes software developers, AI engineers, and domain experts in pharmacy and supply chain.
Trust & compliance
Trust and compliance.
- Privacy Policy — what we collect, how it's used, and your rights.
- Terms of Use — acceptable use and limitations.
- Data protection — scan images and form data handled under a documented data-protection statement.
- Security practices — encryption in transit and at rest, with role-based access controls.
- Company registration — TrueDose Technologies Ltd. Reg. No. RC 7892145
Required disclaimer
TrueDose assists verification and does not replace regulatory laboratory testing or professional medical and pharmaceutical judgment.
Request a pilot
Request a pilot.
Tell us about your pharmacy, distribution network, or agency. We'll be in touch about a pilot.
chinonso@truedose.site- No spam — we only reply about pilots.
- Your details are kept under our privacy policy.