AI TESTING, EVALUATION & ENABLEMENT · NAIROBI, KENYA
Find Out Where Your AI Breaks
Before Your Customers Do
Independent testing, evaluation and red-teaming for AI products, plus hands-on AI implementation for teams that want it working in production, not just in a demo.
Fixed-scope pilots · Written, reproducible reports · NDA before any data is shared · Based in Nairobi (UTC+3)
WHAT WE DO
Five Ways We Help
Testing comes first. Building and data work follow the same standard.
AI Testing & Evaluation
We test chatbots, RAG systems and AI agents against the standards you care about: factual accuracy, grounding in your documents, tone, refusals and correct tool use. You get a reusable test set and a scored report, so every later model or prompt change can be checked the same way.
AI Red-Teaming
Structured adversarial testing: multi-turn attacks, prompt injection, data leakage and unsafe outputs. Each finding comes with steps to reproduce it, a severity rating and a suggested fix. We use OWASP’s GenAI security guidance as a reference point.
Multilingual & East Africa Testing
Does your assistant cope with Swahili, Sheng and English mixed in one message? Does it understand local names, places and payment habits? We test it with native speakers and report what fails.
Applied AI Implementation
Customer support assistants, WhatsApp automation, document workflows and internal knowledge tools. We build them as fixed-scope sprints, with testing included and a proper handover to your team.
Specialist Data Services
Annotation, review and evaluation data for projects that need language or domain expertise. We take on small, carefully run projects. We do not offer high-volume labelling or content moderation.
Who This Is For
Teams that ship AI to real users and need to know how it behaves
AI startups
Shipping LLM features and wanting an independent view before launch or after a model change.
Companies adding AI
Putting assistants into support, sales or internal workflows and needing them to behave.
Teams preparing for regulation
Needing documented test evidence ahead of EU AI Act obligations.
Firms entering East Africa
Wanting to know how their product handles Swahili, local context and mobile-first users.
How an Engagement Works
Fixed scope and fixed price, agreed before we start
STEP 1
Scope
A call to understand what you built, who uses it and what “good” means for you.
STEP 2
Pilot
A small, fixed-price test of one system, so you see our work before committing to more.
STEP 3
Report
Findings with steps to reproduce, severity ratings and suggested fixes, in plain language your team can act on.
STEP 4
Retest
We rerun the same tests after your fixes, or on a monthly basis as your product changes.
Testing and the EU AI Act
The EU AI Act’s transparency duties for AI that talks to people or generates content have applied since 2 August 2026. The rules for high-risk systems were pushed back to December 2027, but conformity work takes time, and documented test results are part of it.
We run the tests and document what we find. We are not a law firm, and we do not certify compliance.
How We Handle Your Data and Our People
- An NDA and data-processing terms are signed before any data is shared.
- Work happens in accounts and tools you approve, with access limited to named people.
- We delete your data at the end of the engagement if you ask us to.
- We follow Kenya’s Data Protection Act and will work to your GDPR requirements.
- We do not take on content moderation or graphic material, and everyone on a project is paid fairly.
- If your procurement requires a certification such as ISO 27001 or SOC 2, tell us at the start and we will say plainly whether we hold it.
Typical Engagements and Starting Prices
Every project is quoted after a scoping call. These are starting points, in US dollars.
Pilot
From USD 1,500
Testing or Red-Team Pilot
One system, fixed scope, and a written report with steps to reproduce each finding.
Assessment
From USD 3,500
Full Evaluation or Red-Team Assessment
Wider coverage of your system and a reusable test set your team can keep running.
Retainer
From USD 1,200 / month
Monthly Retest
Regular reruns as your models, prompts and data change.
Build
From USD 2,500
AI Implementation Sprint
A fixed-scope build with testing and handover to your team included.
Want your team trained first?
Our Digital Transformation Training and Enablement programmes cover AI tools, data and change leadership, and can lead straight into implementation work.
Frequently Asked Questions
Do you certify that our AI is compliant?
No. We test your system and document the results. Whether it meets a legal requirement is a question for your lawyers or a notified body, and our reports can feed into that work.
Can we work under our own NDA?
Yes. Send it before the scoping call and we will review it first.
Which languages can you test in?
English and Swahili, including messages that mix the two. If you need another language, ask and we will tell you honestly whether we can cover it.
Do you do high-volume data labelling?
No. We take small specialist projects where language or domain knowledge matters. For large-scale labelling you are better served by a larger vendor.
How is pricing done?
Fixed-scope pilots and project fees, or a monthly retainer for ongoing testing. Starting prices are listed above. We confirm the exact fee after a scoping call, in USD, GBP or EUR for international clients.
Start With a Scoped Pilot
Tell us what you are building and what worries you about it. We will come back with a fixed scope and price.
Email: sales@metadata.co.ke · WhatsApp: +254 792 672500