Wakari AI Consultancy
Strategic AI consulting for an accessibility platform founder on a tight budget: feasibility, low-cost validation, and the vendor-management skills to avoid expensive mistakes.
A non-technical founder, a fixed budget, and a real risk of spending it on the wrong development partner.
Wakari's vision was an AI-powered accessibility platform that would replace a spreadsheet-based questionnaire with a natural-language chat interface. The founder, Ofentse, had the product clarity but not the engineering background. The budget was fixed and small, which raised the stakes on every decision.
The risk profile was clear from the first call. Picking the wrong development partner could eat the budget without delivering the platform. Trying to do too much too soon with a frontier model could push the running cost beyond what the business case supported. Skipping validation could lock in a design that residents would not use.
- Fixed budget, no flexibility.
- Non-technical founder needs to manage the build, not be at the mercy of it.
- Concept must be validated cheaply before committing to any expensive development.
- WCAG 2.1 AA is the accessibility bar the build has to clear.
Cheap validation first, founder fluency second, the right build partner third.
We explored low-cost paths first: a custom GPT through OpenAI, low-code prototyping, proof-of-concept builds. The idea was to validate the spreadsheet-to-conversation pattern with real people before committing to a full build. Two of those prototypes shipped, got feedback, and shaped the eventual platform requirements without burning the budget.
Alongside the prototypes ran founder education: how language models work, where they fail, what they cost to run, and how to recognise a development partner who is selling something the project does not need. That fluency carried through to vendor selection and ongoing partner management once a build team was chosen.
- Custom GPT and low-code prototype paths for concept validation.
- Founder-level AI literacy programme covering model selection, cost, and limits.
- Red-flag framework for development-partner evaluation.
- Validation pathway sized to derisk each step before commitment.
Platform built and launched. Budget held. Founder running the next phase with the vocabulary to do so.
Concept validated cheaply
Custom GPT and low-code prototypes proved the conversational pattern landed with users before any expensive build started.
Budget held end to end
Phased validation and partner selection kept the engagement within the original fixed budget. No over-runs, no scope creep.
Founder fluency
The vocabulary to brief a developer, evaluate a quote, and challenge a recommendation. Ofentse now manages the platform without an engineer translating.
Platform live
The accessibility platform shipped to its intended audience. The pattern proved transferable to similar low-budget AI builds in adjacent sectors.
Consulting with Reuben from OpenKit was super helpful. His expertise in Generative AI and Large Language Models provided critical insights into our project's technical and business aspects. Reuben offered clear guidance on feasibility, budgeting, and timelines. His dedication and passion for AI development was evident as he communicated complex concepts and helped us understand ways AI could be used in our project. I highly recommend Reuben and OpenKit for their professionalism, expertise, and commitment to client success.
Ofentse Lekwane Founder, Wakari · Wakari
How we delivered it.
Stack
Capabilities
Compliance
From scoping to live.
- Founder AI educationHow language models actually work, where they break, what they cost. The vocabulary a non-technical founder needs to manage a development partner without being talked over. Month 1
- Concept validationCustom GPT and low-code prototypes used to test the spreadsheet-to-natural-language idea without committing to a build budget. Month 2
- Partner selectionRed-flag framework for evaluating development partners, scope documents, and a validation pathway that minimised budget risk. Month 3
Read another story
All case studies
Bring your team's next AI project to a 30-minute call.
No deck. We listen, sketch a delivery shape, and tell you honestly whether AI is the right tool for the problem.
