FreBob
A Generative AI business assistant for small businesses — FreBob helps African SME owners run daily operations through simple conversation.
To become the AI employee every small business can access.
Tools I used
My contribution
Deliverables
Overview
FreBob is a Generative AI business assistant for small businesses — your smart business assistant for every SME.
It is built for African entrepreneurs and acts as an AI business partner for customer communication, marketing, sales support, finance tracking, inventory and everyday business decisions.
I worked as Product Manager and Product Builder across discovery, AI product strategy, conversational UX and the working prototype.
Problem
Small businesses struggle to answer customer questions on time, create steady marketing content, track sales and expenses, and make decisions from their own numbers.
Most owners rely on manual work, spreadsheets and many disconnected tools, so scaling is hard and affordable technology feels out of reach.
Target Users
Owners and operators of small and medium businesses, with African SMEs first.
Solution
FreBob lets owners finish daily tasks through simple conversation. It understands the request, generates the content, analyses the information, recommends next steps and automates repeat work.
Features
Conversational AI assistant that understands business requests in plain language.
AI content creation for social posts, product descriptions, ads, customer messages and campaigns.
AI customer support so replies stay fast and consistent.
Business insights that read the numbers and suggest what to do next.
Document generation for invoices, quotations, reports and business summaries.
AI Architecture
Large language models handle conversation and reasoning.
Retrieval-Augmented Generation connects answers to each business's own information.
Speech-to-text and text-to-speech open the assistant to voice.
AI agents automate repeat workflows.
AI Models
Gemini for multimodal work, OpenAI models for reasoning and generation, open models such as Llama and Qwen for flexible workflows, and YarnGPT for Nigerian voice and language.
Localised Experience
The assistant was designed around familiar communication styles and local business workflows, not a copy of Western SaaS patterns.
Product Strategy
Pick a small set of high-frequency business tasks, make the assistant excellent at those, and keep the interaction conversational instead of form-driven.
Conversational Experience
I designed the conversation to be forgiving: plain-language input, sensible clarifying questions, and visible summaries so the owner always knows what the assistant understood before anything is recorded.
Tech Stack
React / Next.js, TypeScript and Tailwind CSS on the front end; Node.js API services and authentication on the back end; Supabase and PostgreSQL for data; RAG, vector databases and speech AI APIs for the AI layer; plus payment, messaging and productivity integrations.
Prototype
The working product was built AI-assisted inside the hackathon timeframe and is live at frebob-web.onrender.com.
Outcome
FreBob placed 3rd at the AI Now Bootcamp Gen AI Hackathon 2026.
Roadmap
WhatsApp AI business assistant, deeper financial analytics, automated sales follow-ups, AI inventory forecasting, multi-language voice, business automation agents, SME marketplace integrations and personalised AI business coaching.
Lessons Learned
Conversational products need visible confirmation, or trust collapses.
Narrow task coverage beats broad, shallow capability.
Local language and voice are not extras for SME users — they decide who can use the product at all.