Solvit AI Strategy Session

Overview

Solvit runs a network of field inspectors across Kenya, Uganda and Tanzania. Martin Kiarie, the CEO, wanted to know where AI agents could take work off his team without replacing them. We ran a two-hour discovery call on March 25, produced twelve raw discovery documents, then shipped a full proposal package: system architecture, seven agent specifications, a ROI model built from Solvit's own Zoho numbers, a phasing roadmap, a change management brief, a data model, the formal proposal, a one-page executive summary, and an interactive microsite deployed at kaara.works/solvit. After Martin reviewed the material he asked us to start with a strategy session first rather than a full build. We rescoped to a five-hour in-person workshop in Nairobi on Friday April 10, 2026 for six Solvit leadership attendees, generated a session brief, a quotation, and a signed service agreement as PDFs, and built a second lean proof microsite at kaara.works/solvit/lean-proof.html covering the OpenClaw and NemoClaw phase zero plan. Along the way we surfaced two insights that reshaped the engagement: the nine-hour approval lag is a blind FIFO queue problem, not a slow approver problem, and the 73.4% scheduled-to-inspected drop is driven by solver hoarding behavior tied to pay-per-inspection incentives.

Key Features

• Two-hour discovery call with Solvit CEO Martin Kiarie
• Twelve discovery documents covering people, tech, problems, metrics and data flow
• System architecture document for the Solve It IQ agent layer
• Specifications for seven Claude-powered agents across approval, intake, scheduling and insight
• ROI model built from Solvit's Zoho dashboard numbers
• Phasing roadmap from lean proof through platform integration
• Change management brief and PostgreSQL data model
• Interactive proposal microsite deployed at kaara.works/solvit
• Lean proof microsite with six pipeline visualizations and the OpenClaw plus NemoClaw plan
• Five-hour in-person AI strategy session for six Solvit leaders in Nairobi
• Session brief, quotation and signed service agreement PDFs generated from Python
• Pain point synthesis and per-attendee dossiers for the Nanyuki working session
• Blind-queue and solver-hoarding insights that reshaped the proposed agent priority order