The Problem
A B2B sales operation needed a continuous flow of qualified leads — without standing up a research team and without paying for generic lists that go stale within a week. The South African market doesn't have the same off-the-shelf lead databases the US and EU enjoy, so manual research was the default, and manual research is expensive and slow.
They needed a system that could discover companies, enrich them with the right context, score them against an ideal customer profile, and route the best ones to the sales team — every day, without human intervention.
The Approach
We built a fully autonomous pipeline that runs unattended on a daily schedule.
The system discovers South African businesses across multiple sources, enriches each one with firmographic and contact data, scores them against the ICP using LLM-based reasoning rather than rigid rules, and routes qualified leads to the sales team's CRM with full context. Disqualified leads are tagged and parked, not lost — so the system learns from itself over time.
This is the same kind of autonomous agent system Single Source runs internally for its own pipeline. We eat our own cooking.
The Outcome
Hundreds of qualified leads generated per cycle, with zero manual research overhead. The sales team starts each day with a fresh, scored, contextualised pipeline instead of an empty CRM. Quality stays consistent because the scoring layer keeps adapting.
Most importantly: the cost-per-qualified-lead is a fraction of what manual research or generic list providers charge.