Civ Robotics / SolarIntel

Sales started from cold company lists, not live projects.

Civ sells robots into solar and infrastructure construction. The team's highest-value question was simple to state and slow to answer: which real projects are breaking ground, which contractors are on them, and who is the person to call.

Answering it by hand took about an hour per lead. That hour was the bottleneck for the entire pipeline.

Start with the most painful workflow, earn trust, then widen.

I partnered directly with the CEO and COO to translate how the company actually worked into a system the team would use. We started with lead generation because it hurt the most and the outcome was easy to measure.

Once the first system was in daily use, the same approach extended to a product knowledge ontology and to customer support.

Parallel sub-agents, retries, and citations on every record.

Each target is researched by a multi-agent pipeline. Parallel sub-agents cover permits, public reporting, and context sources. Failed work is retried instead of silently dropped. Every claim carries a citation, so a rep can check the source before picking up the phone.

The output is not a report to read. It is an actionable record: the project, the contractors involved, and the responsible person.

One shared source of truth for people and agents.

I connected Civ's sales, support, and operational tools through automated ingestion. The data lands in a self-updating product knowledge ontology that employees and agents read from the same place.

That layer is what let the system grow past lead generation without re-plumbing everything for each new use case.

• An AI operating system customized for GTM lead generation, product knowledge, and customer support, in daily team use.
• A multi-agent research pipeline with parallel sub-agents, retries for failed work, and enforced citations.
• Automated ingestion across sales, support, and operational tools into a self-updating product knowledge ontology.

Actionable leads

4,000+Generated by the research pipeline.

Research per lead

~1 hr → <5 minManual research replaced by cited agent output.

Revenue supported

$200K+Attributed to workflows the system runs.

Adoption

Daily useAcross the sales and support team.