TRM Labs has introduced two new AI agents, Noor and Vera, designed to work as a team for competitive intelligence. Noor gathers and structures competitive briefs from external sources every Friday, while Vera conducts quality assurance by cross-referencing claims against source material. Previously, competitive updates were done monthly, often leading to stale information in sales conversations. Now, the process is weekly, with a human product manager performing a final spot-check before updates reach battlecards. The two-agent pipeline ensures high accuracy, as Vera catches nuances like misinterpreted feature tiers or incorrect claims about TRM's own capabilities. Early challenges included stale sources slipping through, self-referencing errors, and over-eager battlecard updates, which were fixed through mandatory date verification, cross-referencing internal resources, and training agents to distinguish between evergreen differentiators and weekly intel. Noor and Vera do not replace human judgment but significantly reduce manual effort, from hours to under five minutes of human involvement per cycle.

Key facts

• Noor gathers competitive intel; Vera reviews for accuracy every Friday.
• Two-agent pipeline replaces monthly manual updates with weekly automated ones.
• Human product manager does final spot-check before battlecard updates.
• Early fixes: date verification, internal cross-referencing, selective battlecard edits.
• Agents do not replace human judgment on deal-specific priorities.