Skip to content

Sales operations

Reference build

Two agents that hand sales a qualified shortlist

A multi-agent system where a research agent gathers company intelligence and a separate qualification agent scores fit against an ICP — saving reps 85% of their research time.

85%less rep research time

At a glance

Duration
17 weeks
Team size
1 person
Engagement
New build
Project type
AI & automation
Industry
B2B SaaS

The situation

The challenge

Reps spend more time researching accounts than talking to them, and a single agent asked to both gather and judge tends to rationalise whatever it found rather than score it honestly.

What we did

Separate the gathering from the judging, so the qualification step reads evidence it did not choose.

The calls that mattered

  • Research and qualification are different agents

    One collects, one scores against the ideal customer profile. Splitting them stops the system from grading its own homework.

  • Scores carry their evidence

    Every fit score links the signals behind it, so a rep can disagree with the reasoning rather than just the number.

What changed

reduction in rep research time
85%reduction in rep research time
specialised agents
2specialised agents
  • 2 — research and qualification, separated deliberately

Reps get a scored, evidenced shortlist instead of a research task, cutting time spent on account research by 85%.

Services used

  • Research agent with source-attributed company intelligence
  • Qualification agent scoring against a configurable ICP
  • Rep-facing shortlist with evidence links

What we would do differently

Every project has one of these. Publishing it is the point — a case study with no regrets in it is marketing, not evidence.

Multi-agent was justified here specifically because the two jobs have different failure modes. It is worth resisting everywhere the split does not buy that.

Services used

More work

Ready to start?

Start a project like thisWork