Skip to main content
All systems operational

Case study

AI sales agents · enterprise (under NDA)

Anonymous eval traffic becomes context for AI sales agents

An enterprise vendor under NDA runs a platform where AI sales agents and human reps co-pilot outbound. Leadpipe showed which accounts were actually on pricing and agent-governance pages so playbooks fired on real behaviour, not cold guesses.

Part 01

The company

The customer sells software for modern outbound: configurable AI sales agents that draft and sequence first touches, alongside human sales reps who own judgment, objections, and enterprise nuance. Buyers compare orchestration, safety rails, CRM sync, and brand voice before they book a deep dive. The company name and product brand are under NDA for this write-up.

Part 02

The challenge

Traffic was strong on agent-builder docs, pricing, and “automation vs. human outreach” comparison content, but few visitors identified themselves. RevOps saw sessions; GTM could not say which RevOps or sales leaders had come back three times.

Weekly high-intent accounts (indexed, ~8 wk)

Agent-led experiments need firm context. Without visitor identity in the same systems the agents read from, automation sounded clever to everyone and personal to no one.

Part 03

How Leadpipe helped

They ran Leadpipe on the marketing site and sent identified companies and contacts to the CRM with URL history and repeat-visit timing. Rules tagged accounts that combined pricing with integration or security views so human reps and AI sales agent playbooks branched on the same facts.

Weekly high-intent accounts (forms + chat vs + Leadpipe, ~8 wk)

100 idx

Forms + chat baseline

200 idx

Same cohort + Leadpipe

Reps got shortlists with page context; orchestration consumed the same fields for agent sends, with human-in-the-loop over a set confidence threshold. Consent and region stayed ahead of scale.

Part 04

What moved the numbers

In about eight weeks, weekly identified high-intent accounts about doubled versus forms and chat, mostly from repeat visitors who never clicked a CTA.

Outbound performance (same cohort)
MetricBeforeAfter
Weekly high-intent accts~100 idx~200 idx
First-touch reply~5–6%~12–15%

First-touch reply rate on Leadpipe-sourced outbound landed near 12–15% versus ~5–6% on cold persona lists; AI sales agent variants that referenced real page paths saw roughly a third fewer unsubscribes in the same cohort.

First-touch reply rate

Cold persona lists

~5–6%

Leadpipe-sourced

~12–15%

Pipeline reviews opened with “they hit agent governance and pricing twice this week,” cutting call prep and aligning product, marketing, and outbound on one narrative, without naming the vendor under NDA.