Confidential
Confidential·Architect and operator·2026 · 4 months

An outbound acquisition pipeline fully driven by AI agents.

Daily ICP job-listing intel, agent-driven enrichment and scoring, personalised icebreaker generation, multi-channel sequencing across LinkedIn and email, real-time pipeline supervision. One operator runs the whole thing.

An acquisition system that doesn't scale with SDR headcount.

Any B2B services business hits the same equation: pipeline regularity depends on the number of people you pay to prospect. The client wanted the opposite, a system where regularity depends on infrastructure rather than headcount. A common problem for most growing B2B businesses.

Funnel modelling before the first line of code.

First deliverable: a twelve-page scoping document. Reachable ICP market sizing, prioritised buying signal, expected stage-by-stage conversion rates, contact volume needed to generate N monthly meetings, marginal cost per booked meeting. The system was only built after explicit sign-off on the numbers.

An orchestrated architecture, operable by a non-developer.

N8N workers and Python scripts for intel and enrichment, Claude agents for ICP scoring and icebreaker writing, orchestrated sequencing across LinkedIn Sales Navigator and Smartlead, pipeline and performance tracked in Supabase. Fully documented and handed over to a non-developer operator after a single training session. No residual dependency on engineering.

What it produces.

30 to 40
LinkedIn connection requests per day
<48 h
between signal detection and first contact
1
operator runs the whole thing
Let's talk in 30 minutes.
Got a similar project?
Pipeline d'acquisition IA | Ceres Broker