The method
The problem I solve: pipeline leaks
A sales pipeline rarely loses its prospects all at once. It loses them through small leaks: a prospect who waits for a reply and goes elsewhere, context dropped between two channels, a follow-up nobody sent, an appointment booked and then missed. Their sum shows up in revenue.
These leaks open between your tools and your teams. My job: installing the AI and automation systems that give every prospect a fast reply and every case an owner and a next action, while your salespeople keep the relationship.
The operating loop
Every system I deploy runs the same loop: see what is really happening → bring the information together → prioritise → assign the next action → accelerate → verify the result.
The stages of an engagement
- Baseline & Blueprint (paid): mapping the real journey: volumes, delays, channels, owners, SLAs, architecture, risks, metrics. Delivered with a firm quote. No engagement starts without this baseline.
- Productised pilot: fixed scope, fixed price, minimal integrations, acceptance criteria set in advance, team training included. A pilot addresses one bottleneck, not ten.
- Managed system: monitoring, failure tracking, SLA control, performance reporting, bounded optimisation. Operable through procedures, not through one person's memory.
- Expansion: a second pipeline stage, a second system, multi-site, only after measured proof.
The offers themselves: the services page.
What makes a system a system
A module without a trigger event, a source of truth, a bounded action, a human owner and a metric is a demo, not a system. Every critical case needs a visible state, an owner, a next action and proof of result.
What gets measured
Direct metrics are promised: time to first reply, share of requests handled within SLA, contact rate, qualification, next step accepted, time to human takeover, lost requests.
Business effects are observed: sales, margin, NPS, reviews, loyalty. Never promised without defensible causality.
Conditions for a good pilot
A pilot launches when the problem is frequent and costly, an internal sponsor owns the journey, the minimum data is accessible, a direct metric can move within a short window, the expected human actions are explicit, and stopping or escalating is clean.
What I refuse
- Revenue or NPS promises without demonstrable causality.
- Replacing the existing information system on the first pilot.
- Unlimited support without a defined scope.
- Financial, contractual or security decisions made by the AI on its own.
- Collecting sensitive data without a defined purpose or retention period.
- "Adding AI" without an identified journey or an accessible decision-maker.