What I can deploy
An entrepreneur, a salesperson by nature and an engineer by method, I design and install the AI and automation systems that move a prospect through to a sale, then a client through to loyalty. Your salespeople keep the relationship; the system takes the first reply, qualification, appointments, follow-ups and case tracking.
The building blocks below are grouped by pipeline stage. They share one skeleton: a trigger event, a source of truth, visible state, a human owner, a next action, a clean escalation path and a measurement. They sit inside your existing tools without replacing them. The exact scope is set during the Baseline & Blueprint, from the real journey, never off a catalogue.
Acquisition: first reply and qualification
Every inbound request (form, message, missed call, social channel, email) gets an answer within minutes, at any hour. The system qualifies it, routes it to the right person with a usable brief and tracks it until a human takes over. A request outside the scope escalates. This is Speed-To-Lead.
Upstream, the website itself can answer: visitors ask their question, the answer comes from a knowledge base you own, and the system cites its source rather than improvising. The same content is exposed in a format ChatGPT, Perplexity, Claude or Gemini can read and cite.
What it measures: time to first reply, contact rate, qualification rate, lost requests.
Appointments: booking, reminders, attendance
A qualified prospect is offered a slot within the conversation, in the right salesperson's calendar. The system confirms, sends a reminder before the meeting, offers a new slot when the person doesn't show and records the outcome. The salesperson walks in with the context already gathered.
What it measures: booking rate, attendance rate, time from request to appointment.
Sales: follow-ups, case tracking, handover
A quote that has been sent carries visible state, an owner, a next action and a deadline. The system follows up on a set rhythm, escalates when things stall and hands over to the salesperson as soon as a decision needs a human: negotiation, objection, commercial gesture. Context travels with the case from one channel to the next.
What it measures: follow-up delay, unanswered quotes, conversion rate per stage.
Client relationship and retention: after-sales and complaints
After the sale, the same principle applies to the client's case: service job, order, complaint. The client knows where their request stands, a human stays accountable, nothing stops for lack of a follow-up. One end-to-end example is Claim-to-Cash: from a declared claim to a repaired vehicle and collected funds, tracking assessment, quotes, parts, repair, return, invoicing and payments.
What it measures: processing time, stuck cases and for how long, closed complaints.
Steering: cockpit and measurement
A single view of what matters, fed automatically by the blocks above, instead of the spreadsheet one person updates by hand and nobody trusts any more: journey metrics, the state of open cases, what is stuck and since when. Recurring jobs run on schedule and raise an alert when they fail.
Other building blocks
Repetitive production chains (reporting, market monitoring, file preparation) run by agents, with human validation before anything goes out. I deploy them when they prepare a salesperson's work, never on their own.
The tools I work with
I'm neither a reseller nor a vendor's integrator: tools are chosen during scoping, starting from yours. Day to day: Claude, Codex, ChatGPT, Gemini and the MCP protocol; TypeScript, Next.js, Python, Node.js, GitHub, Vercel, Cloudflare, Playwright; PostgreSQL, Neon, Google Sheets, Obsidian; n8n, Notion, Trello, Google Workspace, Telegram, Cal.com; Resend, Canva, Remotion, OBS, Google Search Console, Google Analytics.