Skip to content
Phi-Lean.ai – True Lean software automation

Phi-Lean.ai · True Lean software automation

Flow first, then AI.

Phi-Core plugs specialised AI agents into the tools your company already uses: e-mail, calendar, CRM, ERP, accounting. They prepare the work, you approve it, it goes out. And before plugging anything in, we look at how work really flows in your company.

  • Hosted in Europe (GDPR)
  • Human validation before sending
  • Your data stays yours

A quote request, on a Tuesday evening

  1. 9:40 pmA family asks for a language-stay quote on your website.
  2. 9:42 pmPhi-Core qualifies the request and prepares the quote in your house style.
  3. 8:05 amYour adviser reviews it and approves in one click.
  4. 8:06 amThe family has its quote, before calling your competitor.
See this case in detail

Phi-Core: any business event in, finished work out

Phi-Core plugs into the tools an SME already uses, puts specialised AI agents to work and hands back results ready to use. You validate before anything goes out.

Triggers when something happens

Phi-Core the engine and its agents

Outputs what lands on your desk

Triggers when something happens

  • Inbox & messagingE-mail, Teams / Slack, WhatsApp message
  • Forms & websiteQuote request, contact form, chatbot
  • CRM & calendarNew lead, deal moved, meeting booked
  • Files droppedInvoice, CV or order lands in a folder
  • ERP, orders & shop floorNew order, overdue invoice, machine stop
  • Schedule & web watchEvery Monday 7:00, news, public tenders

Phi-Core the engine and its agents

Phi-Core

Connects your apps · runs the agents · delivers

  • AgentProfile Analysis
  • AgentCompany Understanding
  • AgentSynthesis
  • AgentYour next one

Human in the loop: you validate before anything goes out.

Outputs what lands on your desk

  • Drafts ready to sendReplies, follow-ups, posts in your tone
  • Briefs & synthesesClient brief, meeting prep, weekly digest
  • Documents generatedQuote, proposal, report in your house style
  • Systems kept up to dateCRM, ERP, accounting, dashboards
  • Tasks & approvalsRouted to the right person, one-click OK
  • Smart alertsAnomaly flagged with the next best action
Download the diagram (PNG)

Yesterday’s delighter is tomorrow’s basic

The Kano model separates what customers take for granted, what satisfies them more the better you do, and what delights them. Over time, delighters become basics. Quote response time is making that journey, fast.

Quote response time

  1. Before 2025

    A few days

    Performance

  2. 2026

    A few hours

    Delighter

  3. 2027

    A few hours, even minutes

    Basic

Phi-Lean projection, based on the Kano model (Kano et al., 1984).

Squeezed from above and below

From above
Large groups, with mapped processes and AI, lower their prices and move onto your ground: advice, proximity.
From below
One consultant with ten AI agents offers what a team of ten used to, with no office and no staff. There will not be three of them; there will be thousands.

After M. Gourmand, “The Systemic Attrition Spiral”, Phi-Lean Press, 2026

55%

of large European companies used AI in 2025, against 17% of small ones.

Eurostat, 2025, companies with 10+ employees, EU-27

Before automating, we come and see

Plugging AI into a broken flow just makes your defects faster. So the diagnostic starts on the ground, with your teams: people and flow first, and only then AI, where it really pays back.

  1. 01

    Map

    the processes, on the ground

  2. 02

    Remove

    the waste, release capacity

  3. 03

    Target

    where AI will really pay back

  4. 04

    Plan

    the AI roadmap

  5. Ready for AI

Then one flow at a time: the first one live, measured, then the next.

Three ways to get it wrong

Past revolutions, from robotics to the Internet, show three recurring mistakes.

  • WaitingYour competitors learn every month.
  • Plugging AI into a broken flowYour defects, only faster.
  • Automating to cut headsThe know-how leaves with them.

After M. Gourmand (2026), “The Systemic Attrition Spiral”, Phi-Lean Press.

Our flow analysis method, on phi-lean.com

Ten use cases, ten trades

All use cases

Typical scenarios, built from situations we meet in small and mid-sized companies: a dental practice, a travel agency, a kitchen retailer, a workshop. Yours will start from your flow.

Your AI is you

In 2024, Air Canada was ordered to compensate a passenger whose chatbot had announced a refund rule that did not exist. The airline argued the chatbot was a separate entity; the tribunal disagreed. What an AI says on your behalf commits you: that is why nothing goes out without you.

CBS News, 19 February 2024; Moffatt v. Air Canada, 2024 BCCRT 149.

  • Human validation before sending

    Every quote, reply or entry is put to you ready to go; you approve, adjust or reject.

  • Hosted in Europe (GDPR)

    Your flows run on infrastructure hosted in Europe, in line with the GDPR.

  • Your data stays yours

    It serves your work and nothing else. You decide what is kept.

  • Your tools, not ours

    Phi-Core plugs into what you already use. No migration, no new software to learn.

Five signs of a dinosaur company

Tick what sounds like your company, or that of a business owner you know.

Tick the signs that apply.

Talk about it
Five signs of a dinosaur company

Rooted in France and the UK, alongside you across Europe and beyond

Map: main presence in France and the United Kingdom; secondary presence in Belgium, Luxembourg, Switzerland, Spain, Portugal and Italy; projects in the rest of Europe, the United States, Brazil, China, the Philippines and India.
Manuel Gourmand

A platform by Phi-Lean

Phi-Lean.ai is the automation platform of Phi-Lean, the firm founded by Manuel Gourmand: ENIM engineer, MSc Lean Enterprise from the University of Buckingham, fifteen years in operations and supply chain in France, the UK, the Netherlands, Brazil and the United States. Here, flow consulting and technology talk to each other.

Discover the firm on phi-lean.com LinkedIn (opens in a new tab)

Which of your flows deserves to be automated?

Thirty minutes is often enough to spot the flow that will pay back most, and what needs fixing before any AI is plugged in. No commitment.