FR US
NewAn AI setter that calls your leads back at the time they picked

Software and AI agents built for the way you work.

Libreo automates the repetitive work of service businesses (client follow-up, replies, callbacks, reporting) with custom tools that your team supervises and that are measured in results.

Let's talk about your project See the case studies
5 systems in productionHuman oversight built inHosting in the US, the EU or on-premise
libreo · agent consolein production

Agent activity · today

live
Hours saved · this week31.6 h+12% vs last week
Tasks handled41898.1% with no rework
Human approvals234 pending

In production

ClaudeOpenAIMistral AIMakeGoHighLevelHubSpotTwilioPython

The building blocks we orchestrate · picked to fit your constraints.

§ 01 · Valuemaxxing

No tokenmaxxing. Valuemaxxing.
“Maxxing” is the art of optimizing everything. Libreo applies it to what actually matters: hours given back to your team, clients who get followed up, qualified appointments on the calendar. Not to how much AI you burn through.

FIG 0.1

Built around your business

Every tool starts from how you work, your documents and your rules. Not the other way around.

FIG 0.2

Supervised by your team

Agents prepare, propose and execute. Your people sign off on anything that commits the business.

FIG 0.3

Measured in results

Hours saved, response times, appointments booked: we track what the tool delivers, every month.

Read: tokenmaxxing vs. valuemaxxing, what AI should really maximize

§ 02 · What we build

Eight ways to give your team its time back.

Hover over a tile to see the tool at work. Each solution runs on its own or combines with the others in a single project.

Voice agents

AI setters & speed-to-lead

Your leads get a call back at the exact time they picked, are qualified against your criteria, then handed to your closer with an appointment already in your CRM.

Callback at the time they choseQualification on your criteriaAppointment booked in your CRMAudit PDF sent to the closer
hover ▸ See the AI setter case study
AI assistants

Assistants for your clients and learners

Explain, grade, follow up, motivate, using your content and your tone of voice.

See the Yoocancare case study
Data analysis

Your numbers, as clear answers

Ask a question in plain English, get a figure and a next step.

See the data analysis service
Automation

Workflows that run themselves

Form, CRM, quote, follow-up, invoice: no more copy-pasting between apps.

See the automation service
Custom software

Software that fits your field work

Jobs, files, schedules: all your information in one place.

See the custom software service
AI receptionist

A front desk that never misses a call

Bookings, common questions, transfer to a human when needed. Nights and weekends included.

See the voice agents service
Custom models & private AI

AI that stays in your building

Open-source models installed on your own machines, for data that must never leave.

See the private AI case study
Lead reactivation

Revenue hiding in your old leads

The agent re-engages dormant contacts who opted in, by text and by phone, and books those who raise their hand.

See the lead reactivation service
§ 03 · Under the hood, in plain English

What a Libreo agent does, step by step.

Scroll down: the interface builds itself as you go. An example client follow-up agent for a service business.

01

It plugs into your tools.

Email, CRM, calendar, files, business software: the agent connects to what you already use, with precise permissions. No need to replace anything.

02

It reads, sorts and understands.

Every request is read, categorized and summarized according to your rules: client type, urgency, next action.

03

It acts, under your control.

It replies, books appointments and schedules follow-ups. Anything that commits the business (a quote, a discount) waits for your green light.

◆ Human approval
04

You measure what it delivers.

Every week: hours saved, response times, appointments booked. Valuemaxxing, in practice.

agent · client-follow-up
connectors.yaml connected
# Sources allowed for the “Client follow-up” agent
sources:
  - crm: gohighlevel     # contacts, appointments
  - calendar: google     # availability
  - email: imap          # incoming requests
  - files: /quotes       # read-only

permissions:
  write: [appointments, notes, follow_ups]
  human_approval: [send_quote, discount]

report: every Monday 8:00 AM
Incoming requests · sorted 5 new
REQUESTTYPEPRIORITY
Quote · Miller WoodworksQuoteHigh
Timeline question · Mr. RossInfoNormal
Complaint · Nova StudioSupportHigh
Invoice reminder · Morgan CPABillingLow
Sign-up · J. DialloSign-upNormal

→ summary sent to the team · 5 requests, 2 high priority

Agent actions running
Reply sent · Mr. Ross · timeline confirmed
Appointment booked · Miller Woodworks · Thu 10:30 AM
7-day follow-up scheduled · Morgan CPA

Nova Studio quote · 8% discount proposed. Approve sending?

EditApprove
Weekly report W45
Hours saved31.5 h
Response time6 min
Appointments14
Hours saved per week

§ 04 · Case studies

Five systems in production.

Every project started from a real need and runs today for its users. Figures are dated and kept up to date. Most of these clients are in France: the methods carry over as-is.

In productionProfessional training

Yoocancare: four AI assistants supporting every learner

900+ learners supported: Claire explains the lessons, Clémence grades assignments, Constance checks in every Monday and Prudence helps them launch their practice.

4assistants
900+learners
24/7answers for learners
Read the case study
In productionDigital services

Netcoaching: a digital agency where agents produce and humans coach

Websites, ad campaigns, blog posts and Google Business Profiles: production is largely automated, with human sign-off at every key step.

200+clients supported
140+deliverables produced
48 hfor a small-business website
Read the case study
In productionSales pipeline

AI setter: every lead called back at the time they picked

For an ad campaign offering a free audit, the agent calls leads back, qualifies them and books the closer's appointment in GoHighLevel.

100%callbacks on time
38%leads qualified
0manual data entry
Read the case study

All five production systems and the ready-to-deploy solutions

§ 05 · Method

One useful tool, fast. Then we expand.

No big, fuzzy project: we start with one specific task and a measurable result, and widen the scope once real usage proves the value.

  1. 01

    Understand

    A call to walk through how you work and spot what eats up your time.

    30 min · free
  2. 02

    Scope

    A first scope, a measurable goal and a clear budget.

    1 week
  3. 03

    Build

    Development and testing on your real data, with your users.

    2 to 6 weeks
  4. 04

    Go live

    Deployment, team onboarding and first adjustments.

    1 week
  5. 05

    Measure and grow

    Results tracking, maintenance and new features.

    monthly plan
§ 06 · In their words
“I used to spend my Mondays chasing learners one by one. Now Constance does it, and I only handle the situations that really need me.”
CSCarine SimonHead of Training, Yoocancare (translated from French)
“The AI setter calls every lead back right on time, even on Saturdays. Our closers only get qualified appointments now, with the audit already attached.”
TMThomas MartelHead of Sales, 3Xpansion (translated from French)
olivier-abou · founder
Olivier Abou, founder of Libreo
§ 07 · Who you'll work with

Olivier Abou. Linguist by training, builder of governed AI agents.

PhD in natural language processing, author of 100+ technical books and a tech journalist for 15 years, I started out in the AI of rules and dictionaries.

That is still my method: each agent works on a narrow domain, with your sources and your rules. The result: answers you can verify, and far fewer errors. One point of contact, from the first call to go-live.

My method: governed AI My background
§ 08 · FAQ

What business owners ask us first.

How do you keep AI errors and “hallucinations” in check?
By design. Each agent works on a narrow domain, from your documents and validated business glossaries, with rules written in advance and a step-by-step reasoning process it must follow. When information is missing, it says so instead of making something up. My method →
Do we need to replace our current software?
No, in most cases. Agents connect to the tools you already use: CRM, calendar, email, business software. Custom development is only proposed when no existing tool will do.
You're based in France. How does that work for a US company?
Everything runs remotely: video calls, shared workspaces, deployments on your accounts. France is about six hours ahead of New York, so calls are scheduled in your morning, and the work moves forward while you sleep. Quotes and invoices are in US dollars.
Where is our data hosted?
It depends on how sensitive it is: with a US or EU cloud provider, or directly on your own servers with an open-source model. We decide together during scoping.
How much does a project cost?
A first, focused tool usually runs between $2,000 and $10,000, followed by a maintenance and improvement plan starting at $99 per month. You get a firm quote after scoping.
Who stays in charge of decisions?
You do. Agents prepare and carry out the tasks we define together. Anything that commits the business (sending a quote, a discount, a sensitive reply) can require human approval.
What does “valuemaxxing” mean?
It's the opposite of tokenmaxxing, which means using as much AI as possible without measuring what it returns. Valuemaxxing maximizes the value produced: time saved, clients better served, appointments booked. Read the article →
Let's talk about your project

Tell us about the task that eats up the most time.

A 30-minute call is enough to know whether a tool can take it on, and what it could give you back.

Let's talk about your project

hello@libreo.io · (815) 365-3829 · Lyon, France