Home — ADVISORE, data and decisions

The observation

The data is there, the decisions wait

In most of the organisations we meet, the information is already there: ERP, SQL Server, PostgreSQL or MySQL databases, spreadsheets, business applications. What is missing is the short path between a management question and a reliable answer.

There is a great deal of talk about artificial intelligence, and very little industrialisation. Consumer tools have shown what is possible; what survives in production rests on other foundations: scoped data, explicit entitlements, written usage rules and trained teams.

What follows is a route: the observation, the five offers, the agents, on-premise models, training, decision intelligence, the industries, the method, our values, the founder. Read it in order, or go straight to what concerns you.

Our offers

Five offers, one single route

They follow one another rather than sitting side by side: make the data queryable, automate what deserves it, decide where the models run, train the teams, then frame usage through governance that actually applies.

Generative decision intelligence

Ask your question in French or English and get an answer built from your own databases. Entitlements respected, queries logged, no impact on production.

See the offer

Business AI agents

An agent receives an objective and carries out the sequence of actions it requires in your applications: search, check, enter, notify. It works under explicit permissions, with human validation for sensitive actions.

See the offer

On-premise LLM models

Open models running on your own infrastructure: confidentiality, sovereignty, predictable costs and independence from API providers. An internal gateway filters, logs and caps usage.

See the offer

Training and awareness

Four tracks — executives, business staff, IT and data teams, awareness workshops — built on your real cases. No tool is of any use if it is neither understood nor used.

See the offer

AI audit and governance

An inventory of what actually circulates in the company, a written usage policy, an entitlement matrix, continuous controls. A framework that can be enforced, not a decorative charter.

See the method

See the full catalogue of offers

Business AI agents

An assistant answers. An agent executes.

The difference is not technological, it is operational: an agent acts inside your systems, touches data and changes records. That is why it requires a stricter framework than a conversational assistant.

An assistant answers

You have to ask it the question, at the right moment, with the right words. It produces an answer: you check it, you carry it into the application, you type in the result. The value is real, but it depends on the user being available and paying attention.

An agent executes

It receives an objective, breaks the task down, calls the tools it is allowed to use, checks what comes back, tries again if needed, then reports what it did. You remain the judge of the outcome, but you are no longer the operator.

Seven business areas concerned

CRM and sales

Meeting preparation from the customer history, opportunity updates, follow-up on quotes left pending, account summaries.

Accounting and finance

Statement reconciliation, detection of missing supporting documents, preparation of statements, payment deadline tracking.

Human resources

Shortlisting applications against explicit criteria under supervision, interview preparation, leave and training tracking.

Production

Work order tracking, reporting of consumption variances, downtime entry, preparation of production reviews.

Logistics and procurement

Order and delivery tracking, comparison of supplier offers, stock-out alerts, import file preparation.

Customer service

Answers to recurring questions from approved documentation, ticket creation and follow-up, documented escalation to a responsible person.

Management reporting

Consolidation of key indicators, explanation of variances, a summary note built on identified and dated sources.

Three non-negotiable conditions

Explicit permissions

Every agent has an identity, a data perimeter, a list of tools and written limits on what it may do. It does neither more nor less than what is defined.

Complete logging

Every tool call is traced: who, what, when, on which data and with which result. What is not traced cannot be run in production.

Human validation

Sensitive actions — payment, commitment, sending to a third party — still require a person's approval. Without exception, even when the agent is right.

See the business AI agents

On-premise LLM models

Your models, your infrastructure, your data

Hosting the model instead of consuming an API changes three things: what may leave the company, the shape of the cost, and who decides on changes.

Confidentiality

The requests and documents being processed never leave your infrastructure. That point alone often decides the matter.

Sovereignty

You know where the data is, who can reach it and under which jurisdiction it is processed — and you can demonstrate it.

Predictable costs

An amortised infrastructure investment rather than a variable expense that follows usage and someone else's price list.

Independence

Changing model becomes an internal technical decision again: you are no longer bound to a third party's commercial policy.

When a public API remains the better choice: low and irregular volumes, non-sensitive data that may leave under a contractual framework, an immediate need for the best available model, or no server room and no operations staff. In both cases the same rule applies: whatever leaves the company is filtered, logged and decided by a written policy.

See the on-premise models

Training and awareness

Four tracks, four audiences

Executives, business staff, IT and data teams, awareness workshops: every session starts from your real cases and ends with an action plan.

Executives — understand and decide

What these systems can do today, what they cannot do, and what that changes for your organisation, your data and your business lines. How to judge an investment and set the framework for use.

Business staff — use it day to day

Framing a useful request, checking an answer, spotting a plausible mistake, documenting what was done. Exercises on your documents, your indicators and your procedures.

IT and data teams — run, administer, secure

Deploying and monitoring a platform or an on-site model, managing entitlements, reading the logs, measuring quality, responding to an incident. Hands-on work on your own environment.

Awareness workshops — open the subject

A session designed for one department or one management team: demonstrations on your processes, a collective mapping of the tasks to automate and those to protect, answers to legitimate concerns.

Training is not an add-on: it is the condition for use. A system that nobody knows how to question, check or challenge produces neither value nor safety.

See the training tracks

Generative decision intelligence

Query your databases in plain language

SQL Server, PostgreSQL, MySQL: your teams ask their question in their own words and get an answer built on your data, within the access rules in force. Every answer leaves a trail — question asked, query executed, data consulted — and no process writes to production systems.

Your existing entitlements apply: nobody sees beyond their own rights.
Every answer is auditable end to end, down to the query actually executed.
Read-only access, execution limits and timeouts: production is never touched.

Tier	Scope	Duration	Indicative investment
P0 Diagnostic	Framing, data audit, prioritised use cases, target architecture	4–6 weeks	XAF 9–15 m
P1 Benchmark & POC	Technology comparison on your real data, measured POC	8–10 weeks	XAF 25–45 m
P2 Pilot platform	Semantics, business agents, 20–50 pilot users	10–14 weeks	XAF 60–120 m
P3 Industrialisation	Security, monitoring, backup, training, go-live	12–20 weeks	XAF 90–250 m
RUN Operations	N2/N3 support, maintenance, cost supervision, new indicators	12 months minimum	from XAF 2.5 m/month

Indicative ranges, excluding taxes, confirmed by quotation after framing. Every tier ends with a decision: continue, adjust or stop.

See the GenBI & AI technical offer

Industries we serve

Eight industries, eight different contexts

We do not sell the same demonstration to everyone. Each industry has its own page: issues, use cases, typical pilot scope, governance and indicators tracked.