Finance
- Trigger
- Reconciliation differences at period close
- The agent
- Analyses the differences, gathers the supporting evidence and proposes the matching, with an explanation.
- The human
- The accountant approves the entries.
Agentic AI advisory
Let your organisation’s intelligence rise.
Ostenia helps leadership teams decide which processes to entrust to AI agents, take them into production and keep them under control. A partner at every stage, from the first decision to results.
Ostenia at a glance
- 20+ yearsof experience for each of our three partners, including a leadership career at Deloitte and then EY
- 45 daysto know where AI agents create value for you, at what cost and with what risks
- Independentof vendors: no licences or platforms to sell – only what serves your case
- DTS ConseilOstenia is its subsidiary, dedicated to agentic AI
Why now
AI agents no longer just answer. They act.
An agent is given an objective, works across your applications, carries out the steps and reports back. Once software acts on behalf of the organisation, the question is no longer technical – it belongs to leadership.
Yesterday
Assistants that answer
You ask a question and get a text back. People do everything else.
Today
Agents that act
They read a file, cross-check the data, prepare the decision and execute what they are entrusted with.
The rules
Obligations that apply
The EU AI Act has been coming into force in stages since 2025. Better to map your uses before you deploy new ones.
The question for leadership
Who controls what?
Which processes to entrust, on what terms, under whose responsibility: this is where we come in.
Self-assessment · 10 questions · 5 minutes
Where does your organisation stand? Start here.
A maturity profile, a reading across five dimensions and three priorities. No sign-up: your answers stay in your browser.
What place does agentic AI hold in your strategy?
- The topic has not yet been raised at executive committee level.
- It has been discussed, with no decision taken.
- Priorities are set, but with no dedicated budget.
- A roadmap is approved, with a budget and a sponsor.
Your answer opens the rest of the assessment.
Expertise
Six areas of expertise take you from the first decision to an agent in production.
- 45 days
Agentic roadmap
In 45 days, a clear view of where AI agents will create value in your organisation – at what cost, and at what risk.
Agent design and deployment
We design and integrate AI agents that work inside your existing tools, and take them into production – under human control.
AI governance and trust
Hold your agents to the same standards as your people: a defined role, a named owner, clear limits and full accountability.
- 12–36 month trajectory
IT strategy and data architecture
Systems and data built to carry AI: a clear diagnosis, a target state and costed scenarios.
AI programme management
Bring data and AI programmes back on track – and set new ones up to succeed.
- One day
Executive seminar
One day for your executive team to understand agentic AI, select its first use cases and set the rules of the game.
The partners
Every engagement is led by a partner, from the first conversation to results.
Strategy, finance, platforms: three careers in major organisations, brought together around your decisions.
Meet the partnersApproach
Decide first. Build next. Govern always.
- Stage 1 · 45 days
Decide
Assessment of your data and systems, costed use cases and a 12-month roadmap signed off by leadership.
- Stage 2
Design
Agentic workflows, model selection, human decision points, control rules and an evaluation protocol.
- Stage 3
Deploy
A measured pilot on a live scope, then a phased roll-out to production, side by side with your teams.
- Stage 4
Govern
Ongoing monitoring of quality, cost and risk, team upskilling and regular reviews.
At every stage, leadership signs off before moving to the next. Agents remain under human control.
Use cases
A good use case fits in three lines.
A trigger, what the agent does, what the human decides. Three typical examples, framed the way we work on them.
Legal
- Trigger
- A contract received from a supplier
- The agent
- Compares each clause with your standard clauses and flags deviations and risks.
- The human
- The legal team decides what to negotiate.
Customer service
- Trigger
- An incoming request, by email or web form
- The agent
- Handles simple requests and prepares a complete file for the others.
- The human
- An adviser takes over sensitive cases.
Frequently asked questions
The questions to ask before you start.
Our data is not ready. Should we wait?
No. That is often the case, and it is what the first weeks are for: establishing which data is enough for a first agent, and what needs preparing for the next ones.
We already work with an IT services firm or a vendor. What do you add?
An independent perspective, on leadership’s side. We sell no licences and no platforms: we help you choose, frame and control – and we can work alongside your existing providers.
What happens to our documents and data?
They remain yours and are used only for the engagement. We sign a non-disclosure agreement if you wish. The choice of models and hosting, sovereign options included, is yours: no imposed lock-in.
Does the EU AI Act apply to us?
As soon as you use or deploy AI systems, it applies to you to some degree. It classifies uses by level of risk: some, such as recruitment or creditworthiness assessment, are deemed “high-risk” and carry specific obligations. We inventory and classify your uses before we build.
Who actually does the work?
A partner leads every engagement, from the first conversation to results. Together, the three partners bring what a successful agent requires: a sound decision, a sound investment and a sound system.
Where do we start?
With a one-hour first conversation with a partner, free of charge and with no obligation. Depending on your situation, we then propose the 45-day agentic roadmap or a one-day executive seminar.
Our principles
Our principles
“What your organisation knows, your agents put to work.”
People decide
Every agent has an owner and clear limits. Decisions that commit the organisation remain human.
Vendor-independent
We sell no licences and no platforms. We recommend what is right for your situation.
Measure what matters
Quality, cost, speed, risk: every agent is evaluated before and after go-live.
The simplest thing that works
No agent where a simple rule will do. No demonstration without a next step.
Insights
Our insights are short, and written for decision-makers.
Abdoul Seck, Abdoul Diallo and Thierry de Lapasse
Let’s talk about your first agents.
A one-hour conversation with a partner, free of charge, to understand your situation and give you a candid view of where to start.