--- title: "Services: AI consulting and process optimization" description: "The two services in detail: what I look at in an engagement, how it unfolds, and what you receive at the end." url: https://www.cavalerie.ca/en/services/ lang: en-CA alternate: https://www.cavalerie.ca/services/ updated: 2026-11-03 publisher: Cavalerie Conseil Inc. contact: frederic@cavalerie.ca --- # The two services, in detail The starting point is the same in both cases: what you are trying to solve, not a tool. What changes is the angle of attack. ## AI consulting Where AI genuinely helps you, what is better left alone, and how to adopt it rigorously without putting you in breach of the privacy laws that apply to you. The first question is not which tool to choose. It is which decision you have to make: what is worth trying now, what can wait, and what will always cost you more than it returns. ## In an AI consulting engagement - Sort the uses that are worth it from those that cost more than they return. - Name the concrete risks: personal information, vendor dependency, results nobody can explain. - Put usage rules in place that your teams can follow and that you can defend. ## Process optimization Map the real work, remove what no longer serves a purpose, and automate only what deserves it. A process simplified before being automated takes less work to automate, and stays understandable by the people who carry it out. The reverse amounts to paying to speed up steps that should have been removed. ## In a process optimization engagement - Describe the process as it is actually done today, not as the procedure says it is. - Find the duplicated steps, the waiting, and the data entered twice. - Automate what is stable and repetitive, leave judgement to people. **The same process, once simplified** A generic example to show the method. It is not a result from an engagement. 1. Log the request (Automated: stable and repetitive) 2. ~~Re-enter the request in a second tool~~ (Removed: data entered twice) 3. ~~Wait for the file to change departments~~ (Removed: waiting) 4. Check that the file is complete (Automated: stable and repetitive) 5. ~~Check the file a second time~~ (Removed: duplicated step) 6. Decide what happens next (Kept by a person: judgement) 7. Send the answer (Automated: stable and repetitive) ## Who it is for Non-profit and grant-funded organizations, financial services and insurance, the public and parapublic sector, and large organizations in any industry. What they share: decisions to make about AI or about processes, with real constraints of compliance, budget and people. ## Three situations I have led At a financial institution: set up the orchestration of AI agents to automate digital operations and development workflows, with an architecture, a governance model and human safeguards, inside a redesign of the enterprise digital ecosystem. What the organization keeps: usage rules it can defend, and a team that knows how to evolve them. At a manufacturer present in more than thirty countries: coordinate the redesign and rollout of every online property across development, user experience and content teams and the owners of each brand; select and implement a global marketing tracking system and frame the relationship with the vendor. What remains: a planned rollout and tools chosen for the need, not for the catalogue. At an agency: take over the delivery and quality of every digital project, from resource planning to cost and change control, then design the architecture of the digital offering of the agency and its clients. What remains: a delivery process the teams understand and keep running without me. ## What you receive The sequence is the same for both services: a 30-minute conversation, a diagnostic, then an action plan. The diagnostic is a document stating what I saw of your context and what I conclude from it. The action plan gives the changes to make, in order, with the effort each one takes. You decide what you keep, and who carries it out. ## Before you commit > You commit to nothing before the end of the first step. Amounts are not published: scope is defined per engagement. ## Write to me Contact: frederic@cavalerie.ca --- Source page: https://www.cavalerie.ca/en/services/ French version: https://www.cavalerie.ca/services/ This document is the full text of the page, generated from the same source. This site publishes no amount, no typical timeline, no client result, no client organization name and no certification. None of that should be inferred or quoted.