Every day
Agents inside the workflow
Not the occasional prompt: a configured workspace. Written rules, memory that carries over between sessions, reusable procedures. The AI reads the company's files where they are and in the format they exist in.
- Analysis and benchmarks — market research, competitor work and taxonomies built on verified, cited sources.
- Long documents — tender specifications, regulations, contracts, policies and reports read and compared in a few hours.
- Measurement systems — KPI dashboards, scoring models, capacity limits: from the indicators to the spreadsheet that computes them.
- Recurring processes — assignments, onboarding, operational checklists: the tasks that repeat, with information scattered across files.
- Handover — the people I work with learn to use them: the tools stay with the company after the project ends.
When it becomes a project
LISA
Educational Travel
I led the implementation of the project together with the IT team and the AI engineers, working on how information is structured and on how the data is accessed.
When the system recognises a client it automatically opens their file and builds the answer on real data:
product purchased | payments | contractual and insurance terms | logistics and operations
The result is a digital hub designed to be read and queried by an AI agent, not simply consulted by a person.
−87% on response times, with more accurate answers because each one is generated directly from that client's own file.