We do not name our clients or publish their figures without written consent. What we describe here is the architecture and design choices—in other words, what makes it possible to assess our work.
When a project is signed in the CRM, a sequence is triggered: the document folder structure is created, the tracking file is initialized, the management database is updated, and the relevant team is notified. The difficult part was not the sequence itself, but its reliability: the trigger is idempotent, so a webhook replayed by the platform never creates a second file, a common flaw in this type of integration.
An inbox containing tens of thousands of messages had become unusable. We audited volumes and senders, created a folder taxonomy suited to the actual business, categorized the archive, then set up sorting rules. A decisive choice: the bulk-processing phase was carried out using our tools, but the long-term system is a script attached to the client’s inbox that runs on its own every night. No subscription and no dependency on our infrastructure after delivery.
A system designed to be deployed for multiple clients under their own brand: detection of overdue payment dates, a graduated follow-up sequence, and immediate exit from the cycle as soon as payment is recorded. The exit logic is the sensitive point: following up with a client who has already paid costs more than the invoice.
Monitoring calls for tender, filtering based on specific criteria (sector, amount, area, keywords), and alerts only for notices that are truly relevant. The challenge with this type of system is not capturing everything; it is avoiding a stream that no one will read after three weeks.
An assistant used to draft, revise, and publish through conversations in an instant messaging app. Every version is retained, earlier versions can be restored, and an editorial context memory grows over the course of exchanges. A rule built into the architecture: no publication without explicit approval, along with protection against duplicate publication in the event of a technical replay.
A dashboard that monitors the availability of a group of websites and the presence of expected technical elements on their pages, with alerts when there is a discrepancy. It also lets us know that a delivered system is still running, rather than hearing it from the client.
Describe it to us in a few lines: which process, how many times a week, and how many people are affected. We will reply with an initial assessment of what can be automated and what cannot.
Response within 24 business hours · Quote after scoping · Confidential discussions