Projects built with the same no-code / low-code stack.
Automating Business Data Enrichment & Geolocation
Context — The startup mapping references candidate vendors whose actual business activity needs verification.
Need — Automatically verify each candidate company's activity status, and retrieve geolocation data to put them on a map
Selected initiatives — Built a process connected to the public open data Enterprise Search MCP to automatically verify each company's activity based on its SIRET number. Used the API BAN Base Adresse Nationale to retrieve location data. (data.gouv.fr)
Empower teams without exposing the database
Context — As the AI landscape database grew more complex, the Hub’s non-technical team needed to interact with the data and workflows without directly modifying the underlying Airtable structure.
Need — Create a simpler and safer way for non-technical users to perform day-to-day operations, while protecting the integrity of the underlying database and workflows.
Selected initiatives — Designed and built purpose-driven Airtable interfaces tailored to specific business processes and user needs.
Abstracted the underlying database complexity behind simplified operational interfaces.
Reduced the need for non-technical users to interact directly with the database structure, minimizing the risk of accidental changes and data inconsistencies.
Automating and Scaling Sales Prospecting
Context — A sales pipeline fed by around 20 events / year (1000 to 1200 leads), but never consolidated and rareley acted upon.
Need — Qualify this pool of contacts and modernize a sales process that remained largely manual.
Selected initiatives — Designed a medallion architecture on n8n to ingest raw contacts from various sources (webinars, whitepapers), process them, and feed them back into Salesforce as qualified leads.
Automated sourcing via PhantomBuster from scheduled LinkedIn webinars.
Personalized follow-up messages using generative AI (Gemini, on the client's Google Workspace suite).
Ongoing pilot of a headless CRM operated from Slack, powered by Salesforce Agentforce.
Enhancing Application Assessment with AI
Context — Alongside the digitized application process (Airtable/Softr), each submission is reviewed manually by the jury.
Need — Provide a second, independent and systematic assessment to complement human evaluation.
Selected initiatives — Designed expert agents to score each application with a confidence rating, for comparison against the jury's own evaluations.
Digitizing the Selection & Evaluation Process
Context — As part of the annual update of the French AI startup and provider landscape, the Hub wanted to streamline and secure its selection process.
Need — Replace the distribution of Excel files to jury members — difficult to consolidate and keep up to date — with a unified and secure interface providing real-time access to the underlying data.
Selected initiatives — Designed and implemented an authenticated interface connected to the AI landscape database, enabling selected jury members to evaluate assigned groups of startups through a dedicated interface, with assignments managed from the back office.
Digitizing an Application & Selection Process
Context — Design and digitize a call for applications targeting vendor / HR team pairs, with a two-stage application process: an initial pre-application phase to formalize the project and validate the partnership, followed by a full application including functional and technical documentation.
A dedicated evaluation interface was also developed for the jury.
Need — Avoid excel manipulation, give access to jury members out of the workgroup
Selected initiatives — Designed and implemented the end-to-end application and selection process.
Built a dedicated jury evaluation interface using Softr.
Processed 60 applications throughout the selection process.
Enabled 12 independent jury members, representative of the AI & HR ecosystem, to participate in the evaluation.
Structured the selection into two stages: in-depth review of applications by a focused expert jury, followed by finalist hearings with an expanded jury.
Automating Repetitive Accounting Processes
Context — Repetitive accounting tasks (expense reports, tax provisions) are often still handled manually by freelancers and small structures.
Need — Make these tasks reliable and faster without adopting a complex accounting system.
Selected initiatives — Automated expense report processing: email receipt, data extraction and conversion from PDF, submission to the accounting tool (e.g. Tiime).
Automated VAT provisioning: upon receiving the accountant's declaration by email, the amount is identified and a transfer to a dedicated account is generated via a secure banking API (e.g. Qonto) — with human validation before execution (human-in-the-loop).
Automating Email Triage & Lead Qualification
Context — A shared inbox for the working group receives a heterogeneous stream of requests.
Need — Automatically qualify these emails and only alert the group's co-chairs when action is genuinely required.
Selected initiatives — Chained two LLMs (Mistral): a first pass for analysis, a second for qualification, followed by an "action needed" summary sent to the co-chairs when relevant.
Automating the client Lifecycle in Salesforce
Context — Cross-functional engagement at the intersection of Business Development, Customer Success and CRM operations, in a Salesforce environment that had until then been underused.
Need — Make client lifecycle tracking reliable, down to the individual user level, before considering any finer-grained retention management.
Selected initiatives — Built an automated workflow for individual membership renewals, including three successive reminders and automatic activation of member portal access upon payment confirmation.
Automated member lifecycle processes in Salesforce Flow: corporate membership termination, individual member onboarding/offboarding, and other administrative workflows.
Turning a One-Off Database into a Scalable Business Tool
Context — The Hub’s French AI startup and provider landscape is built on an Airtable-based database, initially designed by an external provider for a one-off call for applications.
Need — Evolve the existing setup into a sustainable operational backbone capable of supporting recurring annual calls for applications, while improving data quality and streamlining data management workflows.
Selected initiatives — Refined the Airtable data model and fields to support recurring annual campaigns.
Cleaned and standardized the existing dataset to improve data quality and consistency.
Implemented Airtable workflows to automate recurring operational tasks.
Used n8n for large-scale data extraction, transformation, cleaning, and reinjection.