The start-up has created an AI-assisted training platform natively integrated into employees’ digital workplace.
What if it was possible to train employees directly from Slack, Teams or WhatsApp? This is the promise of Blify, a young start-up founded in 2025. Based on the observation that commitment to non-mandatory professional corporate training peaks at between 10 and 15%, the start-up has imagined a training platform capable of supporting employees within their daily tools.
To expand its offering and structure its technical team, Blify announced this Tuesday, March 17, a pre-seed fundraising of 1.8 million euros (including a portion of undisclosed debt). The funding round is led by AFI Ventures, with the participation of Kima Ventures (Xavier Niel), Better Angle, NewSchool, Supercapital, Fair Equity and more than fifty business angels.
A platform designed like a learning OS
Blify presents itself as a Learning Operating System capable of creating, distributing and managing internal training, all co-piloted by a multi-agent AI infrastructure. “The tool is designed both for educational engineers and for non-technical profiles, a marketing director or a manager who wants to quickly train his team on a new offer or a new regulation,” explains Clément Lhommeau, co-CEO and co-founder of Blify. In practice, the user imports a document, a PDF, a product brochure, a web page, specifies their objective, and the AI agent generates a structured training plan. In just a few clicks, training is pushed to the employees concerned.
The particularity of Blify lies in its method of distribution. Rather than adding yet another platform to employees’ daily lives, the start-up injects training directly into Slack, Microsoft Teams or WhatsApp. “We don’t try to make people come to us, we go where they are,” summarizes Clément Lhommeau. And the tool doesn’t just answer questions: it proactively pushes content in various formats: flashcards, podcasts generated on the fly, videos or vocal role-playing games allowing you to practice facing scenarios. For sales teams, Blify goes so far as to integrate into the CRM via browser extensions, positioning contextual micro-training directly in the business interface.
An AI agent composed of sub-agents
Under the hood, Blify is based on a multi-agent architecture. Each function of the platform (creation of training, daily exchange with the employee, etc.) mobilizes one or more specialized agents who coordinate between them. “We don’t always use the same models. For some agents, it will be GPT-4o or GPT-4.1, for others a Google or Mistral model,” explains Clément Lhommeau. The choice is made using an MLOps approach: the team tests several models on each task, compares the results on a dataset, and selects the one that offers the best performance. The main technical challenge is to ensure that the chain of agents produces the best overall result.
On the business model side, Blify has opted for a per-user license. “We did not want to move on to variables, as we see a lot in AI. We preferred to remain simple for our customers: they know what they are paying,” explains Clément Lhommeau, without excluding an evolution of the model in the long term. The start-up primarily targets mid-sized companies and SMEs, but has shown itself to be opportunistic with major accounts: Club Med and Hermès are already among the first clients.
The funds raised in recent weeks will be directed almost entirely towards the product. Concretely, the start-up plans to recruit software engineers and product designers to move from a first use case focused on managers to a complete solution for creating, distributing and managing training at scale. The technical ambition is assumed: “We raised these funds to build barriers to entry and build the best multi-agent training system on the market.”
The start-up is part of the edtech wave boosted by generative AI, where fundraising announcements are increasing. In a market that quickly becomes cluttered, sorting will be inevitable.
