AI is a present-day engineering challenge. Across Europe, laboratories, companies and universities are racing to bring AI out of the cloud and onto the devices closest to us like the sensors at the edge of our factories, fields, roads and homes.
This is largely part of the mission of the EdgeAI project, uniting over 40 organisations to accelerate the development, integration and commercialisation of edge AI technologies across five key industrial sectors. As exploitation task leaders, our role at CONVERGENCE is to ensure that the project’s results don’t just stay in internal reports, but they reach the market.
The benefits are significant: faster response times, reduced energy consumption, lower data transmission costs, better privacy and resilience even when network connectivity is lost.
What makes EdgeAI distinctive is its breadth, the project has built real demonstrators across five value chains, each tackling specific challenges where edge AI can create tangible impact.
The often-overlooked challenge is ensuring that the innovations created in a research project find their way into real products, services and companies. This is the work of exploitation. The journey from idea to impact. Often the narrative behind innovation focuses on the lone inventor, the small startup, the founding team in a basement. The challenge is different but no less real when the team is actually a consortium of over 40 organisations, universities, research institutes, large companies, SMEs, spread across a dozen European countries, working in different industrial sectors, each with their own definition of success.
In EdgeAI, some partners measure success in patents filed. Others in papers published. Others in product roadmaps updated, in licensing deals negotiated, in PhD students who graduate with skills the market needs, in standards shaped. A large semiconductor company and a four-person spin-off based in Athens do not have the same exploitation horizon. A Norwegian energy research institute and a French agri-tech company are not talking to the same customers.
And yet they are all part of the same project. They have all contributed to a shared pool of results. And of course, the hard question, what happens to those results next?
When we describe what exploitation is the first impression is always different than the reality. The reality is both messier and more interesting than some people think.
It means sitting with a researcher who has been developing a novel AI algorithm and asking:
who needs this?
who actually has the problem this solves and would pay or change their behaviour to have it solved?
It means working with a company that has built something genuinely innovative and helping them articulate why it is valuable. It means asking about IP strategy before the patent window closes. It means noticing when two partners in different parts of the project have results that would be more powerful combined than separate and creating the conditions for that conversation to happen.
All this, this exploitation process/journey, is intensely human. It requires trust, which takes time to build across a consortium of dozens. It requires a tolerance for ambiguity, because most exploitation paths are not clear at the outset. And yes, it does require a certain stubbornness, a refusal to accept that because something is hard to commercialise, it should simply be left behind.
What EdgeAI is building is not just a set of edge AI technologies. It is also a set of exploitation practices, frameworks and habits of mind that can outlast the project. The questionnaires we designed, the repository we built, the conversations we have facilitated, these are reusable so that the next project, and the one after that, does not need to start from scratch.
As EdgeAI enters its final months, our job is to make sure the results do not stop at the project boundary. The exploitation workshop in Milan marked a turning point, partners came together, reviewed their results side by side, and made deliberate choices about what comes next. That conversation has now shifted into its final, most concrete phase: validating the data, locking in the descriptions, and assembling the full picture of what this consortium has built.
What remains is to turn that picture into a final exploitation deliverable, not just a record of what was done, but a roadmap for what comes after.
The lab is not the endpoint. And the story, finally, is close to being told.
Blog signed by: CONV Team