In this new episode of MuzzOsphère, Antoine Gigomas welcomes Éléonore Breton, Chief Operating Partner at Newfund, a seed-stage VC fund that backs more than 120 startups in France and the United States.
We turn to a figure still rare in venture capital: the Operating Partner with front-line experience. Neither a financier nor a consultant, Éléonore Breton structured international Customer Service operations at PayFit and grew the customer care team at Matera from 5 to 50 people, before joining Newfund to put that experience to work for founders.
Beyond the theory of startup support, this conversation explores concrete questions: what difference does it make for founders to have across the table an investor who has actually operated? How did Newfund build an operator-investor approach that gets involved in tools, processes and execution? And what mistakes do founders keep repeating in the scale phase, which a seasoned eye can spot months in advance?
the operator-investor approach applied to seed
Newfund invests at pre-seed and seed, with Franco-American DNA. As Éléonore Breton explains in the episode, the fund has around 120 portfolio companies (about fifty in the United States, seventy in France) and $400 million in assets under management, with a healthcare specialisation carried by the EKA fund. The house conviction: for many French startups, especially in health tech, most of the market lies across the Atlantic. So the fund supports its portfolio companies through that crossing, from practical matters (visas, the founder's relocation) to introductions and getting into accelerators. She cites StartX, which four portfolio companies have gone through.
What sets Newfund apart is how involved it gets after the investment. Éléonore Breton describes a deliberate choice by general partners Patrick Malka and François Véron: rather than a platform model that quickly steers founders towards outside providers, the fund bet on senior profiles who have operated in startups and scale-ups, able to see a mission through to completion.
We get our hands dirty, we do the work for the founder, we don't just make recommendations.
In practice, the operating team produces deliverables: scoping a product roadmap, configuring a CRM, setting up reporting dashboards. The work sometimes comes close to RevOps, which the team fully embraces whenever the impact justifies it.
tailoring support from pre-seed to scale-up
Newfund holds its portfolio companies for a long time (five to ten years depending on the case), which means it has to cover a large part of the value chain. With very early-stage startups (fewer than 10 people), Éléonore Breton works directly with the founders, often on several fronts at once. She gives the example of PixiScience, a startup that automates the marking of maths papers for university professors: defining the product roadmap, scoping the product manager role, then actually hiring the PM (found through her own network).
With more mature companies, the work narrows to one specific vertical. She is leading the modernisation of Camping Car Park's tools (an investment made more than ten years ago, 160 people): rolling out HubSpot for 70 people and setting up an AI voice agent for a customer service team that handles between 1,000 and 2,000 calls a day in high season. Between these two extremes, Newfund also offers “blitz” missions (two sessions of two hours with a founder on structuring a team or a pricing policy).
AI inside a fund, from internal tools to agent pricing
Newfund describes itself as a data-driven VC and has four people dedicated to tech; what the team rolls out at portfolio companies, it also applies to itself. Éléonore Breton cites Eureka, a web app built by Julien, a member of her team, with Claude Code in a month and a half. The tool identifies the relevant key opinion leaders and corporates within the scientific network (through jointly filed patents and co-authored publications). Designed first for the healthcare fund's investors, it has since been opened up to founders. Alongside it, Dust agents prepare meetings and deal reviews, saving hours of work.
The arrival of AI is also creating new requests. Two founders came to her almost at the same time with the same question: how do you price an AI product? Éléonore Breton attended dedicated talks (including one by Chargebee), then turned them into a method every portfolio company can reuse, distinguishing three models: flat subscription, usage-based and outcome-based (billing for the task the agent actually completes).
The only way you can challenge the model, realise its limits and get the best possible answer is to first sit down and think for yourself, before using an LLM.
That is the thinking behind her view on human value: she sometimes asks her team to handle certain tasks without generative AI, to preserve critical thinking. In her view, two skills remain hard to replace: methodologies (prioritisation, architecture best practices, testing, technical debt management) and the perspective needed to make trade-offs.
the recurring mistakes founders make in the scale phase
The first blind spot, according to Éléonore Breton, comes down to a hidden cost: lack of communication. As a company grows, top management inevitably drifts away from individual contributors, and the vision does not filter down well into operations. Her recommendation is direct: overcommunicate, through several channels, both in writing and verbally.
Second pitfall: hiring too fast. She stresses the managerial courage it takes to part ways with someone who is technically excellent but toxic for the rest of the team (this is where, with a wink, she slips in that Muzzo is precisely a way to avoid bad hires). Third point: structure early, without waiting to reach 100 people, especially since generative AI now makes it possible to lay solid foundations quickly; a writing culture is becoming essential, since a model performs better the more complete and well-curated the material it is given.
Finally, prioritisation. Many teams want to do everything and deliver several half-finished projects rather than one or two carried through to the end. She advocates OKRs to align teams and make trade-offs, and the RICE method (Reach, Impact, Confidence, Effort) to settle product decisions. The parallel with technical debt comes up again: piling it up to move fast can cost months of rework, or even force a complete rewrite.
the guest: Éléonore Breton
A business school graduate, Éléonore Breton started out as a category manager in mass retail, then at a statistical research firm working on customer satisfaction. She went on to spend two years at PayFit (2018-2020) as Customer Success Operations Manager, in charge of customer service operations across the four countries where the company operated, in the middle of hypergrowth. She then joined Matera for four years: first as Head of Customer Care (joining as the fiftieth employee, she grew the team from 5 to 50 people while the company went from 50 to 200), then, after a career change, as a software engineer for two and a half years. She is now Chief Operating Partner at Newfund.
key takeaways
- Newfund's operator-investor approach relies on people who have operated and go all the way to execution, not on hands-off advice.
- The same fund supports teams of fewer than 10 people and companies of 160, with missions of a very different nature.
- AI is changing the day-to-day of the job: internal tools built in a few weeks, operational agents, and new questions such as pricing AI products.
- The costliest mistakes when scaling are silent ones: under-communication, hiring under pressure and lack of prioritisation.
- Human experience keeps its value where AI executes: methodology, critical thinking and judgement calls.
