Biography:
Romain Serres is the co-founder of Tilt, a French energy startup. Tilt’s software helps electricity utilities (who sell/distribute electricity) predict demand and run more efficient, cleaner operations. On the other hand, it helps electricity consumers predict their consumption and lower their bill. Tilt just raised a $5M seed round.
Prior to Tilt, Romain worked as a data scientist at French carpooling startup Blablacar and co-founded Kiro, a healthtech startup. He holds a master’s degree from Arts & Métiers (in France) and Columbia (in the USA).
His master's thesis analyzed how the state of New York used machine learning in its electricity grid operations.
Your previous startup leveraged AI in the health space. Tilt does so in the electricity space. What differences do you see between building AI for health versus for electricity?
The type of data you’re working with is different. Health data is text-heavy, whereas electricity consumption is numbers-heavy. That changes how you build your AI. With electricity, you want your AI to predict what the next number (KWh or MWh) of consumption will be, depending on the variables it’s ingesting (ex: weather forecasts).
Health and electricity are similar in that AI adoption has lagged on both, because the risk associated with mistakes is high. If Google’s AI hallucinates when summarizing your Google search, no one gets hurt. If an AI hallucinates when summarizing your blood tests’ results, that’s a serious problem. It’s the same for electricity: if your grid breaks down because your AI messed it up, pandemonium ensues.
In both sectors, AI founders have to explain how many more mistakes humans make compared to AI, rather than justifying the AI’s mistakes. It’s like self-driving cars. Even if we statistically prove they are safer than human drivers, the lack of agency associated with self-driving cars crashing renders the leap scary.
What problem is Tilt solving?
A country has energy needs (for industry, mobility, appliances…) for which it taps multiple energy sources. These sources include fossil fuels (coal, gas, oil), as well as nuclear, renewables…
Within that energy consumption, France has seen a rise in electricity use, via consumer-led switches such as petrol-fueled to electric vehicles, or gas boilers to heat pumps. In 2023, electricity made up around 27% of France’s energy consumption. This is important in the fight against climate change, because the vast majority of electricity is produced by low-carbon energy sources (in France, that’s majority nuclear, followed by renewables).
Electricity use and the volume of renewable energy within that electricity production are growing concurrently. This means that the electricity grid needs to adapt to renewables, which are intermittent energy sources. A solar panel doesn’t produce electricity when it isn’t sunny. You can’t control that (hence the critical role storage of renewable energies plays). It’s easier to control how much electricity your coal generator produces, by simply adapting how much coal you feed it.
Our software helps electricity utilities (who produce/sell electricity) and consumers better predict electricity demand, orchestrate electric assets, and monetize their flexibility (a concept we’ll explain later).
By doing so, Tilt makes the energy transition profitable (savings and even revenue from electrified assets), reduces the instances where the grid has to tap fossil fuels to palliate for lack of available renewables, and participates in balancing the grid.
What did your MVP look like?
We were commissioned by Total, a French energy giant, to study how they could predict their clients’ electricity demand. We also went through their startup accelerator.
The fact that such a company acknowledged that this was a problem became sufficient proof of concept to raise our first investments.
How would you summarize Tilt’s product?
We have three main functionalities. Our clients use one or multiple depending on their needs.
The first is an algorithm that predicts electricity demand. This helps utilities optimize their operations, buy the right amount of electricity, and offer more competitive prices.
The second enables clients to pilot their electric assets (heat pumps, electric vehicles, charging stations, batteries) from afar. This gives utilities more control on how to balance their portfolio (production/purchases to consumption/sales) and over their operations when working with distribution system operators (that manage specific, local parts of the grid).
Both functionalities enable the third one: the monetization of flexibility. This requires a bit of context.
France’s electric grid operator, RTE, remunerates electricity consumers that help it handle periods of high demand. Simply put: your company agrees to reduce electricity consumption when RTE is struggling, and RTE pays your company for your help. Our third product helps our clients plug into those RTE schemes. The technical term for this is “demand response”.
Our first two functionalities are complementary to our third. By anticipating your own electricity needs, you can more effectively work with RTE and thus make more money.
That third product implies connecting with RTE, a state-owned giant. How heavy is that process?
We’re in a special category, as we help companies integrate with RTE schemes, but we don’t actually produce any electricity. We’re about 15 companies in the same basket, half of them subsidiaries of big energy companies.
RTE and us (a small startup) have different definitions of timeline and urgency. This creates some friction. Then, there’s the sheer technical complexity of the integration. We’re not talking about a simple API connection here. We’re talking about connecting to the network handling the entirety of France’s grid. Orchestrating that integration requires time and resources.
RO insights: similar schemes in Mexico
Other grids around the world monetarily incentivize their end consumers (companies and consumers) to help reduce pressure on it.
In Mexico, the national grid operator (CFE) remunerates actors who sell back surplus electricity to it. Niko, a startup that sells solar panels, helps its client plug into the scheme.
Here’s how Raffaele Sertorio, Niko’s co-founder, explains:
“One of the administrative complexities of switching to solar is dealing with the CFE. Generally, houses are connected to CFE’s grid. At the end of the month, CFE calculates how much you contributed to the grid versus how much you consumed. You are charged accordingly.
The difference with solar is that you now have a house that produces electricity and can contribute to the grid, so the equation flips. Your house’s CFE meter needs to become bidirectional to account for that change. In many cases, your house’s meter is out of date and needs to be replaced.
This entire process represents a 3-6 week administrative process (including the potential installation of the new meter), which we take care of for our clients.
[...]
Lastly, there’s political risk. The CFE, which is government-owned, is in a monopolistic position. They can do what they want. If they drastically reduce the amount they pay for the solar energy our clients pump back into the grid, that erodes our value proposition. I’m not too worried because Mexico needs more energy, CFE knows that and it isn’t in their interest to curtail the growth of solar. The new government has been saying quite the opposite.
That being said, CFE represents a lot of jobs (so it’s a vote machine) and the effective deployment of solar could theoretically reduce CFE’s headcount. That’s something we’ll have to aptly navigate (although we’re far from the scale for that to be a problem).”
Excerpt from Niko: scaling solar panels in Mexico, originally published in The Realistic Optimist
Who are your main client personas?
There are two main ones.
The first are companies consuming electricity and/or managing assets (ie: a factory, a building, a network of EV charging stations…).
The second are companies selling and distributing electricity (utilities).
How do you make money?
SaaS fees on our orchestration software, and a cut on the money we help clients make via the RTE schemes.
What’s a contrarian decision you took which you stand firm on?
In the early days, multiple people told us to verticalize. For example, build Tilt but only for a real estate use case. We probably would’ve raised faster that way.
We made the decision to go broad from day 1 and leave the door open for inbound opportunities. Our thinking was the following: the underlying tech is the same, regardless of the asset we’re working with. We didn’t want to close any doors.
That approach has started to bear its fruits, as we’re receiving an increasing number of inbound demands from different verticals (ex: geothermal heat pumps).
How have you structured your sales team?
We focus on a well-defined, narrow client persona for a couple of months at a time. Our sales teams will only speak to those types of clients over that period. This helps us go deep and understand if this persona is worth chasing. If it is, we have abundant data on how that persona thinks, works…
That’s one of my biggest takeaways so far. You need a different sales messaging and strategy for each client persona. You need to know what makes them tick and how to navigate each organization internally.
What challenges do you face when selling?
First, we’re interacting with hyper-operational people. They’re handling real-time electricity consumption. They’re spending a lot of time fighting fires, fixing things… It’s hard to sit them down and take time out of their day for a macro topic like ours. They also aren’t people that are used to being sold to, so it’s tough to get them to open up about their challenges.
Second, we carry the burden of shifting mentalities. In this sector, electricity consumption has traditionally been something you endure and figure out as you go. The idea that you can regain control over that process seems impossible to many.
Lastly, we integrate with our clients’ most precious asset (their electricity). They simply cannot take any risks on it. This creates some initial resistance.
Would you ever move into making hardware?
No. It’s a completely different craft. If a client asks us for hardware, we’ll refer them to other companies that specialize in that.
Do you have competitors?
On a European-level yes, such as Flower in Sweden and Axle in the UK.
They’re good competitors to have, because the market for what we’re doing is so new and so large that we can only learn from them.
We’re not at a stage where we lose clients to a competitor. We’re at a stage where we lose clients because they don’t fully understand what we do yet.
What’s your thorniest operational challenge?
Finding qualified talent.
Generative AI startups have raised a lot and siphon off great AI talent (they can offer higher salaries). Electricity may seem an esoteric sector to some. In a way, that acts as a filter because we only attract talent that’s passionate about the subject matter more than AI itself.
We’ve had a really hard time finding women talent in this space, which creates a gender-imbalance in the team. That’s something I want to fix.
RO insights: startups' structural gender-imbalance
Worldwide, the startup sector suffers from scandalously low female participation. This is due, in part, to the sector being heavily-networked based, which perpetuates pre-existing male networks.
Ukraine’s startup scene has been an interesting case recently. With many men at war, women took a more preponderant place. That doesn’t mean the structural problems disappeared though.
Here’s how Elena Mazhuha, a VC at Flyer One Ventures, explains:
“Less men in the ecosystem has led to more women in higher positions, both as founders and executives. As the law bans men from leaving the country, the ecosystem’s foreign representation (at conferences, etc…) has mostly been led by women.
I wouldn’t say it’s changed much on the VC side, though. Sifted reported on the exodus of women from VC, and the same reasons apply to Ukraine as well. The sector has been muted for a couple of years, as M&As and IPOs remain scarce. It’s hard to have a baby while traveling to conferences half of the year. Even harder if there’s no significant financial reward at the end of the tunnel.
VC sometimes still suffers from sexism. For women in VC, a career at Google looks more and more attractive. In VC, the glass ceiling seems thicker.
How do we solve this? While unpopular, I think quotas aren’t a bad idea. I don’t like them, but we need to force the sector’s hand. We need more women in middle-leadership positions: these are often filled by partners’ networks which, invariably, are overwhelmingly male.”
Excerpt from Ukraine’s startup ecosystem today, originally published in The Realistic Optimist
What’s the biggest structural risk to Tilt?
We operate in a highly-regulated industry. This is exciting. The more an industry is regulated, the higher impact it has on a country. This means it’s an industry worth improving.
Since it’s regulated, we’re at the whims of politics. For example, the price RTE pays out for demand response schemes fell due to a governmental decision. Since we take a cut of those payouts, it hits our revenues.
This is another reason we’re sector-agnostic. It’s practical to hopscotch between different client personas, as one might lose in pertinence following a regulatory decision. Being attached to only one would be dangerous.
We’ve had to develop regulatory savviness early. As a startup, you can be extremely early in your space and bulldozer your way to create new regulation (like Uber did). But we’re the opposite: we’ve had to take regulation very seriously, very early because we simply cannot operate outside of it.
RO insights: governmental decisions boosting one's startup
In some cases, a governmental decision can have an unintended positive effect on one’s startup.
Here’s how Dami Olawoye, CEO of Nigerian solar financing startup Rivy, explains:
“With regards to our business specifically, many fail to grasp the paradigm shift of the recent fuel subsidy removal.
A bit of context: for years, Nigerians have been paying an artificially low-rate for fuel, as the government heavily subsidized it. In 2023, the government decided that those subsidies were financially unsustainable and cut them. The price of fuel skyrocketed.
For people and businesses running on fuel-powered generators, solar became appealing almost instantly. This is the tailwind we’re riding and it’s a powerful one. People are switching to solar out of necessity.”
Excerpt from Rivy: solar financing in Nigeria, originally published in The Realistic Optimist
The EU touts itself as a single-market. How does this play into Tilt’s expansion plans?
Once you’re in with RTE, you are connected to the wider European market such as EPEX Spot (short-term electricity trading) and PICASSO (tasked with balancing the wider EU electricity grid). In the future, we’d like to help our clients to plug into EU-wide demand response schemes, similar to the ones RTE offers nationally.
That being said, an expansion to countries outside of France would require a new integration with the target country’s grid operator (ie: the local RTE). Energy laws aren’t the same in each country, either.
But we already can work with utilities outside of France, selling our consumption forecasts algorithms and our asset orchestration capabilities in order to optimize positions on the energy markets. We do so in the Netherlands and in Switzerland.
What’s your take on the renewable energy transition? Are we going fast enough?
France has the particularity of being a world leader in nuclear, with 38.6% of its energy derived from it.
While it’s a powerful energy source and doesn’t generate CO2, nuclear energy is embellished to the end consumer. Our nuclear industry is dependent on uranium, which we import. That is a risk. Recently, a military coup in Niger disrupted the activities of Orano, a French mining giant with uranium contracts in the country. Praising nuclear energy as the solution to attain “energy independence” is nonsensical.
Nuclear isn’t as clean as it’s presented to be either. The mining of the uranium itself has an environmental impact (conveniently, the mining happens outside of France). The dismantling of nuclear waste is something we haven’t quite figured out yet. While it’s overwhelmingly safe, the damage should something go wrong (ie: Fukushima) is terrible. The more nuclear sites we have, the more that risk statistically increases.
I’m a fan of solar panels. They don’t disrupt people’s living environment, as we can just put some on everybody’s roofs. I’m also a fan of hydro power (powered by dams), which is France’s primary source of renewable energy.
You studied at Columbia, in the US. Why come back to France to start your company?
It’s non-trivial to start a company in a country that isn’t yours: there are steep cultural differences to overcome. Just the fact that paperwork isn’t in your native tongue matters. I felt more comfortable starting a company in France, especially given the French startup ecosystem has boomed over the past decade.
We have fantastic talent, and more funding than before. There’s also a tinge of European patriotism here: I want my continent to grow its own champions and rival our American friends.
Coincidentally, I also wasn’t in the startup mood during my studies, so I wouldn’t have started a company then.
The Realistic Optimist’s work is provided for informational purposes only and should not be construed as legal, business, investment, or tax advice.