This season on Build Mode, we’re diving into what it takes to build a world-class founding team. This week we’re exploring different kinds of co-founder dynamics and what it takes to build a startup with your family. Building with a family member or spouse comes with major benefits like built-in trust and an “always-on” mentality, but it can also create challenges when all the startup risk is coming from one household. 

Build Mode’s Isabelle Johannessen sat down with Hala Jalwan and Alessio Tresanti, spouses and co-founders of Rivio, an AI procurement startup. Jalwan and Tresanti both believe in going all in on their ideas and loved building things together, from community events to cross-country road trips. They found they naturally took simple ideas and blew them up to their most epic potential. So when they got the idea for Rivio, they were confident that they would both be able and willing to commit fully. 

As Rivio has grown, they have two main takeaways: First, co-founders should have clearly defined lanes. Second, it’s a good idea to bring in a third co-founder as a tie-breaker. 

Rivio’s third co-founder and CTO is Leo Larrere. “It’s great because honestly it fits perfectly into this relationship,” Tresanti said about Larrere. “It’s obviously a three-co-founder relationship. He’s also the one that brings sanity to the conversation and can draw the line sometimes.”

In the second half of the episode, Johannessen talks with Anna Sun, the co-founder of Nowadays, an AI co-pilot for corporate event planning that she launched with her sister Amy shortly after graduating from MIT. Sun spoke about how the two built their team out of friends and former co-workers, and created a culture that’s based in community. There’s a built-in trust, not only between the sisters-turned-co-founders but also throughout the team as a whole. 

“Because we’re sisters, we trust each other so much that I remember even previously, when I would start ideas with friends, you always feel like, ‘Oh, I don’t want to step on the other person’s toes,’ or ‘I don’t know if this feedback is too direct,’” Sun said. “But because we grew up in the same household, we have a lot of the same values, and we’re very direct to each other. We don’t want to waste time.”

These conversations shed light on how founders can build a truly effective and happy team as long as there’s a foundation of trust, clearly defined ownership, and a willingness to navigate conflict respectfully. 

Apply to Startup Battlefield: We are looking for early-stage companies that have an MVP. So nominate a founder (or yourself). Be sure to say you heard about Startup Battlefield from the Build Mode podcast. Apply here.  

TechCrunch Disrupt: We’re back for TechCrunch Disrupt on October 13 to 15 in San Francisco, where the Startup Battlefield 200 takes the stage. So if you want to cheer them on, or just network with thousands of founders, VCs, and tech enthusiasts, then grab your tickets.

Isabelle Johannessen is our host. Build Mode is produced and edited by Maggie Nye. Audience Development is led by Morgan Little. And a special thanks to the Foundry and Cheddar video teams.

Electric vehicle drivers and smartphone power users have been salivating over the prospect of silicon anode batteries, which promise to dramatically boost energy density and lower charging times.

Several companies have been working on silicon anodes over the last decade or so, and the tech has started creeping into consumer electronics. Wearable maker Whoop, for example, uses materials from Sila, while Group14‘s batteries can be found in a range of smartphones.

But the real prize is the EV market, which dwarfs consumer electronics by an order of magnitude, according to Benchmark Minerals. To break into this space, however, startups need to produce silicon anode material in far larger quantities than they have been so far.

To hit that scale, Group14 on Thursday said it had started production at its BAM-3 factory in South Korea. The facility is capable of producing up to 2,000 metric tons of silicon battery materials annually, enough for 10 gigawatt-hours of energy storage, or about 100,000 long-range EVs.

“It’s a big deal for us, and I think it’s a big deal for the industry, too,” Rick Luebbe, co-founder and CEO of Group14, told TechCrunch.

The BAM-3 facility broke ground as a joint venture between Group14 and SK, the Korean battery manufacturer. SK owned 75% of the project, but sold its stake to Group14 last summer. 

“SK has had their own challenges — financial and reprioritizing their battery and battery materials strategies all at the same time,” Luebbe said. “It did open up a great opportunity for us to acquire it from SK.”

The startup has been working with a number of companies, including Porsche’s battery division Cellforce Group, StoreDot, Molicel, and Sionic. Porsche has also invested in Group14 through its venture arm.

Most modern batteries use carbon as the anode material. It works well enough, but scientists have long known that silicon, which can store up to 10 times more lithium ions, would be better for energy storage if they could only solve some vexing durability problems: Pure silicon anodes are prone to swelling and crumbling in short order, making them unsuitable for repeated charging cycles over several years.

Group14’s answer is a hard carbon scaffold that holds minuscule silicon particles in place, preventing the anode from swelling or crumbling. That scaffold is shot through with nanoscale holes that allow the lithium ions and electrons to pass through. It also helps the anode charge quickly without breaking down. 

Some of Group14’s customers, like Sionic, are using silicon anodes to boost energy density by up to 50%. Others, like Molicel, are focused on utilizing silicon’s fast-charging capabilities, including a design that can take a battery from flat to full in just 90 seconds. 

That sort of application of silicon anodes could upend the EV market. Chinese EV maker BYD is already aiming to build that sort of capability: It last week revealed a new battery pack that can “flash” charge from 10% to 70% in five minutes. (Luebbe is convinced BYD is using silicon-carbon in its new battery. “It has to be,” he said.)

If charging networks can accommodate such an EV, range anxiety would be a thing of the past. Today, automakers have been striving to deliver 300 miles to 400 miles of range mostly to alleviate consumer concerns, but hitting those numbers requires large batteries that add bulk, heft, and cost. Flash charging that can deliver meaningful range in seconds could allow carmakers to slim down battery packs, saving cost and weight.

“I’ve got a Rivian with a 130 kilowatt-hour battery in it, which is ungodly expensive,” Luebbe said. But with flash charging, concepts like inductive charging at stoplights — which might seem outlandish today — start becoming more feasible, he said. “You’d never think about charging ever again.”

Flash floods are among the deadliest weather events in the world, killing more than 5,000 people each year. They’re also among the most difficult to predict. But Google thinks it has cracked that problem in an unlikely way — by reading the news.

While humans have assembled a lot of weather data, flash floods are too short-lived and localized to be measured comprehensively, the way the temperature or even river flows are monitored over time. That data gap means that deep learning models, which are increasingly capable of forecasting the weather, aren’t able to predict flash floods.

To solve that problem, Google researchers used Gemini — Google’s large language model — to sort through 5 million news articles from around the world, isolating reports of 2.6 million different floods, and turning those reports into a geo-tagged time series dubbed “Groundsource.” It’s the first time that the company has used language models for this kind of work, according to Gila Loike, a Google Research product manager. The research and dataset was shared publicly Thursday morning.

With Groundsource as a real-world baseline, the researchers trained a model built on a Long Short-Term Memory (LSTM) neural network to ingest global weather forecasts and generate the probability of flash floods in a given area.

Google’s flash flood forecasting model is now highlighting risks for urban areas in 150 countries on the company’s Flood Hub platform, and sharing its data with emergency response agencies around the world. António José Beleza, an emergency response official at the Southern African Development Community who trialed the forecasting model with Google, said it helped his organization respond to floods more quickly.

There are still limitations to the model. For one, it is fairly low resolution, identifying risk across 20-square-kilometer areas. And it is not as precise as the U.S. National Weather Service’s flood alert system, in part because Google’s model doesn’t incorporate local radar data, which enables real-time tracking of precipitation.

Part of the point, though, is that the project was designed to work in places where local governments can’t afford to invest in expensive weather-sensing infrastructure or don’t have extensive records of meteorological data.

“Because we’re aggregating millions of reports, the Groundsource dataset actually helps rebalance the map,” Juliet Rothenberg, a program manager on Google’s Resilience team, told reporters this week. “It enables us to extrapolate to other regions where there isn’t as much information.”

Rothenberg said the team hopes that using LLMs to develop quantitative datasets from written, qualitative sources could be applied to efforts to building datasets about other ephemeral-but-important-to-forecast phenomena, like heat waves and mud slides.

Marshall Moutenot, the CEO of Upstream Tech, a company that uses similar deep learning models to forecast river flows for customers like hydropower companies, said Google’s contribution is part of a growing effort to assemble data for deep learning-based weather forecasting models. Moutenot co-founded dynamical.org, a group curating a collection of machine learning-ready weather data for researchers and startups.

“Data scarcity is one of the most difficult challenges in geophysics,” Moutenot said. “Simultaneously, there’s too much Earth data, and then when you want to evaluate against truth, there’s not enough. This was a really creative approach to get that data.”