A Few Things We're Focusing on in 2025

2024 was a wild ride! We closed and launched the Fund in October, got ourselves a new office, and welcomed some new team members. We're taking the same energy into 2025 and are super pumped to dive in. Here are a few things we’re focusing on as we start the new year:

🧑💻 Developer platforms that aren’t just for developers.  The first generation of developer platforms like Twilio, Stripe, and Github weren’t built with knowledge workers in mind. Yet, these knowledge workers are increasingly finding themselves in the technical hotseat, leaving a void in the market for new tooling and platforms that don’t require the user to have deep technical expertise. We see a clear opportunity for startups that are leveraging the modern AI stack to reimagine products for new user personas across code development, code testing, DevOps, SRE, and more.

🤖 Voice agents transform labor in vertical markets. We’ve all endured the agony of a customer support call - 37-minute hold times only to be routed through multiple agents and end the call with no clear resolution. 2025 is poised to be the year of real-time voice applications, reshaping this experience entirely. Why now? With multimodal LLMs like GPT-4o enabling real-time reasoning across audio, vision, and text, the voice infrastructure layer has reached a "good enough" threshold to outperform current call center standards. The traditional voice technology pipeline is evolving into a more integrated speech-to-speech (STS) model. By eliminating the need to translate back and forth between audio and text, STS models will deliver significant improvements. We expect the initial applications to focus on high-value transactional calls, particularly in sales and customer support - across a range of industries, including healthcare, home services, restaurants, debt collection, hiring and staffing, financial services, and more.

💸 24/7 Money.  After a cold and dark winter for fintech entrepreneurs, 2025 is going to be a turning point. With late stage juggernauts like Chime, Revolut and Stripe barreling towards the public markets, fintech will be front and center in the headlines. While we expect innovation to come in many forms, the defining theme for this year will be Instant Payments. As banks, corporates and startups take advantage of the maturing payment rails and stablecoin infrastructure, the applications of ‘always on’ money will cascade through the ecosystem. From instant settlement in capital markets to same-day paychecks for workers, the new ways builders will utilize advanced payment rails will transform markets at a fast and furious pace. 

🧨 Advanced infrastructure in AI fuels an explosion of new applications.  In the post ChatGPT era, an unprecedented building frenzy occurred with startups cropping up to capitalize on what might be a once-in-a-lifetime “why now” moment. But waves of them failed just as quickly as new frontier models emerged, obviating or commoditizing products. The ground was shifting so fast that some VCs even claimed to give up on software investing because AGI would drop the marginal cost of software to zero. There are indications that AI capabilities are slowing down, due either to decaying scaling laws or simply bumping up against resource limitations on data, compute, and energy. This suggests it’s unlikely AGI doomsday will come soon, but it also means that now is the time for the application diffusion to explode as the contours of the model’s abilities solidify. We expect founders will find ways to take already magical LLMs and turn them into real value for society. AI is still changing the world, but in 2025, it will be less hyped and more grounded.

📺 New revenue models emerge for content. The relationship between AI businesses and IP holders is contentious (to put it lightly). Content creators feel as if they are digging their own grave by producing data that feeds into models that may replace them. Traditional revenue models based on advertising are breaking as LLMs open up a new modality of user engagement. This evolution is spawning new marketplaces between stakeholders as well as new forms of sponsored content in text or visual AI outputs. Creators need to be compensated or they will either refuse or be unable to produce more data, which is increasingly in short supply as AI models saturate the existing data pools.

🏰 Data moats reign supreme. With the rise of LLMs, models have an insatiable appetite for data, but we’re running out of net new data sources. Many startups are getting a new lease on life by pivoting to alternative revenue streams—they’re realizing that all the data they’ve been inadvertently capturing over the years can be used to fine tune proprietary models or be sold to foundation model trainers. Businesses that find innovative ways to gather data are going to build defensibility. There are even people screen recording and keylogging their computers all day in order to collect data to train their own personal AGI one day. More and more of the latent information in the world will be captured and systemized because it is this very information that will prove to be among the most valuable data assets.

📱Consumer investing returns from the dead. The death of the breakout consumer app has been well documented. Google, Apple, Meta, and Amazon have historically controlled all of the distribution chokepoints and past breakouts required luck, timing, and approximately $2B from a rich corporate parent (read: TikTok)! But, the times, they are a changing. Antitrust issues have ensnared all of the major Internet platforms so they have to be more permissive in democratizing distribution. AI enables business-level utility for individual users without a heavy services component. We expect a renaissance of consumer startups (well beyond social) resulting in a new era of consumer apps in a range of categories — from entertainment to productivity to personal finance and beyond.

If you are working on any of the above, we'd love to hear from you.