Excited to announce we have closed $20m for @frachtisvc Fund I, a crypto-native pre-seed fund supporting builders across decentralised infrastructure, AI, middleware, and consumer apps.

BREAKING: Excited to announce we have made our first investor hire at @frachtisvc - @apriori0x @apriori0x is eloquent, crypto-native, sharp, creative and an original thinker. His discipline in evaluating deals, care for venture as a craft, structured first-principles thinking, passion for research and operator experience make him a perfect fit for our investor team. Welcome to the team @apriori0x 🙂

An underrated opportunity right now is figuring out how to measure subjective agent work. Crypto has historically relied on objective mechanisms such as: - Proof-of-Work to prove computational work was expended - Validity rules to determine whether transactions and state transitions follow the protocol - Slashing to penalise provable misbehaviour like double-signing None of these mechanisms measure whether a buyer received what they intended to purchase. We have some infrastructure for resolving bespoke disputes, but we do not have sufficient infrastructure for determining whether a scoped job completed by an agent actually met a buyer's expectations at scale. Existing approaches solve adjacent problems such as: @Kleros_io and @UMAprotocol - can adjudicate disputes but rely on human participation and is designed for infrequent disputes @bittensor - evaluates miner performance within subnets, not whether a buyer's intent was executed @virtuals_io - has evaluator agents, but evaluators are selected per task and there are no economic guarantees tied to the outcomes (yet) A missing primitive that is yet to be built is a way for two agents that do not trust each other to scope a job, agree on conditions for success, execute the work, evaluate the outcome and release payment at scale. Lots of open questions around how to build this, such as how to: - Resolve disputes at machine speed - Keep sensitive information private during negotiation and evaluation - Prevent collusion between agents using the same model providers - Evaluate outcomes where quality is subjective - Determine who bears cost when agent fails In general, I think liability will be important here. If agents have capital and/or reputation at stake, they can be penalised when they fail to deliver against clearly specified commitments. I do think pricing the risk is hard though. In saying that, if a protocol can accurately price the probability that an agent delivers a specified outcome and that pricing improves with every buyer–seller interaction, it will become more valuable over time. The data could be used to build a kind of matchmaking moat, e.g. understanding which agents reliably deliver which outcomes for which buyers. It's an early space but I expect mechanisms for evaluating subjective agent work will be critically needed soon. As agents take on more economically valuable work and expect to be paid for it, we will need a way for two agents that don't trust each other to agree on when a job is done.

We’re at an incredibly interesting time where crypto technology can be applied to solve some of the increasingly evident problems in the AI space. Right now: ⁃ Security is worsening over time as AI capability improves ⁃ Trust in frontier labs is eroding over time as they spread fear to the masses about what their technology is capable of ⁃ Unwarranted surveillance is becoming an issue for society as companies look for any way to collect new data to feed intelligence Crypto was built for this moment. Crypto was created at a time when trust in institutions had eroded and people were suffering the consequences of poor decisions made by a small number of individuals who held too much power within those institutions. Crypto has always offered a fairer way to participate in the economy without restrictions imposed by centralised institutions. Similarly, I expect crypto tech will offer us a fairer way of owning and using our intelligence long term. This structural shift is early and just getting started. It is clear that the research we have been working on in crypto for the past decade is becoming more important than ever. Such as: ⁃ coordinating untrusted counterparties operating open source software using incentives ⁃ proving and verifying work ⁃ computing over private data ⁃ mechanisms for self-custody We’ve already had some amazing teams come from crypto-native backgrounds and make an impact in AI, such as @NousResearch and @PrimeIntellect. I believe this is only the beginning. There is still a huge opportunity for crypto entrepreneurs to tackle some of AI’s hardest problems.

Excellent article from one of my favourite teams in crypto @Decentralisedco @shutterbugsid @joeljohn The article presents a true lay of the land on what early-stage crypto venture looks like right now. Was great to contribute our view at @frachtisvc on the growing market for agent-native trust and why crypto-native investors are primed to underwrite it. Recommended reading 📚

This experiment at @JoinEdgeCity was indeed very cool. Was awesome to use @indexnetwork_ to find relevant connections whilst attending the event. Found it particularly cool watching Index agents negotiate with each other for their principals. As a VC, I wanted to meet pre-seed startup founders at the event. Index knew my thesis exactly and connected me to founders I didn't know would be relevant to my thesis (e.g. neurotech founders). In other words, Index knew my intent better than I did. Index is a completely novel, privacy-preserving, agent-driven discovery network. Very cool use-cases across investor / founder connections, employment, dating and beyond. The product is going live very soon. Worth the follow!

Have seen a number of high-quality pivots from our portfolio companies in the last few months. If something isn’t working, there has never been a better time to go back to the drawing board and build something else of value, especially given how easy it is now to build with AI. Bear markets force founders to push the limits when it comes to innovation and ambition. The best founders respond to the harshest conditions. If you’re building something right now in a market that is too small, or your product is undifferentiated and/or at high risk of commoditisation, I would use the rest of this year to figure out another uniquely valuable path for your business. Start by: ⁃ planning, researching and talking to customers ⁃ using your network to make first early sales ⁃ leveraging AI to increase velocity across the entire business ⁃ refining your story ⁃ being more efficient with your burn Do this and you are setup well to come out the other side and win.

We're still on the hunt for a full-time Investment Analyst to join our team @frachtisvc Looking to recruit someone based in either NYC or SF If you're a crypto-native, original thinker, who is connected in niche networks of talent within the crypto and AI space - apply below: frachtis.com/careers/

In the next 12 months, the crypto industry will see an incredible pool of talent go live with venture ideas. Founders who demonstrate commitment to the industry in very challenging times, think outside the box, dare to try new things and have a decade plus of experience in crypto, are naturally going to be resilient, long-term players (perfect for venture) as opposed to short-term players pouring in when times are good and everyone is making money. Crypto is not going anywhere. It’s actually the best time to be building with original ideas right now. If you can align your thesis with the right VC and raise enough funds to prove PMF to get to the next round, it’s a phenomenal time to be optimistic. You’ll also team up with people who are committed to the cause and in it for the right reasons. I have really enjoyed seeing who is still dedicated to the industry this year. Some incredibly talented researchers and engineers are still floating around experimenting with ambitious new ideas. This cycle will look nothing like previous cycles. We will have an entirely new economic actor transacting onchain (AI agents). There is a massive opportunity to rebuild the stack and applications on top from first principles. Don't miss the forest for the trees.

I often geek out to crypto podcasts and @thebellcurve was one of my favourites even before @MikeIppolito_ invited me to co-host with @MylesOneil It was an honour recording the weekly roundup shows over the last 6 months with Mike and Myles. @MikeIppolito_ 's founder experience at @Blockworks and @MylesOneil's operator experience at @deltadotnetwork allowed us to break down the intricacies of running and growing crypto businesses in this new crypto era. We unpacked a variety of topics over the months on the podcast - from the rise of social, to the resurgence of decentralised AI, to how crypto has changed over the years. The show has helped me immensely as an investor at @frachtisvc to form concrete views about where value will accrue in crypto long term. In this last episode, we dive into crypto’s 1 and 5 year outlook. Hope everyone enjoys! 🎧

Got some news: today's episode will be the final episode of Bell Curve. Can't tell you all how much fun this podcast has been to host, and have so enjoyed diving deep into the weeds over the years. To everyone who listened, thank you. To my co-hosts over the years @pythianism , @im_manderson , @MylesOneil and @0xave deeply appreciate everything. We are going so much higher, salute everyone who's still here and bullish at the bottom 🫡

Teams and ideas within early-stage crypto private markets are getting much better than they were 6 months ago. As AI capability improves and anyone can build anything, software itself is no longer a moat. Apps and infrastructure that only have a technical edge are commoditising fast. What's left standing now are apps and infrastructure that are built on things that AI can't replicate, such as relationships, trust, distribution, network effects and proprietary data on real workflows. Builders are much more astute to this now in crypto and have been working on much stronger / more durable moats from day one. I'm particularly bullish on builders creating and capturing value where incumbents structurally can't build, like credibly-neutral rails for agents. There is more value now than before in building something like credibly-neutral rails for agents, as agents are getting better at doing high-stakes work reliably. As a result, verifiability, security and evaluation of work is becoming something that people (and agents) are willing to pay for. Agent trust is becoming a scarce, priced thing for the first time. It's very early, but demand for this kind of infrastructure is now being tested. In this new world where intelligence is abundant and cheap and where incumbents have entered the playing field, the long-term winners in crypto will be builders who: - stick true to crypto values / primitives - understand how to build and capture value, not just create it - focus on long-term durable businesses The setup is finally here. Builders who are dedicated to the value proposition of crypto and focus on where value gets captured rather than just created, will be long-term winners in this age of abundant intelligence. Very optimistic on the quality of crypto teams now building at the earliest stage and how they think about long-term value. I expect these builders will move the crypto industry forward in no time.

Looking for someone absolutely cracked to join us full-time at @frachtisvc as an Investor / AI Agent Builder We are looking for someone who is: - obsessed with investing, deep research and writing - extremely skilled at building AI agents - able to form differentiated views and defend them This role sits at the intersection of venture judgment, crypto-native research and AI agent execution. We want someone who excels at all three. This is not about credentials. Send us a 2 - 5 min video of AI agents you've built. Show us proof of your work. Apply here: frachtis.com/careers/

Some of the largest crypto funds have raised huge amounts recently, e.g. @a16zcrypto ($2.2b), @HaunVentures ($1b), @dragonfly ($650m) and @variantfund ($222m) - where most of the theses are consensus and related to AI agents using crypto rails. The problem is, so far in 2026 we have hardly seen any valuable agent-driven crypto use-cases. We are at historic lows at pre-seed and seed stage in crypto venture right now. I think a lot of this comes down to the fact that most net-new potential innovation in crypto is at the crypto x AI intersection, but very little value has been created so far from early experimentation. Infra needs to be built, but to be successful, teams are now competing against incumbents with large distribution for the first time, making it harder than ever. Middleware is now too easy for other competitors and/or buyers to replicate. Applications are working out what settlement layer will win, what standard will win and what business model makes sense in an agent-driven world - all of which are very early. We’re also early in understanding the AI problem itself and why crypto is a relevant solution. I think @Tempo will be a great success but it’s still built with incentives of private companies in mind. There could be limitations to what agents on Tempo can do. We saw for the first time last week what censorship in AI looks like. We don’t really know the full ramifications yet and likely will not know for the next 6 months how AI will be governed by centralised forces. 2026 has been a tough year for crypto startups. It happens to coincide with a crypto bear market. I think a lot of this comes down to macro and timing / understanding of the AI problem itself and why crypto matters to solve it. However, there are great early signs of crypto x AI working. In particular, the traction that we have seen in decentralised training (e.g. @PrimeIntellect, @Pluralis) and harnesses (e.g. @NousResearch) is very real. There is a real case for truly ‘Open AI’ to extend beyond just these use-cases in future. Whilst timing has not been perfect yet for crypto x AI, I am very bullish on timing of crypto x AI going into 2027. I strongly believe it is only a matter of time.

Thanks all who came out for an evening with Index and @frachtisvc at @JoinEdgeCity @EdgeEsmeralda!

At @EdgeEsmeralda for the week! Co-hosted an awesome event with our @frachtisvc fam @indexnetwork_ last night - so many epic people around the place. From neuro, to AI healthcare, to biological computing, to serendipity networks Never met so many fascinating people in one location - loving my time here. Reach out if you’re around and would like to chat!

crazy how agents really make you feel like you can clone yourself to make processes 10-100x more efficient spent the last 2 weeks building relationships with talented founders in-person received many deals in the meantime across our website, X, email and Telegram - all of which were assessed by our agents on the backend Now back behind a PC and able to look at what new founders I should be building relationships with next, as signalled by our agents rinse and repeat - fine-tune to make this process more efficient over time

Great start to Consensus Miami at our @frachtisvc founder and investor circle! Cool to hear from many startups about their early traction. Lots happening on the builder side right now. Fantastic time to be creative about solving hard problems. We’re investing - DM me to meet!

Now in Miami for Consensus 🌴 Hosting our @frachtisvc founder and investor breakfast tomorrow morning: luma.com/6ge8xjzc Looking to meet teams here building crypto apps / agents that frontier models and incumbents can’t easily replicate. DMs open.

SF / @Stripe Sessions takeaways 🌉 Crypto x Fintech: - fintech is taking over crypto - they’re raising money without telling investors they’re actually using crypto underneath, crypto founders take note - corporates are extremely interested in stablecoins, not crypto - enterprises guarantee if something doesn’t work they’ll make a customer whole, crypto teams don’t do this enough, which important in the agent era when there’s a lot of risk - imo there’s room for a decentralised network that uses all agent payment standards and is open - @Tempo will be huge for enterprises and massive for blockchain adoption, TBD on timeline for a 3rd party developer ecosystem there - still think some of the best apps in the world going forward will uniquely use crypto rails in a way competitors (like pure AI) don’t to differentiate and provide better UX / value to users AI - AI infrastructure is everywhere, if people think crypto pitches are saturated, try AI - lots of talk about what AI agents will do but unclear exactly what they’ll be doing in outstanding use-cases - somewhat consensus in SF that blockchain can improve trust of AI, yet very little founders have answers as to how to make it happen - becoming more consensus that hardware is a great way to build strong moats in the age of AI - accessing prop data is a huge moat for AI startups, there’s likely industries that are sitting on very interesting datasets in niche markets that will be high growth in the future, which AI startups can target - interesting investment opportunity Startups & Fundraising: - very few founders have good responses for why frontier models won’t eat their startup in the next 6-12 months - the best startups glue incredibly complex workflows together, which require special kinds of permissions to do so and would be hard for competitors to replicate - surprising amount of trad VCs have dedicated crypto teams - talent in SF is best I’ve ever seen, barely had 1 bad pitch across crypto or AI — overall I’m very bullish SF and the intersection of crypto, FinTech and AI. The vibes are high in SF. People are curious in how technologies will intersect and care about distribution / moats. Could see some generational winners here being built at the intersection if not already, then in the future. Now headed to Consensus Miami! DMs open for builders keen to meet there 🌴