Draper Associates

Day 5 of the Iran War: First our hearts go out to the folks in the region. Human tolls are real, regardless which side you are on. After catching up with a few friends currently in the Gulf Cooperative Council (GCC) countries, here's what we should all know re: what is happening on the ground, opinions on how long it will last, and implications for startups. 🇦🇪 UAE Hundreds of drones hitting Dubai, Abu Dhabi, but thankfully with limited damage and casualties. Air defense system working very well. But closed air space has created emotional stress and logistical nightmare for expats who feel trapped, and is damaging the brand in the short term. 🇸🇦 KSA US targets (embassy and Aramco) getting hit, but people generally unharmed. Nothing major. 🇱🇧 Lebanon... drawn into this war in a BIG way. 🇴🇲 Oman: Aside from US bases hit. Muscat airport running smoothly. FYI, the European countries have set up travel desks at airports in GCC, but US travelers are mainly stranded due to State Department budget cuts. Many believe the war will be soon coming to an end because: 1. US munition running low. In fact, they may have chosen to evacuate US embassy instead of wasting patriot missiles on defending it. US makes 600 missiles/yr, and we had about 1000 in stock. Remember, we still need munition as deterrent for other conflicts such as Taiwan. 2. Stock market reactions. KOSPI crashed and recovered. But if the markets globally take a dive, politicians may notice and act. 3. Missile volume seems to be declining. This is the most important fact. Others believe we have let the bees out of the beehive, and there is no quick way out. While good that the US took out the Iranian dictator Khamenei, the bad news is that Khamenei's son is rumored to run the country. Iran is not Venezuela 2.0. Political leadership was weakened, but the assassination triggered the contingency plan - many decentralized groups of rogue militant soldiers have been activated to launch drones and missiles at its adversaries. And the implications for startups: 1. All bull markets end in unexpected ways, and this way will not be an exception. If this war does not end cleanly, we may have some weird inflation or fear triggered bear market. Common concerns around private credit, or AI circular deals are priced in. But I worry the second order effects of a prolonged war may surprise people. Politicians will continue to deny this, but eventually markets will figure out. 2. Global military spending trends are real and undeniable. Expect strong secular trends to override underlying economic cycles. Watch out for anti-drone technologies (Aurelius Systems), defense AI Models (ScoutAI), military orchestration (NODA AI) , advanced surveillance technologies (ICEYE US), ... and many more to come. For more from the ground, good reporting here from Herb Scribner of Axios: https://lnkd.in/gUkh_UPq

MightyFly now has — 3 full-scale aircrafts built — 400+ autonomous flights completed — $1M+ in revenue — A Special Airworthiness Certificate — A $220M LOI for intra-island delivery — A $50M healthcare contract signed And they just closed a $10M round. Traditional logistics needs warehouses, ground crews, runways, fuel networks. Every new route is a capital project. MightyFly's hybrid eVTOL carries 100–500 lbs up to 1,000 miles and makes multiple stops in a single flight with no runway and no ground crew sitting around waiting. The $319B expedited delivery market is going to be won by whoever cracks the unit economics of autonomous middle-mile logistics at scale. Manal Habib and the MightyFly team have been heads-down proving this works in the real world. Congrats to the whole team. Excited for what's next.

The polygraph hasn't changed since the 1920s. That’s 100 years with no innovation. Polygraphs are stressful and require a trained examiner, and they produce false positives that should make any investigator nervous. AI Seer just ran a 118-person IRB-approved lab study in Singapore. Their Multi-Spectral Reality Detector measured pupil dilation contactlessly and hit 83% accuracy. More importantly, it produced 10x less random variation than electrodermal activity (EDA) (the next-best polygraph measure). That's not a small gap. A researcher from Taiwan's Criminal Investigation Bureau independently validated this in a separate trial at Central Police University. When pupil dilation and EDA were combined, both false positives and false negatives dropped. The results are being presented at a peer-reviewed forensic science conference, but it gets even more interesting... The automated mode runs about 10x faster and 10x cheaper than a traditional polygraph with a human examiner. The tech is already patented in Taiwan, with patents pending in the US and Singapore. This is worth paying attention to.

The problem: 95% of enterprises report zero ROI on their agentic AI investments. Not because the technology doesn't work. It does. But because current deployment models force an impossible choice: → Roll out generic agents at scale that don't fit how people actually work → Or hand teams frameworks to build custom automation—requiring engineering armies most companies don't have Neither approach scales. Both ask organizations to predict how work happens instead of understanding it first. The vision: What if agentic systems could start by learning from work as it really unfolds—before trying to automate it? What if enterprises could build adaptive systems that evolve with their people, not around them? We invested in Liminal, and they’re building that future. Their Behavioral Agent Automation Platform (BAAP) observes how work happens across an organization, finds friction points, and builds automations based on real behavior, not guesses. It’s orchestration infrastructure that connects individual productivity to enterprise transformation. Automations emerge from reality, not assumption. The impact: → Intelligent automation that shapes itself around real usage → Workflows that evolve naturally instead of being imposed → Secure, governed scale without human bottlenecks → Systems that continuously improve rather than decay with complexity Every major technology era creates a new layer of infrastructure. Agentic AI needs orchestration infrastructure. BAAPs make intelligence adaptive. This is the orchestration layer that will define the next generation of enterprise agentic AI.

Steve Jobs: First-time entrepreneur Mark Zuckerberg: First-time entrepreneur Larry Ellison: First-time entrepreneur Don’t let a VC firm tell you that you need to be “more experienced”.

Longevity is not a healthcare sub-sector. It is an emerging investment vertical with civilizational upside. Draper is backing the founders building it. We're looking for founders building the next Biocures.

This program is the American dream. America’s Startup is the platform for the best and brightest to create companies that have never been seen with ideas that have never been thought of. There are opportunities that are far beyond our imagination. Thanks to Scripps News for having Tim on to talk about America250’s opportunity for college and university students and to FOX for helping get the word out. https://lnkd.in/emf2kVkD

One sure way to curing cancer is attacking the MYC gene. By the numbers, this gene family is responsible for 70% of cancers, and has resisted all attempts at treatment for more than 50 years. Any drug/therapeutic that can do this is a highly investable cause. But…investing in drug discovery biotechs, especially for these tough targets, has historically been a hard sell for VCs. Even with AI. Simulations have been either too slow to be useful, or too inaccurate to be trusted. And the real cost of that time only amplifies when you’re looking to treat undruggable targets. What we’ve needed for some time is someone developing drug design superintelligence. That’s what Ten63 Therapeutics and Marcel Frenkel, PhD have built. They have the world’s first “large quantum chemical model” which allows them to run 100x faster quantum-level simulations, generating TRILLIONS of potential drug treatment options per protein.  Why this matters, besides faster therapeutic development: Data generation. Other drug discovery companies going after these hard-to-treat targets fail because their AI needs a starting point to analyze. When there’s no positive starting point to learn from, they can’t go anywhere. Ten63’s platform doesn’t need existing data. It generates novel molecular insights all on its own. Which they’re using to attack MYC (and other difficult disease targets). The cure for cancer will be here in our lifetimes. To any and all patients waiting for a breakthrough… their team is hard at work for you. Congratulations Marcel and Team Ten63 on your fundraise - proud to back you yet again!! - and great to have you on the cap table with us Chugai Venture Fund (CVF) Gates Foundation RYSE, K5 Global, Cape Fear BioCapital Black Opal Ventures FundBox Duke University Duke Capital Partners More about the fundraise in the Triangle Business Journal: https://lnkd.in/g3D9V5Hc