Summer 2026 AI Fellows is officially underway. The kickoff set the tone: great energy, thoughtful conversations, and a room full of people ready to build. Thanks to all the fellows, and to the teams across our portfolio companies who helped make the kickoff such a strong start. Interested in joining a future cohort? Apply here: airtable.com/app7xaCYEc3Nxww…

A few weeks ago, we hosted our first Taste & Distribution dinner, bringing together 20 founders, marketers, and GTM leaders for an evening of good food and even better conversation. A big thank you to everyone who joined us and made the first one so special. We’re excited to keep bringing this community together. Want to join the next dinner or our GTM Slack community? Comment “taste” or tag someone who should be at the next table.

Congratulations to @ac_edwards_1, @wellingmax , and the entire team on raising $450M at a $2.6B valuation. Proud to have partnered with @CuspAI from day one. This is only the beginning, and we’re excited to continue supporting the team as they build something truly exceptional.

@EntendreAI Joins @moonpay to Build the Agentic Financial Infrastructure We led Entendre's seed on the thesis that every new financial rail requires AI to reconcile, report, audit, and close on top of it. The agentic financial infrastructure needs its own accounting layer. We met @kvreem while he was at CloudKitchens, now Atoms, and before that Workday and WeWork, building the financial systems enterprises depend on. An engineer turned product expert, he built Entendre's AI agents to turn high-volume transaction streams into clean, audit-ready books automatically, doing continuously what used to take analysts and spreadsheets. Now that layer gets to grow at MoonPay's scale, with Kareem joining as VP of Applied AI. Congratulations to Kareem and the Entendre team. We're excited to back the best founders spinning out of the most exceptional cultures, building AI that becomes category-defining infrastructure. More on why we backed Entendre, and the Axios story, below.

News: Neion Bio announces their $23M Series A led by @caffeinatedcap! Neion engineers chickens to lay eggs that produce complex biologic proteins, in days, at a fraction of what a traditional facility costs to build. The why goes beyond cost. Resilient, domestic supply of critical medicines is a national biosecurity question now, and this is one of the few teams actually building toward it. Proud to deepen our relationship since day one with Dimi Kellari, Sam Levin, and the team. More to come! Read the story in @endpts in comments.

News: Announcing Basata's $21M Series A There is perhaps no space more in need of the benefits of AI agents than healthcare. In such a consequential space, the decisions of "what" and "how" to automate are paramount. Basata is building differently, by showing up, sitting with the healthcare administrators doing the work and building around what they actually need. That empathy compounds: Serving 500k+ patients nationwide (100k in the past month alone!). Today we're announcing Basata's $21M Series A in @TechCrunch! Read more about our conviction in Basata here 👉 basisset.com/insights/basata…

Perspectives: Investing in the Open Source Agent Stack (part 2 of 2) The agent era structurally favors open source and it's creating opportunities for founders and investors alike, both in the the infrastructure beneath AI, and in the products built on top of it. Below AI, tools are emerging across security, reliability and agent-native primitives. Above AI, an application layer is emerging, resulting in products that combine models, harnesses, and user experience to own particular workflows or audiences. These layers reinforce each other. As more companies adopt open-source agents (above AI), the demand for security, reliability, memory, and primitives (below AI) grows. As below-the-AI infrastructure matures, it lowers the barrier for more application-layer products to reach production quality. This is the flywheel. Open-source models and harnesses commoditize the foundation. Application-layer companies specialize for audiences. Infrastructure companies make the whole stack reliable. Each layer accelerates the others. The agent era is creating an enormous new market. The companies that win will be the ones that solve the hardest problems at each layer of the stack, for both open and closed ecosystems, with the kind of operational depth that no LLM can generate from scratch. If you are building at any layer of this stack, we would love to hear what you are working on. Check out the full piece from @_changxu in the comments.

Perspectives: The Future of Agents is Open Source (part 1 of 2) The agent era rewards systems over single models, and those systems reward the control that open models bring. The more agentic software becomes, the more valuable openness becomes, not because open source is ideologically pure, but because modular systems compound. As agent harnesses mature, work is decomposing into reusable primitives. Planning. Exploration. Context compaction. Code editing. Bash execution. Verification loops. Memory writeback. Error recovery. It's quickly becoming clear that the best model per primitive is not the best model overall. And having full control over prompts, context and inference, coupled with the ability to fine-tune on your workflows with open models, boosts performance and radically reduces costs. Check out the full piece from @_changxu below.

News: @parasail_io announces their latest $32M Series A Parasail is building the world’s fastest, most cost-efficient AI inference network. We knew they were onto something based on the sheer number of companies that came to us and told us about why they were switching workloads to Parasail. Today Parasail hosts GPUs in 15+ countries and generates over 500 billion tokens a day. Read more about Parasail in their TechCrunch feature 👉 techcrunch.com/2026/04/15/pa…