A year ago, LLMs hadn’t drastically transformed software engineering. A lot has changed! Engineers are now delegating much (most?) of code authorship to agents, and the SDLC is being redesigned from scratch. Here are the emergent patterns: https://t.co/29PRaYnkdc
I hate pitch meetings. So I take at most 2-3/week. The rest of my time is spent meeting researchers and engineers. Anywhere else this would be sacrilege (coverage goals?!) but it’s the only thing that works for me. Find a place that works for you.
Where you start your VC career can make or break it. Most people choose wrong. My thoughts after nearly twenty years of hiring and promoting junior VCs.
Free Radicals is back! Today’s guest is @ElliotHershberg, partner at @AmplifyPartners and author of the popular blog Century of Biology.
We talk about GLP-1s as a breakthrough moment for biotech, why drug development is starting to behave like software, and how falling discovery costs could finally free biotech startups from selling themselves to pharma.
Other topics include Elliot’s "massive markets, medium prices" thesis, the rise of consumer and "n-of-one" medicine, and Sid Sijbrandij @sytses going founder mode on his own cancer.
Thank you to @SynBioBeta for hosting us at their conference, where we recorded this episode, and several others we’ll be releasing soon!
Follow me, @FreeRadicalsBio & @EricDai_BioE for weekly episodes with leaders in biotech and longevity.
0:00 Intro
2:12 What era of biotech are we in?
3:58 GLP-1s: the once-in-a-generation breakthrough
7:04 Inside Lilly's trillion-dollar valuation
9:09 The Gillette Era of GLP1s
9:54 The Bryan Johnson Era of Biotech: Biohacking goes mainstream
12:55 The $1T question on Consumer Biotech
14:33 Patent cliffs, fast followers, and building for "me-last" therapies
18:20 Disintermediating pharma: HIMS, Loyal, and the standalone biotech
19:41 Why do biotechs sell to pharma in the first place?
23:03 Longevity: a science problem or an ecosystem-alignment problem?
27:14 Why GLP-1s came from pharma
29:52 Why Elliot is Bullish on Standalone Biotechs
33:03 Is Drug Development Becoming more like Software? Not if upside is capped at <$10B
41:30 The geroscience hypothesis and longevity as massive markets, medium prices
45:36 Treating age related diseases or aging as an indication
47:55 How do you value curing aging?
Hey everyone — big day for us at Skiplabs: Skipper Beta is live 🚀
Skipper is a closed-loop coding agent. Instead of constantly going back and forth with the AI, you give it a spec and it iterates internally until it produces a working software service.
We believe this is where AI-assisted coding is heading, and we’re excited to finally share what we’ve been working on behind the scenes.
Start building with Skipper: https://t.co/XYooJgn0hp
On episode 55 of Generationship, Rachel (@rachelchalmers) talks with Sunil Dhaliwal (@dhaliwas) of @AmplifyPartners about investing across major platform shifts, from cloud infrastructure to AI and digital biology. The episode explores how infrastructure, tooling, and scientific computing may shape the next decade of innovation. Tune in!
https://t.co/AL0xpOFDpa
1/ Coding agents are amazing. But once you move beyond toy apps into real services, we find ourselves wrestling Claude and Codex to get the output just right.
The reason SOTA agents struggle is the same reason experienced devs struggle: concurrency & state management are hard!
Introducing the Cosmos Coalition
A new global initiative with NVIDIA and leading AI labs to build and open-source frontier world models for physical AI. Runway joins as a founding member, working alongside NVIDIA and a set of leading AI labs to build, share and accelerate world model research and development through a common open ecosystem.
Technical markets usually look confusing before they look inevitable.
@AmplifyPartners, founded by @dhaliwas in 2012, built its reputation backing deeply technical founders before consensus had language for their categories. The firm’s pattern runs through @datadoghq, @fastly, dbt, @temporalio, Chainguard, and @runwayml, with a team built around infrastructure, developer tools, AI platforms, and enterprise software.
The next category-defining companies will not start as obvious markets. They will start as hard technical problems Amplify understood before the room caught up.
Follow this firm. Study their founders. Track their plays.
#VentureCapital #SoftwareInfrastructure #DeveloperTools #TechnicalFounders #StartupFounders
"We are getting more questions on efforts from the frontier model providers to partner with PE firms, given the importance of TTAN's (aka ServiceTitan) PE Partner Motion."
I am getting tired of folks just randomly throwing out that the frontier labs can just replicate vertical workflow distribution, depth, integrations, trust, dispatch logic, payments, financing, and embedded operational muscle overnight.
Focus is a moat too.
I met @taiuti years ago when he was the cofounder of Luma. I knew it was only a matter of time until we got to work together again. Over the years there was only one person he spoke of to be his cofounder: @_bschmidtchen. When we all finally met and heard their vision for Reactor, it was a no brainer to co-lead the Seed with @buckymoore (finally!) and @theamberyang. We’re excited to be doubling down in the A and getting to work with this amazing team. A lot of great things are on the horizon for @reactorworld
Today Reactor is coming out of stealth. We’ve raised $59M in Seed and Series A funding, led by @lightspeedvp, with participation from @AmplifyPartners, @wndrco, @Sky9Capital, and @FPVventures.
Reactor is the platform for building in the World Model era: the infrastructure that lets developers build with them at global scale for the first time. Stream from a frontier World Model to your app, in real time, all in under 10 lines of code.
World Models represent the next major shift in AI: pixels, audio and actions are generated on the fly, in real-time, in response to user inputs, and to the environment. Every time computing has made a shift from passive to interactive, entire industries appeared that didn't exist before. We're standing in front of such moment again.
Over the last 6 months, we’ve assembled an all-star team with alumni from Apple, Meta, Google, Luma AI, Netflix, and Replicate. We're already partnering with some of the biggest names and labs in the world, and hundreds of developers are already building on Reactor.
The World Model era starts now.
Cell therapies of the future will be more complex.
CAR T has delivered extraordinary results for blood cancers. The new frontier is to achieve similar outcomes for solid tumors. This is a massive challenge that will require more genetic modules—and even combinations of different modules working together.
Waypoint is building a platform to enable this level of engineering complexity. We led their Series A to help scale their engineering efforts and to run their first clinical studies.
I wrote about some of the intuition behind the design of their platform:
1. In order to efficiently navigate a large space of possible designs, we can't test new ideas in serial. Waypoint tests thousands of designs in parallel in each experiment by using pool screening.
2. Typically, pooled screening requires a trade-off in the resolution of the measurement captured. Waypoint solved this by inventing techniques to capture powerful spatial readouts from each experiment.
3. High-throughput experiments with high-resolution readouts creates an embarrassment of riches. It's hard for even the best human minds to reason about where to go next in design space. Waypoint built an AI design system that has already generated hypotheses that intrigue cell therapy experts.
You can read more about these ideas here: https://t.co/E7nNJwONsZ
Hot take: a lot of AI infra talk is still pretending every problem is a database problem.
For agents, that’s often backwards. They want quick access to specific files, weird unstructured data, and messy workflows.
The minute you add shared state, permissions, and multi-step retrieval, you enter this “our storage stack was not built for this, mate” territory.
AI Agents are making files cool again.