Early-stage venture discovery and design, incubating new health and wealth ventures. Ex-frog, teach design strategy @nyuniversity, Startup Studio @cornell_tech.

Brooklyn, NY
Joined April 2011
Now streaming: “The Making of an Iconic Movement: The El Primero.” tinyurl.com/tb4fdcs8 Seven years to develop. Nearly lost forever. Introduced in 1969 as one of the world’s first automatic chronograph movements, the Zenith El Primero survived the quartz crisis to become one of the most important chronograph movements ever created. We sat down with Zenith’s Romain Marietta, Laurence Bodenmann, and David Serra to explore its development, the story of Charles Vermot secretly preserving the tooling needed to produce it, and why its original architecture has endured for more than 50 years. Watch now on TeddyBaldassarre.com
2
22
4
355
33,347
🔥
4700 word labor of love. I've seen all the patterns and heard every story about in-house brand/creative studio teams. This is my vision for where brand teams can go (and why I joined @tryramp). Brand systems that behave like jazz. Brand as software.
1
121
What exactly is the strategy for Instinct if they don't plan on charging users? Are they hoping to win in the enterprise? Will be a really difficult play there. Advertising? Affiliate sales from agentic actions somehow? Just hope for an acquisition?
Instinct looking to raise $1B, potentially at a $10B valuation. Compute costs are high and Noah doesn’t want to charge users for the product. Instinct has raised $350M to date, mostly uses open source models and wants to own chips and data centers theinformation.com/articles/…
1
82
This is exactly why agentic commerce is such a huge opportunity for really well-aligned and niche products. Your specificity will be seen by LLMs and your likelihood of recommendation will be much greater than trad search
Personally, I too feel I am discovering more things, places, and experiences than ever before and I am much more likely to buy from more undiscovered, long tail businesses. (What t'was buried on search result page 12,345 is now just one good prompt away.)
79
This is going to continue to be a critical strategy and GTM choice, not just a marketing one. When anything can be built, what you focus on, and more importantly don’t focus on, will give LLMs more conviction to choose you
I used to always advocate for folks to be generalists, and I still think you should behind the scenes. But publicly, I think it’s more important than ever to be known for one thing. Why? Because ChatGPT is becoming one of the main ways clients discover me. I’ve spent a decade building an association with one thing (design subscriptions). Now anytime a client asks an LLM where to go for one, I’m the top recommendation. Not because I’m the best designer, have the widest range of skills, or work with the biggest names. But because my name has become inseparable from the topic due to publishing thousands of pieces of content on the matter. The entire time, while I thought I was building an audience, I unknowingly was building a set of training data for LLMs. Discoverability is changing dramatically, and if you play your cards right, you’ll never have to beg for clients ever again.
95
Bullish systems thinkers
Because this is true, you’re going to see a generation of investors who built their careers as sector, theme, or business model experts shift their focus to the one thing that is evergreen: understanding people.
2
97
I feel like anime in the US would be nowhere near as popular culturally as it is today had it not been for Peter Cullen.
Rest in power, Peter Cullen. One of one. You will be missed.
88
Fuck ya. RMC-9805 next 🙏
Today we approved a first-in-class targeted therapy for metastatic pancreatic cancer – one of the hardest cancers to treat. Today’s approval provides a critical new treatment for patients who’ve had very few options. fda.gov/news-events/press-an…
1
73
🌳
Researchers at Cornell have created an interactive map identifying 1.8 million individual trees in New York City, with a new approach that could optimize the placement of trees for cooling in cities around the world. In the study, researchers identify the types of trees using satellite imagery as well as on-the-ground datasets and 3D data collected with lidar, short for light detection and ranging. The approach was 82% accurate overall and higher for common types of New York City trees. It’s the first “wall-to-wall,” citywide map identifying trees in natural areas and other unsurveyed areas, which together constitute an estimated 65% of the city’s canopy and have not been accounted for on previous maps. The research team will use the map to determine which trees are best for cooling, and the data could inform a host of other planting decisions where the type of tree makes a difference, from reducing allergens to managing invasive pests. Read more: news.cornell.edu/stories/202…
95
there’s an opportunity to create an OpenRouter for time. @deepseek_ai just changes their API pricing so that it’s 50% less expensive if you use it on weekends. Pretty significant savings if you holdoff on using the model during off hours. The idea is it’s not about optimal model selection at the point you are running the task, but instead designing and timing runs based on urgency vs savings. I imagine cost orchestration will continue to play a big role in how companies make sense of AI COGS and margin when the exact same value of output can cost you 50% less if you did it on Sunday vs Monday. Curious if anyone is working on this yet, would love to chat.
1
76
AI slop, but coffee
Blank Street has raised $75mn from investors including private equity firm General Atlantic as the coffee chain pushes to expand on the US west coast and offer new products such as ice-cream in an attempt to become a “daily luxury brand”. With @GeorgeNHammond
2
145
a seller that clearly exposes its own limitations or bad-fit contexts could become more recommendable by LLMs. Sort of counterintuitive in that your website is normally trying to maximize consideration, but with agent-mediated buying, being clear what you stand for (and what you don’t) will give the agent more conviction to make a call.
43
AI is making it easier than ever for scammers to pull a fast one, which is particularly threatening for older adults and especially those suffering from cognitive function loss. There's a tension between independence and the kind of support we want to provide to aging parents. I'm helping explore an incubation in the area of how we help protect older adults from fraud and want to get a sense of product pricing. If you have 2-3 minutes, and are concerned about how your parents protect themselves from fraud, would you mind completing the pricing study below? find.atherethere.com/survey/…
1
12,292
Is the data center backlash mostly driven by concerns related to water, noise electricity, environmental factors? Or is it mostly stemming from an anti-corporation and anti-elite sentiment?
1
67
For vertical AI agent startups, how you categorize yourself could be costing you. Especially if you are early, don’t do so purely based on the vertical - save that for your TAM when talking to investors.
1
68
The whole benefit of being vertical is that you can be specific, but when you are still at too high of an altitude like "healthcare agent", I've found that buyers will compare you to the foundation models in their mind and the agents they can spin up using claude or codex.
1
1
46
Putting a stake in the ground and aligning your positioning and billing to specific outcomes with higher pricing power means instead of being anchored against a Claude Max plan at $200 a month, you can charge $2,000.
28
Hell ya
Moderna and @Merck today announced positive topline results from the Phase 3 INTerpath-001 trial evaluating adjuvant treatment with intismeran autogene, a novel investigational mRNA-based individualized neoantigen therapy jointly developed by Merck and Moderna, in combination with KEYTRUDA® (pembrolizumab), Merck's anti-PD-1 therapy, in patients with completely resected Stage IIB-IV melanoma. Read more:
56
In manual search, the buyer compares a product against their own ideal. In agentic search, the AI is comparing its representation of the buyer’s ideal against its representation of the product. The implication is that now there can be noise on both sides of the match.
1
108