Venture · The investor lens
The winners of the AI era won't be the ones who understand the technology end to end.
They'll be the operators who understand a business well enough to aim the technology at where value is trapped, and who can coordinate across teams to release it. Technical depth is becoming abundant. Judgment about where to point it is not. I look for the same trait in founders, and I hold my own calls to the same standard: write them down, then check them.
Four lenses
How I pressure-test an opportunity
Problem–solution fit
Vitamin, painkiller or infrastructure? How precisely does the solution meet a real, urgent need?
Founder signal
Adaptability under pressure, clarity in uncertainty, and how someone responds to being challenged.
Market reversal
Assume the product already exists. Who would miss it if it vanished, and what would that cost them?
Opportunity mapping
Not a binary invest-or-pass, but a map of current fit, future collaboration and network value.
My calls, scored
I write predictions down. Then I check them.
Anyone can have a take. The test is whether it holds up. Here are calls I published on my blog, scored honestly against what actually happened.
RightAI agents will start evaluating and buying on people's behalf.
Within months, OpenAI and Stripe launched an open protocol for checkout inside ChatGPT, Google followed with its own commerce protocol backed by major retailers, and Visa and Mastercard built payment systems for agent-initiated purchases.
RightBrands will have to market to the AI, not just the person ("B2A").
"Generative engine optimization," structuring product data so AI agents recommend you, became its own discipline, and Wharton researchers began describing "marketing to AI" as a distinct capability.
RightTrust and machine-readable data will beat slogans.
The infrastructure that shipped was about exactly this: verifying that an agent is legitimate, scoping its spending, and feeding it structured product data.
Too fastB2C is dying, and agents will take over the purchase itself.
Agents won discovery first, not checkout. In-chat checkout was pulled back after few merchants integrated, and AI platforms are still a small slice of U.S. e-commerce. Merchants wanted AI to send them shoppers, not to be the cash register.
Published November 3, 2025
Wall Street just bought the revolution: what Morgan Stanley's EquityZen deal means for investing
RightOther major institutions will buy or build competing private-market platforms.
Three days later, Charles Schwab announced a $660M deal to acquire Forge Global. Morgan Stanley closed EquityZen in January 2026.
Half rightRegulators will step in with new rules on disclosure, pricing and access.
New rules came, but in the opposite direction: the SEC proposed expanding retail access to private markets rather than tightening it.
Too earlyTokenized platforms will emerge promising true access, and get acquired too.
Tokenized securities moved onto the SEC's agenda. The acquisitions haven't happened yet. Still watching.
What scoring myself taught me: I'm usually right about direction and early on timing. The infrastructure shows up before the behavior does.
My read now
Ramp, November 2026
A live call, dated today, to be scored next year.
What I noticed
In a year, Ramp went from my LinkedIn feed to airports to conferences. That isn't just brand building. Most of Ramp's revenue still comes from card swipes, and airports and conferences are where business spending happens. They're marketing at the moment of the swipe.
Problem fit
A painkiller for finance teams. But the biggest pain is shifting, from expense reports to AI costs nobody can see. Spend on AI tokens across Ramp's own customers has grown roughly twentyfold in a year.
Market reversal
If AI absorbs more work, companies may spend less on travel and more on AI, and that spend arrives as invoices, not card swipes. Ramp's free AI-cost tool, open even to non-customers, reads to me as protecting the core before it shrinks.
What I'd advise
Double down on accountants. Ramp just entered accounting firms for the first time, and one firm brings every client it serves.
What I'd say no to
Breadth for its own sake. Speed is Ramp's edge, but hundreds of new features in a year spreads attention thin. At a roughly $60B valuation, focus matters more than range.
What I'm watching
Whether the free AI-cost tool turns non-customers into customers, and whether card swipes become a smaller share of revenue before an IPO. It also fits my 2025 thesis: companies are staying private longer, and the public misses the growth.
Written November 2026 · Score due November 2027
Where I learned it
Same questions, different rooms
Venture fellowship
Pegasus Angel Accelerator · Late 2024
A short fellowship where I first learned to read a company the way an investor does. I scouted at MIT $100K pitch nights and Boston startup events, judging the problem, the founder, the market, and what the numbers don't show.
FounderAI
Founder–investor matching platform · London
Built the stage-gated roadmap that checked a company's readiness before it reached an investor. Guided 7 companies toward investor readiness and brought on 3 advisors and angel investors through my network.
Today I use the same lenses advising an early-stage startup on positioning and go-to-market.
Curated rooms, not crowded ones. That's where signal lives.
Sources for the scorecard and memo: company announcements from OpenAI, Stripe, Google, Visa, Mastercard, Charles Schwab, Morgan Stanley and Ramp; SEC releases; reporting from Bloomberg, TechCrunch and industry press, 2025–2026.
