The reading list
Essays, tech notes, and the in-between.
- Product·3 min read
Corporate Venture Capital: Innovation as Investor
Fred Wilson stood on a stage in 2016 and called corporate investing 'dumb' — and for most corporate venture funds, he was right. But GV, Salesforce Ventures, and Intel Capital keep proving the exception, and the difference isn't intent, it's structure. CVC done well isn't a growth engine; it's a sensing instrument — bought optionality on futures you can't build internally. Here's Chesbrough's map of when it works, the design choices that decide it, and the boom-bust history everyone forgets between booms.
- Product·3 min read
The Standup Autopsy
The daily standup has the cleanest derivation in all of agile — peers synchronizing and surfacing blockers in under fifteen minutes — and it's still the most reliably dead ceremony on most calendars. Here's the full autopsy: where it came from, the five ways it dies, three questions to diagnose your own, and the replacements each mapped to the constraint it actually serves.
- Product·3 min read
Innovation by Spin-Out: The Sister-Company Play
Nestlé didn't incubate Nespresso — it exiled it. Separate company, separate CEO, separate headquarters, a sales channel that violated everything Nestlé knew about selling coffee, and nearly two decades of patience before it became a pillar. Sometimes an internal unit, however protected, isn't separation enough: the new business's economics are so hostile to the parent's that the bet needs its own company. Here's when to make that call, what a real spin-out requires, and the case histories — Nespresso, Alphabet's Other Bets, Cisco's strange spin-in machine — that the pop versions get wrong.
- Product·3 min read
Handling Customer Requests: An Intake You Can Defend
The backlog where customer requests go to die isn't a filing problem — it's a decision problem wearing a filing costume. Here's the intake pipeline I actually run: one ledger, five interrogation questions, clustering by problem instead of by feature, and the exact wording of a no that builds more trust than a silent maybe.
- Product·3 min read
40 Million Daily Users vs 5,000: What a North Star Is Actually For
In 2024, Duolingo filed a 40.5-million daily-active-user number with the SEC, up 51% in a year, with more than ten million people holding streaks over a year long. The same year, Rabbit sold roughly 100,000 R1 devices and its founder admitted about 5,000 people used one daily. Both companies had a big number to point at. Only one of the numbers measured value received — and the distance between those two kinds of number is the entire argument for choosing a North Star carefully.
- Product·3 min read
Healthcare.gov: The Rescue That Invented a Playbook
On October 1, 2013, Healthcare.gov launched after roughly three years of work across more than fifty contractors — and collapsed on contact with its users. Reportedly six people enrolled on day one. Two months later it worked for the vast majority of visitors, and the fix wasn't a relaunch or a new contractor. It was a small team adding the things the project never had: monitoring, twice-daily contact with reality, one prioritized list, and small releases shipped continuously. The interesting question isn't why it failed. It's why the rescue worked — and how a two-month emergency became codified government practice.
- Product·3 min read
Innovation Accounting: Funding What Doesn't Fit a P&L
You can build the separate unit, the lab, the intrapreneur program — and watch all of it die in a single budget meeting, because someone asked a two-quarter-old venture for its ROI. The immune system that kills good ideas doesn't always look like a hostile VP. Sometimes it looks like a spreadsheet, applied honestly, by people doing their jobs. Here's the financial machinery that funds new bets without lying about them.
- Product·3 min read
Figma vs Adobe XD: What Losing a Category Costs
Adobe launched XD in 2016 as a fast-follow with the best distribution in design software, and still lost the category so completely that its endgame was agreeing to pay $20 billion for Figma — a deal that died under EU and UK regulators in December 2023, costing Adobe a $1 billion breakup fee and XD its life. Then Figma IPO'd in July 2025, priced at $33 and closed at $115.50. The receipts for losing a category rarely come this itemized.
- Product·3 min read
We Don't Sell Saddles: Positioning With a Moat, and Without One
In 2013, Slack refused to sell 'group chat' — a category littered with corpses — and positioned against email instead; 8,000 invite requests arrived in the first 24 hours of the preview, and the company eventually sold for $27.7 billion. Nine years later, Jasper raised $125 million at $1.5 billion as an 'AI writing assistant' — five weeks before ChatGPT gave the core of that position away for free. Same discipline, opposite outcomes, one lesson: positioning is a choice about what your differentiation is anchored to, and anchoring it to something you don't control is borrowing.
- Product·3 min read
Intrapreneurship: Kickbox, 15% Time, and the Myth of the Corporate Rebel
The Post-it note took twelve years, Google's 20% time quietly died of manager incentives, and the heroic corporate rebel is mostly survivorship bias — we only hear from the ones who weren't fired. What actually produces innovation from within isn't rebellious personalities. It's systems that make permission cheap and evidence mandatory: slack-time policies, a $1,000 card in a red box, and a real venture path instead of a suggestion box.