Parse notes
Sponsored
Every integration comes with the same overhead: OAuth flows, token storage, refresh logic, and provider-specific quirks. It's tedious, time-consuming, and pulls your team away from core product work.
Cloud software valuations have cratered to 10-year lows, with the median NTM revenue multiple hitting 4.1x. Investor confidence in the SaaS business model is fracturing as AI disruption risks threaten the retention rates and terminal value assumptions that powered the last decade of software investing. The path back requires companies to prove stable retention over multiple quarters.
a16z surveyed 100 Global 2000 CIOs and found OpenAI still leads enterprise AI with 78% production usage, but Anthropic posted the largest share gain at +25% and now sits at 44% penetration. The surprise winner in enterprise apps is still Microsoft, not the startups dominating tech Twitter. Average enterprise AI spend rose from $4.5M to $7M and is expected to hit $11.6M this year.
The old model was to sell the software and let the customer figure it out. Businesses now need to do the work to make their AI great before they hand over the keys. AI agents will fail if improperly contained. It is up to the AI vendor to make it work. Customers shouldn't have to get agents to work.
Stop calculating “potential customers × potential price” and start with what the world currently spends solving your problem. That's your actual market today. Everything beyond that isn't market size, it's a thesis about growth. Most founders are stacking 3-5 speculative theses without realizing it, and since each compounds risk exponentially (30% × 30% × 30%), understanding this can dramatically improve how you allocate time and opportunity cost.
AI has collapsed the cost and time of building and testing products, which means feasibility, viability, and usability are no longer the hard parts: desirability is. The critical bottleneck in product discovery is now understanding what customers actually want. Teams need to shift investment toward deeper customer insight and research to answer the fundamental question before building anything.
Sponsored
connects equity, compliance, and stakeholder workflows in one platform - so you can issue shares with confidence, stay compliant by default,
Modern sales CRM designed for B2B startups.
Robynn is an AI marketing platform that helps founders create branded content.
AI agent for real frontend tasks: build UIs, refactor, and integrate.
Big Tech has figured out how to acqui-hire without the acquisition. License the IP, hire who you want, leave the rest behind. No regulatory review, no shared outcomes. Microsoft did it with Inflection, Amazon with Adept, and Google with Character AI. The unwritten oath between founders, employees, and investors is dead. When picking your next company, ask your VC who else they've backed in your category and what happened to employees at the founder's last company.
Microsoft, Oracle, NVIDIA, and Starbucks all started by selling something to a tiny, often invisible market before stumbling into bigger opportunities. The playbook isn't “envision a radically different future and build it,” it's figure out a real need by selling, learn by iterating on what doesn't work, avoid trying to predict the future or convince markets, and stay in the game long enough to catch a demand wave when it hits.
Slack faced issues with feature bloat and complexity by late 2019, as the product accumulated many additions without a cohesive structure.
Chase your passion with a blend of curiosity and practicality rather than as a blind directive.
Clarity of thought stands as a key differentiator for successful founders, influencing all aspects from hiring to sales.
AI could create a new form of advertising that may actually flourish.
We are now shifting into an 'agents as customers' era - companies need to be thinking about how to become the vendor that AI agents prefer.