The weekly top 10 for B2B tech operators · Every Friday
Time to Profit - Probably one of the most important metrics in the post-COVID Civid/free money era. Most startups die not from bad ideas but from running out of cash before reaching sustainability. Shorter TTP forces discipline: fewer vanity bets, tighter PMF proofs, and faster elimination of anything that doesn't compound.
Now is always a good time to front-foot your architecture. For techies: Google actually has a great article on Architecting disaster recovery for cloud infrastructure outages. Here is a good DR Plan article, and here is a good template.
Sure, software is scalable, but Startups don't scale by adding people — they scale by building systems. This breakdown covers how to layer ops, reduce founder dependency, and keep complexity from killing momentum.
Roughly 90% of early-stage financings on Carta now use SAFEs, with post-money versions dominating new issuance. The VC Corner walks through the full history from Carolynn Levy's 2013 five-page draft at YC to today's edge cases founders still miss - stacked caps, silent dilution, and MFN traps.
OnlyCFO lays out 10 reasons software churn is about to spike. AI is pushing tools into the "nice-to-have" territory, and switching costs are dropping. There is probably more coming - 90%+ of AI-driven churn hasn't even hit yet because teams were locked into annual contracts.
Ok - interesting take - the Coding bit is only roughly 20% of an engineer's job. Apply Amdahl's Law: a 50% speedup in coding yields only about a 7% total productivity gain. The real bottlenecks are along the lines of research, architecture, planning, and feedback loops - these are where AI can help, but they still can't replace judgment.
The Mac mini is sold out. The 256GB Mac Studio is unavailable - Apple is ripping through inventory. Om Malik explains why - it's the AI boom. Apple's unified memory architecture, a bet made in 2020, is the only consumer hardware shipping Edge AI-capable devices as SKUs.
Backing up my post a couple of weeks back on taste, Contra Labs ran 15,000 creative evaluations across five domains and found no single AI model leads all three workflow phases. Claude wins ideation, Gemini wins mockup, Grok wins refinement. The real finding: once output clears a quality bar, every model converges on safe, averaged aesthetics. Taste is still the human moat.
Data Driven VC's April market update: public SaaS top-10 median held at 13.8x NTM revenue while the overall median slipped to 2.2x. The correction cycle that began in late 2024 looks about done. Meanwhile, M&A hit $1T+ in Q1 2026, running nearly 2x YoY on deal volume, with fewer but structurally larger deals, and Anthropic's annualized revenue reportedly hit $30B by the end of March (up from $9B at the end of 2025, a 1,400% YoY growth rate!).
SaaStr runs 20 or so AI agents with 3 humans. Total bill: $2,300/month. Their AI VP of Marketing costs $95/month, AI VP of CS $160/month - roles that run $500-800K/year with humans. Revenue swung from -19% to +47% YoY. Lemkin's hindsight rules: budget 0.5 FTE of human attention per production agent.
Get the top 10, every Friday
Curated SaaS and tech insight from around the web, repackaged for people to put to good use — free.