Startups Brag They Spend More Money on AI Than Human Employees (5 minute read)
Startups brag about spending more on AI compute, viewing it as a growth marker, while avoiding hiring humans. CEOs claim AI investments replace human roles efficiently, boasting significant expenses on tools like Claude. The rise of the "AI-native company" highlights a push for minimal staff, potentially leading to unsustainable financial practices.
|
The death of SaaS has been greatly exaggerated (3 minute read)
Narratives surrounding the death of SaaS are, at best, too early. They are more of a reflection of where product leaders see the industry going rather than a decision driven by buyer behavior. The vast majority of spend remains seat-based. Consumption-based revenue has barely increased.
|
The Rise of the Probabilistic Founder (7 minute read)
The probabilistic founder is experimental by default and willing to abandon their priors. They'd rather run ten cheap experiments than commit to one expensive plan. They use quick iteration and deliberately light road maps to get results in days, not quarters. The probabilistic founder still has to be a relentless executor, because an environment that rewards running ten experiments instead of one is unforgiving to anyone who can't actually finish things.
|
|
DoW Contracting for Startups 101 (32 minute read)
The US Department of War is one of the world's largest and most stable customers. It spends hundreds of billions on modern defense systems and technology every year. Defense contracts offer funding and also long-term sustainment. Winning requires persistence, but it results in predictable margins on predictable revenue streams. This post is a high-level primer that provides a rough framework on how to approach selling to the Department of War today.
|
Who Gets an FDE, and Who Doesn't: The Great B2B + AI Debate Right Now (9 minute read)
Forward Deployed Engineers try to understand workflow, find where data is broken, configure for specific use cases, and test, iterate, and catch edge cases before going live. Several vendors are now building toward this with agentic deployment assistants. None of them are at Palantir-grade FDE quality yet, but the vendors who close this gap first will have a very different competitive position than everyone else. The FDE constraint is currently an artificial ceiling on growth - once companies get actual deployment support, they'll switch.
|
Your product has a new user. It's not human (6 minute read)
Agents now evaluate, pick, and operate products on behalf of humans who never open the UI, which makes onboarding, microcopy, and brand invisible to the buyer that matters. Stickiness built on switching cost also dies, since agents remove the pain of switching. Stripe's Machine Payments Protocol means agents can also transact. The piece pushes founders to strip UI, brand, and switching friction away from their product and look at what is left.
|
|
Toki 2.0 (Tool)
Automatically plan and schedule tasks from ideas using an AI agent.
|
|
Pricing pressure will crush you (7 minute read)
AI is not trying to replace your SaaS tool. Claude and GPT are the new ROI anchor in every procurement conversation, which makes every incumbent tool feel overpriced. Competitors multiply because shipping software is cheap, and incumbents bundle aggressively enough that the marginal module ends up priced at zero. The post walks through a real renewal where a Claude integration replicated 95% of a vendor's AI feature at 15% of the token cost, and the customer cut the renewal by 45%.
|
Coding agents ignore their own budgets (3 minute read)
Your coding agent cannot be trusted to watch its own wallet. Token spend across Ramp customers is up 13x since January 2025, and every obvious way to slow it down failed. A live token counter in the system prompt was never referenced across 14,000 agent messages. A request_more_budget tool got zero calls in 5,000 turns. When forced to explicitly approve its own overage, the agent said yes 97% of the time because it was grading its own work. The only setup that actually controlled spend was handing the approval decision to a separate model with its own view of the workspace. Never let the worker audit itself.
|
|
Love TLDR? Tell your friends and get rewards! |
|
Share your referral link below with friends to get free TLDR swag!
|
|
|
|
| Track your referrals here. |
|
|
|
