Sponsored
💬 "Tell me why we keep losing to [competitor] — with quotes from the actual calls."
Parse notes
There's a difference between a job and a task that most AI coverage misses. A software engineer's job is not "type code into an IDE." That is one task. The job is to build useful software, make tradeoffs, debug edge cases, and ship something that works. Once you separate the two, the employment debate gets sharper. ATMs reduced tellers per branch but also reduced the cost of operating branches, so banks opened more, and total teller employment rose. Where in your own market are buyers tolerating less of something they actually want, just because it's too expensive to deliver?
The current sky-high valuations in the artificial intelligence sector are a temporary bubble driven by supply and demand. While developer and prosumer AI tools can achieve unprecedented rapid revenue growth right now, this window will likely close in one to two years as the market saturates and competition intensifies.
Founders now have two roles: keep the installed base genuinely happy while simultaneously building the best agent in their category, managing internal resistance, and not letting customers trap them in their past. CEO resignations are up significantly at the moment as a lot of founders and CEOs have decided this two-job reality is not something they want to do. The rule in B2B has always been to grow or die. Right now, there is less margin for error than ever before.
Landing pages are input materials for agents. Businesses can either choose to make their information easily readable by agents or let every tool scrape and guess. This post discusses how to create landing pages that can persuade humans yet still have a structure that makes it easily readable by agents.
MedScout put one of its best AEs on a projector and recorded everything. Which tabs he opened. Where he paused. What made him say "this one is interesting" versus "this one is not." That tacit knowledge usually lives in someone's head as pattern recognition and never gets documented, which is why most AI account qualifiers fail. The next time someone on your team says "we should automate qualification," the right first question isn't which model to use. It's whether anyone could write down what your best rep is actually doing.
We're looking for the founder of a VC-backed startup to help write TLDR Founders for hundreds of thousands of subscribers. Curators have been invited to major tech events, scouted for Tier 1 VCs, and get early access to unreleased TLDR products. Learn more [11].
Automate business operations, product development, and growth tasks using AI agents.
Automate Instagram DMs to engage users, qualify leads, and manage follow-ups.
Generate formulas, charts, and data analysis in Excel using natural language.
A designer's job is not the Figma file. A PM's job is not the spec. An engineer's job is not the code. Those are outputs, and AI is great at producing them, which is exactly why the hidden part of work suddenly matters more. Sitting with the problem. Rejecting the cheap answer. Making judgment calls the prompt can't make for you. The next time you're hiring for a role and find yourself impressed by how much someone ships, the better question is whether anyone else could have decided what to ship in the first place.
Uber burned cash, courted controversy, and became a massive public company. WeWork did the same things and went bankrupt. The behavior didn't change, only the verdict did. In 2022, investors admitted ZIRP-era excess had pushed companies to build for the wrong environment. The same dynamics are back now under a cleaner label called AI, and "this time we know what we're doing" has historically not aged well.
Removing toll booths between users and the core product strengthens the castle by commoditizing the complements around it.
A "True Product" combines high customer focus with high standardization, resulting in scalable solutions via technology.
A single AI round can now absorb whole funds.