THE THINKING
The methodology, applied.
The same commercial and GTM problems keep showing up. These are the patterns, named plainly, with the fix attached.
Featured this week
Two pieces worth a Wednesday.
The Authenticity Moat: Why Manual Outreach Still Closes (and When To Automate)
Manual outreach is a founder personally sending individual, non-automated messages to potential customers. It still closes better than automated sequences at the early stage, and not because of effort: trust and relationship-building work differently when you're still discovering what converts.
Claude for Small Business closed the integration gap - but only incrementally
Mid-May, Anthropic embedded Claude directly into QuickBooks, PayPal, HubSpot, Canva, DocuSign, Google Workspace, and Microsoft 365, packaging workflows for payroll, invoicing, sales, design, marketing, and month-end close. The internet spent the weekend losing its mind about the new Claude for Small Business.
All insights
Claude Code for Non-Technical Founders: How to Build Without Writing Code
Claude Code builds software from plain-English instructions, so non-technical founders can ship deployed web apps without learning to code.
The AI Startup Failure Report: Why Most AI Startups Won't Survive the Next Model Release
About 40% of AI startups launched in 2024 shut down within two years, but the dividing line isn't model quality, it's whether the company controls something foundation labs structurally can't ship: proprietary data, an embedded workflow, a regulatory path, or a distribution surface. The test: if a stronger model shipped tomorrow, would your product get stronger or weaker? AI didn't change why startups fail; it compressed the timeline.
Ungated Education As a GTM motion: Why Your Best Content Shouldn't Be Gated
Ungated education means un-gating your playbooks, benchmarks, templates, and methodology and treating the content itself as distribution, not a lead magnet. Gated content converts up to 41% of visitors but only 62% consume it, and it's invisible to AI search at the problem-recognition stage. Ungated gets 20-50x more downloads; top B2B now publishes 65-75% ungated. Keep 20-30% gated at mid-funnel.
The Distribution Problem Nobody Talks About: Why Nobody's Finding Your Product
Product-market fit and discovery are different problems; if your product converts but nobody finds it, you have a channel problem, not a product one. Run a channel-demand map: write your buyer's problem in their words, find the rooms where those words already appear, and rank them by intent rather than audience size.
AI advertising disclosure: what ChatGPT's ad platform means for consumers and founders
ChatGPT's ad format integrates sponsored results into AI responses. The Trust Transfer Problem explains why the channel works short-term and why founders should build genuine audience relationships before the regulatory window closes.
The Authenticity Moat: Why Manual Outreach Still Closes (and When To Automate)
Manual outreach still closes better than automated sequences at the early stage because automation scales a motion but can't discover one; the authenticity moat is real while you're learning what converts. Automate only once the motion is proven across 50+ sends, and keep a human in the first reply and the close.
How To Get Your First Customer When You Have No Testimonials
Your first paying customer is a separate problem from product-market fit; land it with a four-step cold approach - a warm list of ten, a zero-proof message that trades proof for specificity, trust you structure rather than claim, and a real price you say out loud.
Your Buyers Are People, and They Buy Like People (aka the B2B Myth)
There's no such thing as a rational B2B buyer. People decide on emotion and justify with logic, even at enterprise prices, so write your commercial narrative to the person's fear and win, not to a faceless org.
Why Building In Public Is Awful (and the Version That Isn't)
Building in public is awful because it makes you perform the work and do the work at the same time, on a treadmill that rewards what performs over what's true; the honest version is narrow - say true things when you have them, and stay quiet otherwise.
Boring wins: why unsexy B2B is the smartest play in 2026
Boring, unsexy B2B problems win because they have almost no competition: nobody wanted to build the fix, and the founder who lived the problem was the customer first. Problem selection matters more than technology selection. Find your problem with three questions: what repetitive past-job work you'd have paid to kill, is anyone building it for that buyer, and can the buyer count the time saved in hours per week.
Startup Pivot Timing: When To Pivot and How To do It In 2026
Wilbur Labs 2026 data shows pivoting startups raise 2.5x more capital and grow 3.6x faster; the real risk is persevering too long on a dead hypothesis.
Agentic AI Is Killing Per-Seat SaaS Pricing
Per-seat pricing was always a proxy for value, and AI agents break it. Seat-based SaaS revenue is projected to fall from 21% to 15% of the market by 2030, so price the outcome your software produces, not the seats it's accessed from.