Build The Inbox Assistant That Saves You An Hour A Day
How to build an inbox assistant with the Claude you already pay for, so it labels mail and keeps thread context.
Everything I have written on building a company alone. I publish on Substack every Tuesday and Friday, and every article is listed here.
Free. 46 articles so far, all listed below. Read them on Substack ↗
How to build an inbox assistant with the Claude you already pay for, so it labels mail and keeps thread context.
Why your early competitors are four forces that keep customers where they are, not the startup you saw on Instagram.
How to pick one distribution channel in an afternoon with three checks, instead of posting everywhere and growing nowhere.
How to tell whether an incumbent will build your idea, using three kinds of gaps: backlog, trapdoor and model.
Why to stop hiring for experience, and three steps that test fit instead: a creative answer, a think-out-loud interview and a work trial.
How to find the one step where your funnel breaks, whether audience, product or price, and test a single fix.
How to scope an MVP so it ships sooner: sell before you build, find the delight, and build to learn.
What four founders with large followings said about making content that gets users, from showing receipts to sharing failed experiments.
A five-step way to build a personal website with an AI agent that does not look made by AI.
Why chasing every channel slows early growth, and how to pick who your product hurts most and channels that compound.
Five plays consumer apps used to get their first users, with the rooms and people to reach in your category.
Building a working app in 5 hours with Floot, and how it compares to Lovable, Base44, Bolt, Replit and Cursor.
Five habits that keep a busy solo founder from ending a month with nothing moved.
How a set of AI skills reads your subscribers, comments and niche to show whether you write for the people who actually show up.
How to build Claude skills that work: do the task first, design the system, order the actions, keep a log, build a loop.
Where a visionary founder tends to leak when working alone, and how to plug it with an opposite hire, blunt advisors and a village.
How to build an AI skill that synthesizes your user interviews so you stop building features nobody asked for.
Why your early competitors are four forces that keep customers where they are, not the startup you saw on Instagram.
Three things building Shezaar taught one introvert about being on camera, learning to code and sharing her writing.
Where stretching the truth as an early founder turns into fraud, looked at through the Phia allegations and the author's own story.
How to set up context and guardrail files so your AI agents stop repeating old mistakes and improve with each run.
How Durva ran user research for Shezaar with no infrastructure: seek criticism, talk to non-users, and keep a standing group of advisors.
Why waiting for approval from people who already dismissed your idea costs you, and why you should build without it.
Which work to keep doing by hand and when to automate, based on your customer count and two simple questions.
How to hire as a solo founder: automate first, then treat culture as a pass or fail gate and skill as a dial.
How to pick one distribution channel in an afternoon with three checks, instead of posting everywhere and growing nowhere.
How to decide whether your startup is still worth pushing through, and why setting kill criteria early saves time and money.
How IKEA sells fridges to sell cabinets, and how to find your anchor product, the friction around it, and the partner who removes it.
Five questions to tell whether your problem will last for years or disappear once better tools arrive.
Why a waitlist is not a customer base, and four moves to build an audience before your product launches.
What SF tech billboards cost, why some work, and how to test your own message for being specific, in their language and repeated.
Why a solo founder needs a structural answer to why you, and the three moats that can give you one.
Why early competitors validate your market, and how to own one unserved slice instead of rebuilding to match them.
Five steps to find your first ten users outside your website, starting with a niche that feels uncomfortably small.
How talking only to people who already agree misleads you, and a three-step check to catch your own bias about customers.
A three-step system for sounding like yourself with investors and customers: a self-audit, pre-call prep and practice.
What being authentic as a founder means in practice, and why a short, true reason for building is your anchor.
What Table For One is: a free letter for early-stage solo founders, and what it will cover each week.
How to build an inbox assistant with the Claude you already pay for, so it labels mail and keeps thread context.
A system for showing up on Reddit consistently, using the 70/30 rule and a scout, sifter and coach, without writing AI-sounding comments.
A set of prompts that takes you from a new idea to talking with potential users in hours, not months.
How to set up Channel Scout, a Claude pipeline that interviews you, scans competitors and picks the channels worth your hours.
How to use a free Claude skill that maps the features and functions of your vibe-coded app so you can find what broke.
How to build a daily personal news brief with Claude that reads your news, social feeds, newsletters and podcasts each morning.
How a non-technical founder, Nomiki, got 950 women on a waitlist in 16 weeks by validating fast and shipping before she was ready.
How Alisha Gupta built a 1,500-person community for creatives without quitting her job, by listening before building and owning a niche.
How Jen Yip has bootstrapped Lunch Money alone for seven years, and why clarity on her ideal users and focus are her real edge.