Semir Jahic runs Salesmotion with his CTO, José. Two people, fully remote, bootstrapped for two and a half years. He's in Switzerland, José is in Spain, and between them they're building a sales intelligence platform, serving enterprise customers, shipping product, running marketing, and handling everything else a company needs to function.
I came across Semir when he commented on a LinkedIn post I wrote about building with Claude Code. He said he'd love to show me what he was doing. I took him up on it, spent an hour watching him walk through his setup, and immediately asked if we could record it properly for the show.
What struck me wasn't any single tool or automation. It was the mindset underneath all of it. Every time he finds himself doing something twice, the question becomes: can AI do this? Not eventually, not after a proper scoping exercise, but now, this week, with a few prompts.
Since February, when Claude Opus came out, he's rebuilt the Salesmotion website entirely from scratch using Cursor and a GitHub-to-Vercel pipeline he taught himself as he went. He's never touched it manually since. Blog posts are generated automatically, pulling from CRM data, Granola call recordings, and customer emails, then enriched with approved customer quotes from a folder in the GitHub repo. He barely reads them before they go out. Traffic has gone up substantially since February, and he's now getting inbound signups through ChatGPT and Claude search that he simply didn't have before.
He built a knowledge base by pointing Claude at his support emails, call transcripts, and the product codebase, and telling it to write the articles. He hasn't written a single one himself. When someone abandons a signup flow on the Salesmotion website, he gets a Slack notification built with a few prompts, and a two-step automated follow-up email goes out. A tool that does this off the shelf costs around $30 a month. He built it himself and owns it completely.
The product side works the same way. When he has an idea from a customer conversation, he prototypes it directly in a branch of the app, shows José what he means visually, and hands it over. What used to be a two-week loop of explaining, building, reviewing, and realigning now happens in an afternoon. They ship at a pace that would have been impossible six months ago.
He was honest about the anxiety that comes with all of this. The moat question keeps him up at night. Anyone could theoretically build what Salesmotion does. His answer to that is focus on customers, stay lean enough to adapt, and don't confuse being VC-backed with being safe. Several better-funded competitors in his space have already shut down. Salesmotion is still going.
He also flagged something I haven't heard many people say clearly: AI costs may not stay where they are. As the big model companies move toward consumption pricing, the founders who've been thoughtful about efficiency from the start will be in a better position than those who've built on the assumption that Claude Opus 4 for every email rewrite will always cost what it costs today.
Where I'm taking this personally: Semir is the first guest in what I want to build into a recurring thread in the newsletter and on the show. Lean teams, AI-first operations, founders doing more with fewer people than anyone thought was possible two years ago. If you're running a small team and doing interesting things with AI, I want to talk to you.
The full conversation is on The AI Revolution Show. Subscribe on Apple Podcasts:
Alex
Founder, Shift AI
P.S. Your 16 most important meetings of the year will happen at Shift AI Europe - with the founder six months ahead of you, the investor who has seen it all, the operator who solved it last quarter. Standard tickets now on sale.
The Shift · Published every Tuesday · shiftai.events
You're receiving this because you subscribed. Unsubscribe any time.