Dear Ben

Andrew Warner asked: "Every time I talk to a founder, I hear about some AI workflow I never would have thought to try. What are the non-obvious ones you're seeing?"Most people use AI to make stuff. The overlooked category: using AI to mirror things back to you. Scoring. Flagging. Routing. That's where the leverage is hiding.Ask your own question at DearBen.ai.--# SHOW NOTES- The Make vs Mirror Matrix: you're probably doing Make — but Mirror is where the blind spots live- Jeff Brewer gets rated on clarity, decisiveness, and interruptions after every single meeting- Dan Shipper built a system that flags when he's avoiding confrontation- Wade Foster (Zapier CEO) scores team meetings against Five Dysfunctions and sends individual feedback- Pete Sena's mega prompt analyzed 13 years of my notes and surfaced patterns I'd forgotten- Chris Brownridge routes meeting actions to Linear/calendar automatically via Claude agent- One meeting. One mirror workflow. Ship it. Stack from there.--# TOOLS MENTIONED- Motion AI Exec Coach (meeting scoring)- Granola (local-first meeting notes)- Zapier (automation)- Fireflies (meeting transcription)- Claude Desktop / Claude Code (local file analysis)- Linear (task routing)--# ABOUT DEARBEN.AIPractical AI for CEOs who don't have time to become AI experts.Ben Fisher — 2x CEO, 3x CTO.A founder-to-founder experiment answering real AI questions from operators running real companies.Ask yours at https://DearBen.ai--# CONNECTWebsite: https://DearBen.aiLinkedIn: https://fshr.co/profileX: https://x.com/skinnyandbaldHosted on Ausha. See ausha.co/privacy-policy for more information.

Work happens in conversations. The problem is it stays there.You hash out a product idea, talk through requirements, make decisions — and then somehow you're supposed to recreate all of that in a document later. That's where everything falls apart.I use AI to extract the work that already happened and turn it into artifacts that keep the momentum going. PRDs, user stories, requirements — pulled straight from the conversation.Ask your own question at DearBen.ai.--# SHOW NOTES• Work happens in conversations — but it stays trapped there• The "missing middle" — decisions made verbally rarely become documents• Extract the work, don't recreate it — AI pulls out what you already figured out• Zero-draft artifacts — PRDs, user stories, requirements from the conversation itself• Pressure-test with clarifying questions — surface gaps before you build• 95% of the work already done — you just need to extract it--# TOOLS MENTIONED• Fireflies.ai — Meeting recording and transcription• Claude Desktop — Extraction and synthesis• Claude Code — Workflow automation• claude-code-requirements-builder — Requirements extraction workflow--# ABOUT DEARBEN.AIPractical AI for CEOs who don't have time to become AI experts.Ben Fisher — 2× CEO, 3× CTO.A founder-to-founder experiment answering real AI questions from operators running real companies.Ask yours at https://DearBen.ai--# CONNECTWebsite: https://DearBen.aiLinkedIn: https://fshr.co/profileX: https://x.com/skinnyandbaldHosted on Ausha. See ausha.co/privacy-policy for more information.

This isn't a Q&A episode — it's for builders who want to see how I actually work.I walk through using Claude Code and Every's skill creator to build the system that helps me prepare for Dear Ben episodes. The process: do it manually first, pay attention to what you repeat, then systematize.Ask your own question at DearBen.ai.--# SHOW NOTES• Do it manually first, then systematize — I did this process by hand several times before building the automation• AI is probabilistic, not deterministic — the agent built the skill but forgot to create the slash command• ChatGPT's Prompt Optimizer genuinely improves prompt efficacy, even for Claude• The LLM asking questions extracts your thinking — it's a brainstorming tool that forces you to articulate what you believe• Four phases: Ideation, Content Development, Proof of Concept, Blind Spots--# TOOLS MENTIONED• Claude Code• ChatGPT Prompt Optimizer• Every's Skill Creator Agent--# ABOUT DEARBEN.AIPractical AI for CEOs who don't have time to become AI experts.Ben Fisher — 2× CEO, 3× CTO.A founder-to-founder experiment answering real AI questions from operators running real companies.Ask yours at https://DearBen.ai--# CONNECTWebsite: https://DearBen.aiLinkedIn: https://fshr.co/profileX: https://x.com/skinnyandbaldHosted on Ausha. See ausha.co/privacy-policy for more information.

Murali asked: "I run a 600-person software dev agency. How do I index our sales material so my team can actually find what they need?"Before jumping to AI, ask: is this a process problem or a familiarity problem? When the people creating docs aren't the ones searching, semantic search creates real leverage.Ask your own question at DearBen.ai.--# SHOW NOTES• First question: is this a structure problem? At 600 people with creators ≠ searchers, it's not• Lindy gives full visibility into prompts — you can tweak and experiment, not just hope it works• The agent builder gets you 85% there; you course correct the rest• Use a smaller model for retrieval — it's not reasoning, save budget for generation• 5-minute setup: Google Drive → Lindy → Slack integration• Future extension: calendar triggers that auto-fetch docs before meetings--# TOOLS MENTIONED• Lindy AI — automation platform with prompt visibility and agent builder• Fireflies — AI note-taker with real-time knowledge base features (bonus)--# ABOUT DEARBEN.AIPractical AI for CEOs who don't have time to become AI experts.Ben Fisher — 2× CEO, 3× CTO.A founder-to-founder experiment answering real AI questions from operators running real companies.Ask yours at https://DearBen.ai--# CONNECTWebsite: https://DearBen.aiLinkedIn: https://fshr.co/profileX: https://x.com/skinnyandbaldHosted on Ausha. See ausha.co/privacy-policy for more information.

Henry asked: "What's your best practice advice for winning AI search? Nobody really knows how to game the system."There's a lot of speculation presented as fact in this space. But practitioners who've been deep in SEO are starting to converge on real patterns. I brought in Chirag from Taco to break down what's actually working.Ask your own question at DearBen.ai.---# SHOW NOTES• LLMs run web searches behind the scenes — if you don't rank on Google, you probably won't rank in ChatGPT• Turn keywords into questions: "dandruff shampoo" becomes "what are the best dandruff shampoos for men?"• Identify where LLMs pull citations from — Reddit, listicles, authoritative publications in your vertical• Create content on third-party sites that LLMs already trust (Reddit threads, guest posts, affiliate sites)• Spam tactics get plugged fast — same as traditional SEO• Some startups have leapfrogged traditional SEO for LLM placement (Lenny's podcast goes deeper)--# FEATURINGChirag Kulkarni — Founder & CEO, Taco (AI-powered SEO agency)Forbes 30 Under 30 (Marketing). Former CMO at Medly Pharmacy (scaled 1→50 pharmacies, $100M+ revenue).• Website: https://taco.co/• LinkedIn: https://www.linkedin.com/in/chiragkulkarni/--# TOOLS MENTIONED• Scrunch• Air Ops• Profound--# ABOUT DEARBEN.AIPractical AI for CEOs who don't have time to become AI experts.Ben Fisher — 2× CEO, 3× CTO.A founder-to-founder experiment answering real AI questions from operators running real companies.Ask yours at https://DearBen.ai--# CONNECTWebsite: https://DearBen.aiLinkedIn: https://fshr.co/profileX: https://x.com/skinnyandbaldHosted on Ausha. See ausha.co/privacy-policy for more information.

Part 2 of the behind-the-scenes series. I take Claude's workflow spec and build it in Lindy — and debug it live when things break.The lesson: check templates first before building from scratch. Lindy has out-of-the-box knowledge base templates I could have started with.Ask your own question at DearBen.ai.--# SHOW NOTES• Check templates first — Lindy has out-of-the-box knowledge base templates I should have used• Debug in public — Google connections drop, folder IDs get lost, agents hang• Multiple approaches exist — agent-built workflow vs. template, different tradeoffs• Token efficiency matters — different workflow structures have different credit consumption• Use Claude to generate test data — I had it create 10 fake sales documents for the demo--# TOOLS MENTIONED• Lindy AI (agent builder, knowledge base templates)• Google Drive integration• Slack integration--# ABOUT DEARBEN.AIPractical AI for CEOs who don't have time to become AI experts.Ben Fisher — 2× CEO, 3× CTO.A founder-to-founder experiment answering real AI questions from operators running real companies.Ask yours at https://DearBen.ai--# CONNECTWebsite: https://DearBen.aiLinkedIn: https://fshr.co/profileX: https://x.com/skinnyandbaldHosted on Ausha. See ausha.co/privacy-policy for more information.

Jason asked: "How do you manage junior developers when they're using AI to build systems more sophisticated than they understand?"The solution isn't more oversight — it's constraints that force understanding. I share three practices that work together.Ask your own question at DearBen.ai.---# SHOW NOTES• Spec-driven development: write what the system should do in plain language before any code• Opinionated frameworks: use Next.js or T3 stack to prevent AI from inventing architecture• Test-driven validation: understand what each test proves, not just that tests exist• Why these three rules create the same constraint: forced understanding• Bonus: invest in a fractional coach for code reviews and AI workflow guidance--# TOOLS MENTIONED• Next.js• T3 Stack--# ABOUT DEARBEN.AIPractical AI for CEOs who don't have time to become AI experts.Ben Fisher — 2× CEO, 3× CTO.A founder-to-founder experiment answering real AI questions from operators running real companies.Ask yours at https://DearBen.ai--# CONNECTWebsite: https://DearBen.aiLinkedIn: https://fshr.co/profileX: https://x.com/skinnyandbaldHosted on Ausha. See ausha.co/privacy-policy for more information.

Jeff asked: "Can AI generate my content, let me approve it in Slack, then auto-post to LinkedIn, X, and Reddit?"The orchestration is easy — scheduling, APIs, multi-platform posting. That's solved. The hard part is content quality. Getting AI output that doesn't sound like AI.Ask your own question at DearBen.ai.---# SHOW NOTES• Why orchestration (scheduling, APIs) is the easy part — every platform handles that• The real bottleneck: content quality and voice consistency• AI writing tools vs. automation platforms: black boxes vs. build-it-yourself• Lindy as middle ground: prompt visibility + Slack review + workflow automation• 15-minute proof-of-concept walkthrough with Claude Sonnet 3.5• Year one cost: ~$2,100 (vs. 260 hours of manual content creation)--# TOOLS MENTIONED• Lindy.ai• Zapier.com / Make.com--# ABOUT DEARBEN.AIPractical AI for CEOs who don't have time to become AI experts.Ben Fisher — 2× CEO, 3× CTO.A founder-to-founder experiment answering real AI questions from operators running real companies.Ask yours at https://DearBen.ai--# CONNECTWebsite: https://DearBen.aiLinkedIn: https://fshr.co/profileX: https://x.com/skinnyandbaldHosted on Ausha. See ausha.co/privacy-policy for more information.