The Startup Ideas Podcast

An AI company gets built live, complete with a CEO, engineer, QA, video editor, and content strategist. The conversation explores agent memory, heartbeat checklists, skills, and QA loops. It also gets into token budgets, recurring routines, shareable AI companies, and the big question of whether agent orgs actually work.

A fast tour of why clean web data is becoming core infrastructure for AI. It digs into the AI agent stack, Firecrawl’s scraping and browser-control tools, and a wave of niche business ideas like price trackers, SEO tools, job boards, research products, and data services. It also explores the wild idea of companies hiring AI agents as employees.

Moritz Kremb, a Berlin AI workflow builder and OpenClaw power user, shares the setup that finally made the tool click. He walks through memory fixes, model choices, heartbeat tuning, browser modes, and security basics. Then he shows two standout builds: a no-slop short-form video system and a conversational CRM wired into everyday tools.

Remy Gaskell, an AI educator and builder focused on practical automation, breaks down how beginners can build AI agents for real business roles. He explores agent loops, context files, memory, tool connections, and reusable skills. There is a live build of an executive assistant, plus a tour of beginner-friendly platforms and how simple workflows can grow into department-level systems.

A new open-source AI tool gets unpacked through loops, experiments, and self-improving code. The conversation explores 10 business angles, from niche agent products and nonstop marketing tests to research services, finance automation, and sales workflows. It also touches on Agent Hub, early tech hype, and a practical path to getting started with Claude Code and Colab.

Oliver Henry, builder and AI practitioner who built Larry OpenClaw, explains how an autonomous agent runs TikTok slideshow marketing to drive app downloads and MRR. He describes the Larry Loop: creating content, reading analytics, iterating hooks and CTAs, and packaging the system as a shareable skill so others can replicate his automated marketing workflow.

A 30-step playbook for building modern SaaS using AI agents, media, and tight niche focus. Tips on mapping customer workflows and spotting repeatable, monetizable tasks. Advice on starting as a service, automating mechanical work with agents, and building a media flywheel alongside product development. Strategies for pricing, onboarding, and scaling by reinvesting and hiring from the niche.

Cody Schneider, growth engineer and co-founder of Graphed, who builds agent-driven GTM systems. He demos running multiple AI agents for LinkedIn auto-response, bulk Facebook ad generation, cold-email pipelines, Notion automation, and live data-driven ad optimization. Short, hands-on walkthroughs show how to wire tools, spin up ephemeral infra, and run autonomous marketing loops.

A founder tests an AI agent across five money-making workflows. Live warm outreach is automated and personalized at scale. Daily competitive intelligence and recurring monitoring run on autopilot. A 50-VC pipeline is built for Series A targeting. A rapid investment memo and content repurposing flow are created in one session.

Vin (Internet Vin), a creator who builds personal productivity and agent workflows, walks through using Obsidian and Claude Code as a personal OS and thinking partner. He demos feeding a vault to an LLM, custom slash commands, tracing how ideas evolve, and connecting vault domains. The conversation covers daily reflection, autonomy via agents, and privacy versus power trade-offs.