OpenAI DevDay: Dots, Agents & $100B Opportunities

Google's most advanced audio models are LIVE, try them for yourself: Gemini 3.8 Live: https://startup-ideas-pod.link/gemini-3.8-live Gemini 3.5 Transcribe https://startup-ideas-pod.link/gemini-3.5-transcribe Gemini 3.5 Live Translate: https://startup-ideas-pod.link/gemini-3.5-live-translate In this episode, I talk with Vas from Varick about what it takes to put AI to work inside a real company. Vas makes the case that AI pays off through process reengineering, and he walks me through the exact method his forward deployed engineers (FDEs) use: interviews, process mining, step sorting, and agents built inside existing systems of record. We go through real engagements, including a $5B public software company and an accounts payable overhaul that cut the cost per invoice from $31 to $6. By the end, you get a clear picture of the FDE role, the business opportunity behind AI roll-ups, and a five-day plan to start on your own. Links Mentioned: FDE Presentation: https://startup-ideas-pod.link/FDE-slides Vas’s Article: https://startup-ideas-pod.link/vas-fde Timestamps 00:00 – Intro 01:31 – Sponsor: Google 03:58 – FDE Overview 05:02 – AI Roll-Ups and Process Reengineering 08:04 – The Personal Systems Analogy 09:55 – Understanding a company’s process (step-by-step) 14:11 – Case Study: $5B Software Company 18:11 – 4 Buckets for Every Step 19:04 – Build Inside Systems of Record 21:36 – Case Study: PE Portfolio 23:43 – Selling to C-Suite Executives 27:07 – Process of Mapping Five NetSuite Companies 28:39 – Example: Accounts Payable Process Map 32:54 – Case Study: 60-Person Accounting Firm 34:51 – When to Use Code, Agents, or Humans 36:00 – Choosing AI Models 38:16 – Sidekick vs Background Agents 40:29 – The 3 Skills of a Top FDE 42:25 – Why FDEs Earn So Much 45:26 – Five-Day Starter Plan 47:39 – On-Premise Hardware Demand 48:36 – OpenAI Private Intelligence 50:20 – The Full Playbook 52:01 – Closing Thoughts Key Points AI pays off when you re-engineer the process first, then build agents into it. Map the real process with interviews, system-of-record mining, and existing docs. Sort every step into four buckets: delete, plain code, agent, or human decision. Build agents inside the tools clients already use, like Salesforce, NetSuite, and Slack. Sell the outcome each buyer cares about, and prove it with before-and-after KPIs. Top FDEs combine domain knowledge, production engineering, AI judgment, and strong communication. The #1 tool to find startup ideas/trends - https://www.ideabrowser.com LCA helps Fortune 500s and fast-growing startups build their future - from Warner Music to Fortnite to Dropbox. We turn 'what if' into reality with AI, apps, and next-gen products https://latecheckout.agency/ FIND ME ON SOCIAL X/Twitter: https://twitter.com/gregisenberg Instagram: https://instagram.com/gregisenberg/ LinkedIn: https://www.linkedin.com/in/gisenberg/ FIND VAS ON SOCIAL Varick Agents: https://www.varickagents.com/#hero-section X/Twitter: https://x.com/vasuman AI Forward Deployed Engineers: https://learn.varickagents.com/fde-in-30-days

I watched almost 60 minutes of Sam Altman on stage at OpenAI Dev Day 2026. Out of 20-plus launches, I pick the three or four that I think can make people billions of dollars in aggregate. I break down Dots, OpenAI's personal agent platform, plus the Decisions API, the Agents API with computer use, and Sign in with ChatGPT. I also share my four-step framework for building in this world and two business ideas I hope someone takes. If you want to build a business and make money around AI, this episode is for you. Timestamps 00:00 – Intro 01:48 – Dots, the Personal Agent Platform 03:29 – OpenAI Doubles Down on Plugins 05:02 – Decisions API 06:43 – Agents API With Computer Use 08:01 – Sign In With ChatGPT 12:36 – Where should you build? 13:46 – Idea: Real-World Work APIs 15:04 – Idea: Analytics for Agent Discovery 17:05 –Closing Thoughts Key Points Dots gives plugin builders a front door to 1.2 billion weekly active users. OpenAI is doubling down on plugins to make ChatGPT the app ecosystem for AI. Sign in with ChatGPT lets users bring their current plan, so a free core app with paid upsells now works. Workflows too niche for OpenAI to build become viable businesses. The strongest businesses in this world own the trigger, the action, and the feedback. Two open opportunities: real-world work APIs and an analytics layer for agent discovery The #1 tool to find startup ideas/trends - https://www.ideabrowser.com LCA helps Fortune 500s and fast-growing startups build their future - from Warner Music to Fortnite to Dropbox. We turn 'what if' into reality with AI, apps, and next-gen products https://latecheckout.agency/ FIND ME ON SOCIAL X/Twitter: https://twitter.com/gregisenberg Instagram: https://instagram.com/gregisenberg/ LinkedIn: https://www.linkedin.com/in/gisenberg/

In this solo episode, I break down the $5 trillion wave of small businesses set to change hands as their owners retire, and why AI agents make this wave a real opening for solo founders. I walk through how Thrive Holdings and General Catalyst buy accounting firms, property managers, and support centers, then run AI agents inside them to lift margins. Then I show how I'd run a one-person holding company: the folder structure, the agent files, the human approval rule, and my average week. I close with the strongest arguments against AI roll-ups, including the one I take most seriously. Timestamps 00:00 – Intro 01:52 – Why the $5 Trillion Shift Is Happening Now 04:52 – Examples: Thrive Holdings& General Catalyst 08:38 – The Fund Playbook 10:07 – The Small-Deal Gap 10:51 – The One-Person Holdco 14:23 – The Folder Structure 16:47 – The Agent Pipeline 18:38 – Inside a Reviewer Agent File 19:41 – My Week Running the Holdco 21:49 – How to Land the First Business 22:24 – Arguments Against AI Roll-Ups 27:43 – Closing Thoughts Key Points About a million small businesses, part of a $5 trillion shift, are set to sell by 2035 as owners retire (McKinsey). AI agents now handle the work these firms run on: data entry, document chasing, status updates, and first drafts. The big funds chase bigger deals, which leaves the small firms open for solo founders and small teams. A one-person holdco runs on shared agents and rules, plus a GM with real upside at each business. A person approves all agent work before it reaches a client. The corrections log, turned into rules every week, becomes the most valuable asset in the holdco. The #1 tool to find startup ideas/trends - https://www.ideabrowser.com LCA helps Fortune 500s and fast-growing startups build their future - from Warner Music to Fortnite to Dropbox. We turn 'what if' into reality with AI, apps, and next-gen products https://latecheckout.agency/ FIND ME ON SOCIAL X/Twitter: https://twitter.com/gregisenberg Instagram: https://instagram.com/gregisenberg/ LinkedIn: https://www.linkedin.com/in/gisenberg/

In this solo episode, I break down the huge business opportunity because Meta has opened Muse, its personal AI agent, to developers. You can now submit a connector, which lets Muse use your service when someone asks it for help, and I think this could be the app store moment for AI. I explain how a connector works, share four startup ideas you can build on one, and cover how to get customers beyond Meta's directory. I also show how I'd build a first version with a coding agent like Claude Code or Codex, and what I'd test before submitting it to Meta for review. Timestamps 00:00 – Intro 01:35 – The App Store Parallel 04:55 – How a Muse Connector Works 07:29 – Startup Idea 1: Lead Gen for Business Suppliers 09:28 – Startup Idea 2: Home Repair Dispatch 11:07 – Startup Idea 3: Paddle Match and Court Finder 12:47 – Startup Idea 4: Family Dinner Planning 14:10 – How to pick an idea? 14:56 – How People Find Your Connector 15:32 – Growth Idea 1: Partner With Creators 16:17 – Growth Idea 2: Product-Led Sharing 17:46 – Growth Idea 3: Connected Marketplaces and Directory Placement 19:34 – Building the First Version 21:38 – Custom Connectors and Meta Approval 23:56 – Where to Start This Week 25:31 – Closing thoughts Muse Connector Prompt: https://startup-ideas-pod.link/muse-connector-prompt Key Points Meta has opened Muse to developers, and a connector lets Muse use your service when someone asks it for help. I look closely at the step in a request where someone needs a business that can deliver and money changes hands. With a small budget, I'd start with lead generation, because I can show a customer a sample before writing much software. I plan distribution around channels I can reach, such as creators, product-led sharing, and connected marketplaces. A coding agent can build the first version, and I still inspect the results and test the awkward requests myself. To start this week, I'd talk to one type of customer about the last time they dealt with the task. The #1 tool to find startup ideas/trends - https://www.ideabrowser.com LCA helps Fortune 500s and fast-growing startups build their future - from Warner Music to Fortnite to Dropbox. We turn 'what if' into reality with AI, apps, and next-gen products https://latecheckout.agency/ FIND ME ON SOCIAL X/Twitter: https://twitter.com/gregisenberg Instagram: https://instagram.com/gregisenberg/ LinkedIn: https://www.linkedin.com/in/gisenberg/

Get Your Complete Financial OS at https://startup-ideas-pod.link/brex_SIP In this episode I speak with Nicholas Cole about the real value of everything you write on the internet. Cole has 15 years of experience as a nonfiction writer and ghostwriter, and he runs the SaaS platform Typeshare. He gives me a simple model with three tiers of content: commodity, personality, and original. He also explains the digital brain, which he calls your personal language model, and he shows how AI repeats your approved language at scale. By the end, you know how to judge the short, medium, and long-term value of a piece before you write it. Start Writing Online In 30 Days: https://startup-ideas-pod.link/ship30 Timestamps 00:00:00 – Intro 00:03:36 – POV is the Moat 00:05:40 – Language As Open Source 00:11:24 – Value Is Relative To The Reader 00:14:26 – Approved Language 00:17:43 – The Value of AI in Writing 00:22:09 – Commodity Ideas 00:24:23 – How To Set Up The System 00:27:58 – Ownership IS Association 00:33:56 – Build your Personal Data Set 00:39:53 – Branded Content vs Founder-Led Content 00:45:17 – What is Original Content? 00:46:35 – Writing Versus Short-Form Video 00:48:58 – Three Types of Hooks 00:49:50 – Timely Content vs Timeless Content 00:53:27 – Voice is 3 dialed settings 00:56:32 – Finding your Voice 01:00:49 – The Company Brain As The Moat 01:07:30 – Closing Thoughts Key Points A point of view creates the moat, because copycats must wait for your next idea. Content sits in three tiers: commodity, personality, and original. Each tier adds different leverage. Ownership equals association. Volume builds it, and personality details make it strong. Your life story is the unmade data set, so AI learns it only from your own writing. Every piece sits on a spectrum from timely to timeless, so match your expectations to the type. Humans do the thinking and the writing. Robots do the repeating and the remixing. The #1 tool to find startup ideas/trends - https://www.ideabrowser.com LCA helps Fortune 500s and fast-growing startups build their future - from Warner Music to Fortnite to Dropbox. We turn 'what if' into reality with AI, apps, and next-gen products https://latecheckout.agency/ FIND ME ON SOCIAL X/Twitter: https://twitter.com/gregisenberg Instagram: https://instagram.com/gregisenberg/ LinkedIn: https://www.linkedin.com/in/gisenberg/ FIND COLE ON SOCIAL X: https://x.com/Nicolascole77 Youtube: https://www.youtube.com/@nicolascole77 Ship 30 for 30: https://startup-ideas-pod.link/ship30

In this episode, I talk with Ryan Vogel about Jev, a new type of AI built for classification. Ryan shows how Jev takes an input plus an output schema and returns a probability for each choice in about 200 milliseconds. He demos Jev sorting 1,700 emails for 18 cents total, then covers lead scoring, support routing, video clipping, and browser control. I push him on the startup angle: find a business with an expensive queue of incoming information and put Jev at the front of it. You leave with a clear mental model, real use cases, and a simple way to try it today. Links Mentioned: Jev/Typeface AI: https://typesafe.ai AI Gateway: https://vercel.com/ai-gateway Timestamps 00:00 – Intro 02:27 – What Jev Is and Why It Matters 04:32 – Email Triage Demo 07:19 – Jev as an AI Decision Maker 15:46 – How to Use Jev in a Business 20:48 – Startup Idea: Local Services Matching and Instant Quotes 22:51 – Use Case 1: Bitcoin Signal Test and Limits 24:03 – Use Case 2: Auto-Clipping Long Videos 25:27 – Use Case 3: Browser Control: Flight Pick in 7.1 Seconds 26:18 – How to Get Access 27:25 – Closing Thoughts Key Points Jev is a classifier: an input and an output schema go in, and a probability for each choice comes out. Ryan's demo scores 1,700 emails for 18 cents total. Each Jev query takes about 200 milliseconds, whatever the input and output structure. Use Jev at any point where a business makes fast, repeatable decisions on incoming data. Keep Jev in an advisory role, and save frontier models for high-intelligence tasks like trading. Instant access runs through the Vercel Gateway, and a waitlist covers direct access. The #1 tool to find startup ideas/trends - https://www.ideabrowser.com LCA helps Fortune 500s and fast-growing startups build their future - from Warner Music to Fortnite to Dropbox. We turn 'what if' into reality with AI, apps, and next-gen products https://latecheckout.agency/ FIND ME ON SOCIAL X/Twitter: https://twitter.com/gregisenberg Instagram: https://instagram.com/gregisenberg/ LinkedIn: https://www.linkedin.com/in/gisenberg/ FIND RYAN ON SOCIAL X: https://x.com/ryanvogel Youtube: https://www.youtube.com/@vogeldev/videos

I sit down with Remy to go through Instinct, the new invite-only personal agent that runs inside iMessage. Remy shares his raw chat history on screen: a haircut booking in Copenhagen, a restaurant reservation, a Bali visa on arrival, and an Emirates Skywards sign-up. We cover the parts that impress us, the points where the agent hits a wall, and the privacy questions that stay open. By the end of this episode you understand what Instinct does today, and you get fresh ideas for personal agents in general. Timestamps 00:00 – Intro 02:08 – Instinct Pros 11:18 – Instinct Cons 13:16 – Simple Onboarding 15:19 – Tools and Connectors 17:43 – Example 1: Booking a Haircut in Copenhagen 20:24 – Example 2: Restaurant Booking and Calendar Entry 23:03 – Example 3: Bali Visa on Arrival and Emirates Skywards 26:00 – Closing Thoughts Key Points Instinct hides the agent complexity behind a phone number and iMessage, so a first-time user starts in seconds. Remy gets it to book a haircut, hold a restaurant table, file a Bali visa on arrival, and open an Emirates Skywards account. A spend-limited virtual card keeps the blast radius small when the agent pays for things. The agent stalls when a task needs a phone app or an Indonesian checkout page. Users report that Instinct keeps copies of email after they disconnect Google, so treat privacy as an open risk. The trusted person network lets one Instinct talk to another, which builds network effects into the agentic era. The #1 tool to find startup ideas/trends - https://www.ideabrowser.com LCA helps Fortune 500s and fast-growing startups build their future - from Warner Music to Fortnite to Dropbox. We turn 'what if' into reality with AI, apps, and next-gen products https://latecheckout.agency/ FIND ME ON SOCIAL X/Twitter: https://twitter.com/gregisenberg Instagram: https://instagram.com/gregisenberg/ LinkedIn: https://www.linkedin.com/in/gisenberg/ FIND REMY ON SOCIAL X: https://x.com/remy_gaskell Youtube: https://www.youtube.com/@aiwithremy AI with Remy: https://www.aiwithremy.com/

Get Your Complete Financial OS at https://startup-ideas-pod.link/brex_SIP I welcome Ras Mic back to the pod to explain the phrase "software factory." Mic shares his screen and walks through the exact system that he runs today. His factory has four steps: isolate, build, prove, and ship. He keeps the whole system in five or six markdown files, so it works with any model and any harness. By the end of this episode, you can boot up your own factory, run many agents in parallel, and trust the code that comes back. Create your own Software Factory: https://startup-ideas-pod.link/ras-software-factory Timestamps 00:00 – Intro 02:17 – Software Factory Definition 03:44 – Why the Software Factory Matters 05:23 – Step 1: Isolate With Git Work Trees 11:34 – Step 2: Build With the Code Structure Skill 14:48 – Step 3: Prove With Evidence-Driven Testing 22:25 – Step 4: Ship With Grep Loop and Greptile 26:52 – The Physical Factory Analogy 29:21 – A Software Factory Is Markdown Files 30:02 – Closing Thoughts Key Points A software factory is a workflow of skills and domain knowledge, so it runs with any model and any harness. Isolate: every feature starts in a fresh git work tree branched from origin main, so each agent keeps its own station. Build: a code structure skill makes the agent write service layer code that a human developer can read. Prove: the agent records a before state and an after state as video, screenshots, or numbers. Ship: Greptile scores the PR, and the agent loops back to build until it earns five out of five. Mic runs up to 15 features in parallel and reviews the visual proof instead of the raw code. The #1 tool to find startup ideas/trends - https://www.ideabrowser.com LCA helps Fortune 500s and fast-growing startups build their future - from Warner Music to Fortnite to Dropbox. We turn 'what if' into reality with AI, apps, and next-gen products https://latecheckout.agency/ FIND ME ON SOCIAL X/Twitter: https://twitter.com/gregisenberg Instagram: https://instagram.com/gregisenberg/ LinkedIn: https://www.linkedin.com/in/gisenberg/ FIND MIC ON SOCIAL X/Twitter: https://x.com/Rasmic Youtube: https://www.youtube.com/@rasmic

I talk with Ras Mic about GPT-6 Astra. We skip the game demos and the 3D toys, and we focus on use cases to earn money or improve products. I share 9 Astra prompts that I posted publicly, and Greg Brockman reposted. Ras then shows his hardware project: he moved from a speaker idea to a parts list, a Blender layout, and merged code in about 30 minutes. The takeaway is simple: use this model for the ideas that felt too large for you last year. Timestamps 00:00 – Intro 01:53 – Astra Overview 04:14 – 9 Astra Prompts 11:48 – Jarvis Speaker Idea 16:21 – Think Bigger with Astra 18:29 – Vibe Coding to Vibe Manufacturing 21:16 – Closing Thoughts Key Points Astra costs more per task, and it uses fewer steps, so the value per dollar stays high. A performance audit moved one of Ras’s apps from 800 ms to 20–30 ms. A security audit on his live payments app found real risks in production. Ras went from a speaker idea to a $561 parts order and a merged pull request in about 30 minutes. Ras’s point: intelligence keeps climbing, and bravery stays flat. Ask for bigger things. The shift that vibe coding brought to software now reaches physical products. The #1 tool to find startup ideas/trends - https://www.ideabrowser.com LCA helps Fortune 500s and fast-growing startups build their future - from Warner Music to Fortnite to Dropbox. We turn 'what if' into reality with AI, apps, and next-gen products https://latecheckout.agency/ FIND ME ON SOCIAL X/Twitter: https://twitter.com/gregisenberg Instagram: https://instagram.com/gregisenberg/ LinkedIn: https://www.linkedin.com/in/gisenberg/ FIND MIC ON SOCIAL X/Twitter: https://x.com/Rasmic Youtube: https://www.youtube.com/@rasmic

I run this episode solo. I explain local AI in plain terms: the model runs on hardware I control, and a cloud model runs somewhere else. I map the four pieces of the local AI landscape — the model, the warehouse, the software, and the workflow — and I define the words that beginners meet first: parameters, tokens, context window, quantization, and GGUF. I walk through the Google open model stack (Gemma 4, Google AI Edge, LiteRT-LM, AI Edge Gallery), compare the other open model families, and show three ways to run a model today. I close with a first workflow you can copy and three startup ideas that use local AI as the wedge. And a special thank you to Google for supporting the podcast. Timestamps 00:00 – Intro 01:35 – The Open Model the Landscape 03:09 – Vocab Decoder 06:48 – Google Gemma Clearly Explained 10:29 – Other Open Model Families 14:20 – Path 1: Run Gemma in LM Studio 18:17 – Path 2: Ollama 20:15 – Path 3: Google AI Edge 21:07 – Hardware Cheat Sheet 21:52 – First Workflow to Build 22:47 – Workflows Before Fine-Tuning 25:06 – Local vs Cloud vs Hybrid Eval 26:33 – Framework for Local AI Startup Ideas 27:22 – Startup Idea 1: Home Health QA Reviewer 29:24 – Startup Idea 2: Offline Field Report Copilot 32:10 – Startup Idea 3: Pre-Send Reviewer for Professional Services 34:47 – Build Your Local AI Lab 37:55 – Closing Thoughts Key Points Ask whether the model is good enough for the job, and the business opportunities become clear. Local AI has four pieces: the model, the warehouse (Hugging Face), the software (LM Studio or Ollama), and the workflow you build around them. Gemma 4 E4B is my practical starting point; E2B fits phones and older machines. Hybrid architecture wins: local does the private first pass, cloud does the heavy reasoning, and a human approves anything important. Start with one repeated workflow — one folder, one model, one output — and run it 10 times. I see a 24-month window to build local-AI-native software for verticals that still run early-2000s tools. The #1 tool to find startup ideas/trends - https://www.ideabrowser.com LCA helps Fortune 500s and fast-growing startups build their future - from Warner Music to Fortnite to Dropbox. We turn 'what if' into reality with AI, apps, and next-gen products https://latecheckout.agency/ The Vibe Marketer - Resources for people into vibe marketing/marketing with AI: https://www.thevibemarketer.com/ FIND ME ON SOCIAL X/Twitter: https://twitter.com/gregisenberg Instagram: https://instagram.com/gregisenberg/ LinkedIn: https://www.linkedin.com/in/gisenberg/