The Startup Ideas Podcast

I'm joined again by Imran Muthuvappa to walk through how to build your own AI Chief of Staff using a tool called Nebula. Imran shows me how to spin up specialized agents that handle the work a real chief of staff would do — surfacing team blockers, tracking project status, holding people accountable to offsite vision goals, running daily agenda briefings, and prospecting ICP leads. We also get into mini apps, model selection for cost efficiency, and why personal software is becoming a real category. By the end, the takeaway is clear: every role now has a "work on the job" component where you supervise yourself and offload tasks to agents.

Links Mentioned:

Try Nebula: https://startup-ideas-pod.link/nebula

Precall Agent: https://startup-ideas-pod.link/precall-agent

Project Status Agent: https://startup-ideas-pod.link/project-status-agent

Lead Gen Agent: https://startup-ideas-pod.link/lead-gen-agent

Timestamps:

00:00 – Intro

02:13 – What is Nebula

02:54 – What an AI Chief of Staff Actually Does

4:26 – Nebula vs OpenClaw vs Hermes

06:09 – Building the Blockage Radar Agent

09:47 – Agent Features

12:04 – Choosing Cheaper Models for Simple Tasks

13:15 – Building Project Status Agent

13:53 – Connecting Tools to Agents

17:38 – Building Vision Tracker Agent

22:07 – Mini Apps and Personal Software as a New Paradigm

25:13 – Building a Daily Agenda Agent and Second Brain Integration

30:25 – Hours Saved vs. Anxiety Reduced for Founders

33:02 – Building the Lead Gen Prospector Agent

39:19 – Final Thoughts: Automate Three to Five Things

Key Points

An AI Chief of Staff handles the boring executive support work — calendar, email, LinkedIn, project status — so a human can focus on decisions.

Nebula lets you build, deploy, and share custom agents through a Slack-like interface, where each agent has its own goals, tools, and system prompt.

Voice input via SuperWhisper or WhisperFlow gets you to roughly 150 words per minute, which Imran calls the biggest productivity lift available right now.

Cheaper models like the Nebula model handle most chief-of-staff tasks well — reserve frontier models like Opus or Sonnet for deep coding or reasoning work.

Mini apps inside Nebula are spinnable web dashboards that connect back to your agents — a glimpse of personal software replacing off-the-shelf tools.

The new skill is judgment: picking which three to five things to automate out of your week.

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/

FIND IMRAN ON SOCIAL

X/Twitter: https://x.com/imranye

Alif: https://alif.build/

I sit down with Meng To for his second appearance on the pod to dig into design md, Google's newly open-sourced format for capturing the soul of a design and porting it across every medium and tool. Meng walks me through a live demo of how he uses design md alongside skills, HTML references, and tools like Aura, New Form, Codex, and OpenClaw to ship landing pages, motion design, slides, and mobile mocks that actually feel custom. We get into the design drift problem with one-shot prompts, why taste is the real moat for builders right now, and how he runs four products as effectively a team of one while iterating a thousand-plus prompts deep. If you build with agents and you want your work to stand out from the sea of purple-gradient lookalikes, this one is for you.

Timestamps

00:00 – Intro

04:00 – What design md actually is

07:17 – Examples: one design DNA across slides, promo videos, motion

09:31 – How to create design system

14:05 – The importance of taste and design

18:28 – Variant, remixing, and skills as ingredients

21:36 – Live demo: creating a landing page with design md and HTML

24:36 – Thoughts on Google Stitch

25:41 – Being fast and at edges is an unfair advantage

29:29 – Midjourney parallels and the queuing flow state

31:44 – Walking through skills (skeuomorphic, 3D, lasers)

34:07 – Now everyone is a designer

36:47 – The full design workflow

38:50 – Iteration versus remix

39:24 – Judgment per minute as the new craft

41:06 – Solo building vs building a team

44:34 – Taste is the moat

48:25 – Building a second brain for design inspiration

50:41 – Closing thoughts

Key Points

Design md is a portable blueprint for typography, color, spacing, and effects that you attach to any prompt to keep design consistent across web, mobile, slides, and motion.

One-shot prompts collapse on page two; a design system carries the soul across every medium and tool you switch into.

Skills work like ingredients (lasers, skeuomorphic, 3D, copywriting), and stacking them on top of design md  is what separates custom work from generic vibe-coded output.

Taste is the real moat right now, and you build it by surrounding yourself with great design and using every product in your niche.

Iteration (90% of the time) keeps a product evolving; remix (10%) takes the same DNA into a new medium or category.

The shift in craft is from moving pixels to making judgment calls per minute, with agents handling the mechanical work.

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/

FIND MENG ON SOCIAL

Aura: https://aura.build

X/Twitter: https://x.com/MengTo

If you want more workflows and tactics to build a business with AI, check out this free workshop: https://www.ideabrowser.com/workshop

I sit down with Andrew Wilkinson and we go deep on how he's restructured his work, his health, and his family office around AI agents. Andrew walks me through Deep Personality (an app he vibe-coded after running psychological screens on himself and his girlfriend), the autonomous SaaS business he runs through agent harnesses like Harbor, and the vector-database setup that lets him query Tiny and his personal holding company like an oracle. We cover where software is headed, why he's pouring capital into TSMC and data center stocks, and the daily AI workflows he's built around health, email triage, and a personalized morning podcast. Listeners walk away with concrete prompting tactics, agent architectures, and a frank read on where the moats are moving.

Timestamps:

00:00 – Intro

01:50 – The OpenClaw Unlock

04:53 – Demo: Deep Personality App

10:38 – Harbor: An Agent Harness For Real Companies

12:30 – Autonomous Companies: Hype Vs. Reality

17:30 – Credibility As The Missing Layer For Vibe-Coded Products

20:14 – Centralizing Data Pipelines

21:35 – Vector Databases

23:22 – Transitioning Companies to Agentic Companies

25:22 – Where Andrew Would Build Today

27:10 – The New Interface

28:21 – Why build now

30:59 – Replacing Adapar: A Networth Wealth Platform

33:29 – Services As The New Software

35:46 – G-Brain Explained and Andrew’s OpenClaws

45:31 – Closing Thoughts

Key Points

Andrew runs a SaaS business called Deep Personality almost entirely through agents, generating roughly $20K of revenue while debugging eats half his time.

Harbor (github.com/geekforbrains/Harbor) gives agents a GUI-style harness — dev, marketing, and support agents that can autonomously merge PRs and adjust ad budgets across PostHog, Meta, and Reddit.

Andrew's family office swapped headcount for a $40K/month Claude bill; his CFO, who had zero coding background, vibe-coded a replacement for Adapar (priced at $50K–$100K/year) in about two weeks.

Vector databases trained on Tiny and Andrew's holding company let him query 132 minority investments, P&Ls, and headcount data conversationally.

For builders today, Andrew suggests aiming for a $1M–$2M product, then parking gains in TSMC and data center exposure given how fast software moats are eroding.

His best prompting tip: ask the model to interview you with multiple-choice questions before generating any output.

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/

FIND ANDREW ON SOCIAL

X/Twitter: https://x.com/awilkinson

Deep Personality: https://deeppersonality.app

Tiny: https://www.tiny.com

Limited BONUS: First 1,000 builders get $1,000. Claim yours while supplies lasts.: https://startup-ideas-pod.link/hyperagent

I sit down with Howie Liu, co-founder and CEO of Airtable, to talk about the agent economy and the launch of HyperAgent. We walk through Sequoia's charts on AI agent deployment, the economics of token-based work versus human labor, and why frontier agents have crossed a threshold that changes how companies get built. Howie then does a live show-and-tell of HyperAgent, including a custom "Greg Isenberg contrarian AI" skill he spins up in real time. This one is for anyone building a solopreneur business, operating a fleet of agents, or trying to figure out where to place their bet in the agent ecosystem

Timestamps

00:00 – Intro

02:22 – Sequoia's AI agent deployment chart reaction

04:41 – Copilot vs Autopilot territory and the $1T+ opportunity

08:13 – Agent economics vs human labor costs

11:12 – Fastest enterprise adoption curve in history

14:48 – The agent command center and fleet of 20 agents

18:03 – What is HyperAgent?

19:43 – Live demo: hyperlocal real estate market reports

22:38 – HyperAgent as the founder, not just the developer

23:21 – Street View, Zillow redesigns, and visual tool power

24:15 – Command center view across a fleet of agents

25:48 – Skills as the key primitive for frontier agents

26:30 – Building the Greg Isenberg contrarian AI skill live

32:31 – HyperAgent vs Perplexity Computer, Manus, OpenClaw, Codex

34:52 – Reviewing writing skill

36:55 – The arbitrage of persistence

41:31 – Confidence milestones: first dollar, $10K/month

35:27 – Reviewing contrarian tweet drafts live

45:05 – Giving the agent feedback and building rubrics

50:15 – Connectors, OAuth, and building custom API skills

53:03 – How to get started with HyperAgent

01:01:54 – Credit giveaway for listeners

01:03:31 – Closing Thoughts

Key Points

Frontier agents have crossed a threshold in the last 4–5 months where they function as true autonomous coworkers, not just chat assistants.

Reframe agent cost by value delivered: a $150 token spend for a board memo beats hours of human time, so anchor on opportunity cost.

The real arbitrage is persistence: 99% of people quit after one shot, while daily practice for 30/60/90 days produces top 1% operators.

Skills are the most important primitive in frontier agents, turning generally intelligent models into domain experts through playbooks.

HyperAgent's differentiation is a low floor plus a high ceiling, with rubrics, LLM-as-judge evals, and fleet-wide observability for scaling.

Aim for $100B companies with under 5 employees, built on fleets of always-on agents mapped to human job roles.

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/

FIND HOWIE ON SOCIAL

X/Twitter: https://x.com/howietl

Hyperagent: https://www.hyperagent.com

Airtable: https://www.airtable.com-

I sit down with Riley Brown to get a hands-on tour of OpenAI's Codex, which he argues is the most powerful single interface for using AI agents today. Riley walks me through how Codex unifies vibe coding, knowledge work, browser use, computer use, and automations into one app, all running on GPT 5.5. I come in as a complete Codex skeptic who has spent most of my time in Claude Code, and Riley shows me skills, plugins, projects, Remotion, Chronicle, and the in-app browser to make his case. By the end, the question becomes whether the era of separate tools for documents, decks, code, and research is collapsing into a single super app.

00:00 – Intro

03:23 – What is Codex

06:46 – Why a GUI beats the terminal for most users

10:13 – Codex: the all in one platform

12:48 – Atlas browser inside Codex

14:21 – Remotion explained and motion graphics workflows

19:28 – Computer use and Chronicle

22:26 – Plugins, skills, MCPs, and integrations

31:57 – Evals, examples, and good outpu

38:43 – Hard questions: who Codex is built for

40:44 – Browser use plays itself in chess

43:20 – Running Claude Code inside Codex

45:58 – GPT 5.5 cost and effort settings

48:50 – GPT Images 2.0

54:09 – Why most people feel overwhelmed by AI tools

57:09 – Three projects to start with on day one and Closing thoughts

Key Points

Codex is positioned as a super app where coding, documents, decks, research, and automations live in one interface, with GPT 5.5 as the underlying model.

The trend across Codex, Cursor, and the Claude Code desktop app is the same GUI pattern: chats on the left, agent in the middle, output on the right.

Plugins offer official integrations like Slack, Notion, Sheets, Remotion, and Canva, while skills are user-created instructions stored as a SKILL.md file.

Computer use and browser use have crossed a speed threshold; the chess demo runs at near-human pace, a leap from earlier "dial-up" feeling agents.

Running Claude Code inside the Codex terminal lets you stack both subscriptions and use each model where it shines.

The biggest unlock for companies is collecting good examples of finished work so agents can match the bar.

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/

FIND RILEY ON SOCIAL

X/Twitter: https://x.com/rileybrown

Vibe Code App: https://www.vibecodeapp.com

Youtube: https://www.youtube.com/@rileybrownai/videos