Janet Maoo

Ex-VC & Startup Operator | Self-taught AI builder | Applying LLMs + agents for how I learn, research and work. USC24

[Janet Maoo @BuiltByJanet]

@NatA556W haha funny

[Janet Maoo @BuiltByJanet]

Grok Bots - my heads of department,
Claude Code - the engineers
Muse - chief of staff (more talk than action)

These are the three agents I use daily. Grok Bot has the best interface from my perspective: I can simply go back to the relevant bot and pick up where I left off instead of searching among tons of chats in Claude.
However, I love Claude Code’s deep research capability.
Muse is the most proactive, reminding me where each task stands and offering emotional support. The bots never seem to get this right. But with @reasonmachines , I can literally leverage the strengths of all three. I delegate tasks to the bots, bots tell Claude Code how to do it and have Muse watch over them, nudging and flattering me at just the right moments.

I can see a future where you just talk to your agent, tag her, and trust her to handle everything online.

[Janet Maoo @BuiltByJanet]

@joshua_xu_ @vercel Congrats! HeyGen's been on a roll. Claude Code + HyperFrames is my favorite combo lately. I used it to turn technical AI terms into a plain-language explainer video, and it finally stuck.

[Janet Maoo @BuiltByJanet]

@petergyang @bot My Grok Bot has basically become my chief of staff. It delegates tasks to my other agents.

[Janet Maoo @BuiltByJanet]

The analytical framework is:
1. What is it, and how does it work?
2. What problem does it solve?
3. How is it used, and should you build or buy it?
4. Value-chain analysis: who are the sellers and who are the buyers?
5. Who are the startups doing this? How?
6. Business model
7. How high is the barrier to entry?
8. Horizontal or vertical? (if applicable)

[Janet Maoo @BuiltByJanet]

As a non-technical person digging into AI startups, jargon is my biggest wall.

I'm always amazed how one "small" technical idea can power an entire company. But I'd learn a term, then forget it a week later.

So I had Claude Code make me explainer videos: plain language, real-life examples. Way easier to remember.

First one: embeddings 👇

[Janet Maoo @BuiltByJanet]

@MilksandMatcha Some of them are genuinely exceptional. But timing and luck are a huge part of it too, being in the right room in the right time. Cheer for them, just don't use them as your benchmark. Careers are long, and most great ones look slow for the first decade.

[Janet Maoo @BuiltByJanet]

@ChrisDaviesHQ Related. The shift from private building to public is underrated for solo founders. Agents make it possible to ship more, but the distribution and feedback still come from humans. I just did the same after overcoming imposter syndrome.

[Janet Maoo @BuiltByJanet]

The best two hours of my week had nothing to do with AI.

I was deep in AI research one afternoon when my friend called. A stray cat was hiding inside a car’s engine.

We couldn’t even see her. Animal control couldn’t help without the car owner, so we left a note on the windshield. Just as we were leaving, he showed up. He said the police had already tried to get her out.

While we waited, we bought two cans of cat food from the corner store, one fish, one chicken. She crept closer, but she was too scared and too quick.

The animal control officer came with a flashlight and a car jack and crawled halfway under the car.

Two hours later, we still couldn’t get her out.

I wish this story had a happier ending.

But I keep thinking about that afternoon. My friend and I dropped our work. The car owner was supposed to be judging his students’ architecture projects. He never made it. I asked if he’d be in trouble.

He just said, “This is life.”

For two hours, nobody was hustling. No deadlines, no AI, no anxiety about what’s next. Just strangers on a sidewalk, trying to help a scared little cat.

I needed that more than I knew.

[Janet Maoo @BuiltByJanet]

@zarazhangrui Building my personal website recently, right time for me, can't miss it.

[Janet Maoo @BuiltByJanet]

I love listening to podcasts, both for insights into startups and for leveling up my language skills. But summaries don’t work for me. I need to leave a trail of what I’ve actually learned.

So I built Podcast Dig, a skill that extracts the audio, transcribes it, and sends the transcript straight into Obsidian. Now I can take notes while I listen and build a knowledge base I come back to again and again.

Over time, my podcast collection has become a resource I keep at my fingertips. When I look up a word I don’t know, I can hear it used in a real conversation, which makes it much easier to remember. Interconnect with my Lexi Ai (the dictionary I built)

[Janet Maoo @BuiltByJanet]

@petergyang Nice. I’ve already got HyperFrames skills set up for motion. Does /video-short work cleanly alongside other video skills, or is it designed as a more standalone “raw recording → finished short” pipeline?

[Janet Maoo @BuiltByJanet]

@bykushnir This is the part most people skip. Building is cheap now, but validating whether anyone actually wants it still isn’t. Have you tried wiring an agent specifically for “kill criteria” — and does it actually change what gets built?

[Janet Maoo @BuiltByJanet]

@reasonmachines Using this already — it fixed the biggest friction I had with switching models.

[Janet Maoo @BuiltByJanet]

Really clear breakdown.

For solo founders or tiny teams, the bottleneck often isn’t the tools — it’s keeping the shared context clean as the agent starts generating more volume.

Have you seen teams succeed by treating the context layer almost like a product itself (versioned, reviewed, not just a dump of notes)?

[Janet Maoo @BuiltByJanet]

@thenamanraj AI agents lower the cost of building, but they don’t lower the cost of attention. Most founders will still underestimate how hard it is to stand out once everyone else can also ship fast.

[Janet Maoo @BuiltByJanet]

@jgreze Feels like we're seeing the beginnings of personal work agents: ones that understand their users and have a subtle sense of what to share.

[Janet Maoo @BuiltByJanet]

Hi Maya, I was definitely one of those girls before. With the encouragement of my friend, I started consistently sharing what I’m building, learning and observing online. I feel braver about voicing my ideas in public. Hopefully, my sharing can inspire others a little bit, if not, at least I crystallize my ideas, and leave a trail of how my thinking evolves over time.

[Janet Maoo @BuiltByJanet]

@sunra_ai Nice launch. The pricing transparency angle is cleaner than most “all models in one place” tools right now.

[Janet Maoo @BuiltByJanet]

I tried multiple video-generating tools and skills last week. As I posted before, I use multiple agents to help me analyze startups, which cut through the chaos and deepen my analytical capabilities by offloading the tedious information collection.

Intuitive, interactive and designable, that’s what I expected from the video. Literally just on prompt. The story of Granola👇