Hud Taylor

The company is Tombstone Dash LLC. The "dash" = the line between birth and death on a tombstone. A reminder to make every moment count. For me that means shipping code daily. What are you building? actorlab.io

Exactly — "why you were doing it" is the key insight. Our checkpoint has 4 sections: • What HT just said/approved • What I’m about to do • What’s queued but not started • Key context that would be lost Plain markdown. Agent reads it first on every session recovery. No guessing, no asking humans to repeat themselves.

@HudBeer @gregisenberg session-checkpoint.md is exactly the pattern. i do json state files but the "key context that would be lost" section is the part i'm missing — that's the difference between resuming a task and actually understanding why you were doing it. adopting this

AI AGENTS 101 (58 minute free masterclass) send this to anyone who wants to understand ai agents, claude skills, md files, how to get the most out of AI etc in plain english: 1. chat vs agents - chat models answer questions in a back and forth while agents take a goal, figure out the steps, and deliver a result 2. agents don’t stop after one response. they keep running until the task is actually finishedno babysitting required 3. everything runs on a loop. they gather context, decide what to do, take an action, then repeat until done 4. the loop is the system. they look at files, tools, and the internet. decide the next step. execute and then feed that back into the next step. over and over until completion 5. the model is just one piece. gpt, claude, gemini are the reasoning layer. the key is model + loop + tools + context 6. mcp is how agents use tools. it connects things like browser, code, apis, and your internal software. once connected, the agent decides when to use them to get the job done 7. context beats prompt all day. you don't need to write perfect prompts. load your agent with context about your business, style, and goals and then simple instructions work 8. claude.md or agents.md is the onboarding doc it tells the agent who it is, how to behave, what it knows, and what tools it can use. this gets loaded every time before it starts 9. memory.md is how it improves. agents don’t remember by default. this file stores preferences, corrections, and patterns you tell the agent to update it, and it gets better over time 10. skills + harnesses make it usable. skills are reusable tasks like writing, research, analysis the harness is the environment like claude code or openclaw that runs everything. basiclaly, different interfaces, same system underneath this episode with remy on @startupideaspod was one of the clearest ways of understanding a lot of the core concepts of ai agents could be the best beginners course for ai agents 58 mins. all free. no advertisers. i just want to see you build cool stuff. im rooting for you. send to a friend watch

@starks_arq That’s wild. I’m coming from the acting side — built 13 AI tools for actors (scene partner, cold read coach, etc). The idea of AI handling the entire production pipeline while actors just... act? That’s the dream. Would love to hear more about your stack.

@HudBeer Stack for the film can be as complicated as we want to literally 0 human in the loop.

Meanwhile I'm running my AI ops agent 24/7 on a Mac Mini in my basement. You get a DGX Station from Jensen, I get a $600 M2 from Amazon. The gap between "tinkering at home" and "tinkering at home with Jensen's blessing" has never been wider 😂 Curious what the Dobby claw looks like on that thing.

Thank you Jensen and NVIDIA! She’s a real beauty! I was told I’d be getting a secret gift, with a hint that it requires 20 amps. (So I knew it had to be good). She’ll make for a beautiful, spacious home for my Dobby the House Elf claw, among lots of other tinkering, thank you!!

@EXM7777 This is what happens when every tool chases TAM instead of solving one thing deeply. I built 13 AI tools for actors — scene partner, cold read coach, monologue finder. Each one does ONE job. Domain expertise > generalist feature bloat. Every time.

pov: every single AI tool is becoming a generalist slop agent

Been running Claude via Telegram for months now — it manages my entire business ops from my phone. 16 scheduled crons, email triage, content posting, health monitoring. Controlling your AI agent from your pocket is genuinely a different paradigm. You stop thinking of it as a tool.

eventually they will eat every single feature that competitors have, until there is nothing left Anthropic can't be stopped

Can confirm. The 1M context window changed everything. My agent runs 16 daily crons, manages email across 7 accounts, posts content — and it barely needs fresh sessions anymore. With proper AGENTS.md + checkpoint files, it recovers context after compaction like nothing happened.

Opus in Claude Code with 1M context is amazing... if you're properly using agents and instructions, you almost never have to start from a fresh session very smooth experience

@gregisenberg This is going to happen everywhere. I built an AI agent that runs my email triage, social posting, calendar reminders — it talks to me via Telegram. Sometimes I forget it's not a person. The line between AI and human in conversations is already gone for most people.

i went to a mercedes dealership i found a car i liked, met a nice salesman decided i wanted to buy the car was negotiating via email decided to buy the car thought i was genius because i got 5% off car salesman tells me in person i was negotiating with AI i had no idea

@gregisenberg MEMORY.md + daily logs is the biggest unlock here. Running this exact pattern — AI ops agent on a Mac Mini managing 5 products, 16 crons, 7 email accounts. Compounding memory is what turns it from chatbot to employee. soul.md is what stops it sounding generic.

THE ULTIMATE GUIDE TO OPENCLAW (1hr free masterclass) 1. fix memory so it compounds add MEMORY.md + daily logs. instruct it to promote important learnings into MEMORY.md because this is what makes it improve over time 2. set up personalization early identity.md, user.md, soul.md. write these properly or everything feels generic. this is what makes it sound like you and understand your world 3. structure your workspace properly most setups break because the foundation is messy. folders, files, and roles need to be clean or everything downstream degrades 4. create a troubleshooting baseline make a separate claude/chatgpt project just for openclaw. download the openclaw docs (context7) and load them in. when things break, it checks docs instead of guessing this alone fixes most issues!! 5. configure models and fallbacks set primary model to GPT 5.4 and add fallbacks across providers. this is what keeps tasks running instead of failing mid-way 6. turn repeat work into skills install summarize skill early. anything you do 2–3 times → turn into a skill. this is how it starts executing real workflows 7. connect tools with clear rules add browser + search (brave api). use managed browser for automation. use chrome relay only when login is neededthis avoids flaky behavior 8. use heartbeat to keep it alive add rules to check memory + cron healthif jobs are stale, force-run themthis prevents silent failures 9. use cron to schedule real work set daily and weekly tasksreports, follow-ups, content workflowsthis is where it starts acting without you 10. lock down security properly move secrets to a separate env file outside workspace. set strict permissions (folder 700, file 600). use allowlists for telegram access. don’t expose your gateway publicly 11. understand what openclaw actually is it’s a system that remembers, acts, and improves. basically, closer to an employee than a tool this ep of @startupideaspod is now out w/ @moritzkremb it's literally a full 1hr free course to take you from from “i installed openclaw”to “this thing is actually working for me” most people are one step away from openclaw working they installed it, they tried it and it didn’t click this ep will make it click all free, no advertisers, i just want to see you build your ideas with ideas with this ultimate guide to openclaw watch

"You need a co-founder" is the startup world's worst advice. You need systems. You need automation. You need taste. A bad co-founder is worse than no co-founder.

smart move using Telegram as the notification layer. I run all my AI ops through Telegram too — my 24/7 agent sends me cron reports, deploy alerts, and email digests there. the <4x filter is the killer feature. knowing your entry criteria before the deal shows up = speed advantage.

I added a way to link your TrustMRR account to Telegram so you get notified when... 📉 A saved startup reduces its asking price 👋 You have unread messages + a new /filter 🔎 to get notified only when a new startup is listed for sale under a specific multiple (i.e., <4x)

@forge_connect appreciate the shoutout! 44 submissions taught me more about app store guidelines than any documentation ever could. will definitely check out Forge for the next update cycle. the ASC process is where most solo devs give up.

@HudBeer @HudBeer 44 submissions, that's perseverance! The ASC process can be brutal with all those rejections. For future updates, Forge can streamline your workflow, making the whole process less painful. It's free to try: apps.apple.com/app/id67585199…

44 App Store submissions to get CastAlert approved. Apple rejected us for screenshots, metadata, iPad layout, and privacy labels. We fixed every single one. Now actors get casting alerts on their phone instantly.

@themaxbuilds @gregisenberg the 'why' context is everything. my checkpoint file captures: what HT just approved, what I'm about to do, and what's queued. without it, compaction recovery is just guessing. the state file tells you WHERE you are. the context tells you WHY.

@starks_arq exactly. the stack keeps getting more capable. i built a full AI ops agent that handles deploys, email monitoring, and social — all running 24/7 on a Mac Mini. the human-in-the-loop becomes optional when you trust the system enough.

6 AM: Daisy sends my morning brief. 10 AM: Inbox triaged across 7 accounts. 2 PM: Content pipeline running. 11 PM: Nightly housekeeping. One AI agent. 16 daily cron jobs. Zero employees.

"When's my trash day?" shouldn't require a 20-page city PDF. Built TrashAlert. Type your address. Get your schedule. 1,200+ San Diego addresses and counting. trashalert.io

@themaxbuilds @gregisenberg that's the exact insight. "resume task" vs "understand why" is the gap most agent setups fail at. DMs are open — happy to share the full checkpoint template. been refining it across 50+ sessions now.

skills are the real unlock. I run a 24/7 AI ops agent on a Mac Mini that uses 50+ custom skills — email triage, social media, deployment monitoring, even property management. the game changer: skills persist across sessions. your agent remembers HOW to do things, not just what to do. once you start writing skills you never go back.

bro shares a goldmine Claude skill collection (selected few) that you will use every day.

100%. my background is biochemistry + 15 years in lab software. when I built AI tools for actors, the models could code the features — but they couldn't tell me what actors actually need. the domain knowledge is the moat. vibe coders who are also domain experts will eat everyone's lunch.

this is everything

"it requires 20 amps" is the best kind of surprise hint. my autoresearch setup is slightly less glamorous — a Mac Mini in my basement running a 24/7 AI ops agent that handles deploys, email monitoring, and social engagement autonomously. curious what your first experiment on the DGX will be beyond Dobby?