
COO @codos_ai | Investing @Delphi_Ventures Prev: @delphi_labs, @IBM, @HPE, @McKinsey, @CarnegieMellon, reformed CFA
Joined March 2015
- Tweets8.9K
- Following954
- Followers21.1K
- Likes30.1K
AI is killing entry level jobs - wrong!
It’s killing the “give me a 6-figure desk job and teach me how to work because I just graduated uni” jobs
But not for people who hustle - post below is a perfect example
AI is creating never-before work opportunities for young people
AI agencies are charging 60-year-old business owners $25,000 for an AI system that takes 35 minutes to set up inside a $20 Claude account
I know because I’ve done over $1,000,000 selling them (and two guys in my group with zero coding skills, Luis and Eric, just closed a $96,000 contract doing the exact same thing)
F*ck gatekeeping it. Here is the entire $25,000 AI setup and the 4 steps to sell it off your laptop this week:
Minute 0 to 10: Pull a rich dinosaur business
Skip broke tech startups. Go to diced. si (pulls verified owner cell numbers for 10 cents a lead instead of paying Apollo $1) and filter for local brick-and-mortar companies doing $3M to $15M a year
Think HVAC, roofing, medspas, plastic surgeons, regional paint plants, or Texas oil and gas yards. These guys are 62 years old, still run their business on sticky notes, and bleed $30,000 a month whenever their receptionist misses phone calls after 5pm
Minute 10 to 25: Turn their website into a live AI voice clone
Before you ever speak to the owner, copy all the text off their website and paste it into your $20 Claude account with this prompt:
"Act as a senior AI voice architect. Extract this company's services, pricing rules, service areas, top 10 customer objections, and booking flow. Output a conversational system prompt for an inbound AI phone receptionist that sounds like a sharp 28-year-old office manager and pushes every caller to book an appointment on the calendar"
Take the prompt Claude spits out and paste it into Lumina (our software that spins up a real working local phone number and branded web demo for any company in 60 seconds)
Minute 25 to 35: The "Call Your Own Business" ambush
Do you realise what happens when a 60-year-old millionaire hears an AI answer the phone in his own company's name?
When you get him on a 15-minute intro call (or automate your outreach through trygtm. com so your calendar fills up while you sleep), don't bore him with slides. Read him the 10-digit phone number you generated and tell him to call it on speakerphone right in front of you
He dials the number. His new AI receptionist picks up in half a second, answers his trick questions about pricing, handles an objection, and texts his iPhone a live calendar invite while he watches
How you deliver the $25,000 build without writing code:
The Claude + Lumina demo gets the owner hooked on Call 1, and you never quote a price on that first call. You bring him back 48 hours later on Zoom with a 2-option roadmap ($25,000 full 4-agent back-office overhaul vs $10,000 core phone and SMS system)
When he wires the $10,000 to $25,000 upfront deposit to plug it into his real CRM, QuickBooks, and phone lines, you don't code the backend yourself. You pay one of our vetted wholesale senior engineers $1,000 out of the client's deposit to hard-code and test the whole production build for you, and you pocket the other $9,000 to $24,000 clean
Right now you face a binary choice with AI. You can sit on the sidelines waiting to get replaced, or you can be the one putting AI into the 33 million old-school companies across America and getting rich off the wave. These 60-year-old owners only install their core AI infrastructure once, and over the next 12 to 18 months that window shuts for good
Reading the 35-minute Claude workflow gives you the blueprint, but actually putting $10k to $35k wires into your bank account takes our live 60-second Lumina voice cloner, our word-for-word 2-call closing script, my $1M+ B2B case studies you can borrow on Zoom, and our 25+ vetted wholesale senior developers building your clients' systems at $0 markup so you never go lone wolf
If you want to partner with us, book a 1-on-1 gameplan call with me or my team. Once you lock in a time on the calendar, we'll audit where you're starting from, pick the fastest niche to land $10k to $30k AI deals in your market, walk through how partnering inside AI Frontrunners works, and hand you a custom 30-day AI roadmap you keep either way
Go to timchao. ai (link in bio) and grab your call spot before the calendar fills up
I like how @dabit3 put it:
It’s outcome engineering
Most of us are managing agents all wrong
We setup agents to do things we already know how to do, then tell them exactly how we want them to do things
This is treating agents like interns
That was fine earlier this year when models were not as capable
But ask yourself - if you just hired a 130 IQ employee, would you just have them do things you already know how to do and tell them exactly how to do it?
If you’re using frontier models with agents, you want to give them goals and objectives for things you don’t know how to do, then empower them to do the things
Hire agents as experts, not interns
Yes you should still have some agent interns to automate boring work to free up more of your time
But if you’re not using agents to do things you don’t know how to do then you’re missing out
Let them go figure out how to do the things that expand beyond your skills and capabilities just like a good leader would when hiring a team
Can't wait for everyone on the timeline to pretend they knew the Riemann Hypothesis and have a take on it
Can you 5x your engineering team's output in 1 month?
Here's how I've seen this happen
Bring in a super AI-native engineer from the outside to:
1. analyze company's SDLC, code, issue tracking and AI setup
2. build a custom plug-in for humans
3. build a software factory for agents
4. push new code through factory with escalation to humans
5. build a dashboard to monitor it all
Within a month you see massive improvement
Obviously not for teams already operating at the frontier, this is more for the teams still early on the AI adoption curve
A new personal AI agent has entered the arena - Replit Amuse
Time to update my report
Companies have spent billions on personal AI assistants. But you rent them, and your memories, chats and logins live on their servers.
We gave Replit Missions a challenge: replicate the leading AI assistant.
github.com/replit/amuse
3 days later, it was done. We used our Security Scanner to find & patch vulnerabilities, and make sure that the end product was safe to use.
Meet Amuse: a free, open-source AI assistant that runs on your own computer. It researches, browses, sets reminders and builds small apps, and it asks before it acts. Your data and logins stay with you.
Overmind is an "AI lab in a box" for agent builders, and my go-to tool for fine-tuning models for agents
Here's what you can do:
> it traces your production agents
> turns those traces into datasets
> runs evals
> fine-tunes smaller open models on your data (you keep the weights)
> hosts them on its own GPUs
> full open source, with a hosted console
Excited to see this now available right through Grok Bot!
Turn Grok Bot into your own AI lab, right from your phone.
• Connect @bot to the Overmind MCP
• Upload a dataset of examples to learn from
• Tell it what to learn and let it fine-tune the model
Over to you: overmindlab.ai
7
24
1,679
Most of us are managing agents all wrong
We setup agents to do things we already know how to do, then tell them exactly how we want them to do things
This is treating agents like interns
That was fine earlier this year when models were not as capable
But ask yourself - if you just hired a 130 IQ employee, would you just have them do things you already know how to do and tell them exactly how to do it?
If you’re using frontier models with agents, you want to give them goals and objectives for things you don’t know how to do, then empower them to do the things
Hire agents as experts, not interns
Yes you should still have some agent interns to automate boring work to free up more of your time
But if you’re not using agents to do things you don’t know how to do then you’re missing out
Let them go figure out how to do the things that expand beyond your skills and capabilities just like a good leader would when hiring a team
And this is basically what we do every day with people
A slide from @AnthropicAI's IPO roadshow. Hits hard and is very true.
AI is soooo much better than how most people are using it.
I’ve said this before - if there’s 1 skill to learn in AI right now it’s making good video
Do this well and you’ll crush it, at least until the models diffuse your skills
For longtime Hermes Agent users, run this command:
hermes sessions optimize-storage --yes
My session db was >1GB and hurting my performance and I didn’t want to delete old session data
The complete guide to Hermes Mobile is now live on Hermes Atlas
My agent did all the work on this, but I've personally been enjoying the beta version of Hermes Mobile and think you will all love it!
The guide includes good videos from @tonbistudio and @AlexFinn
The GTM engineer is the hottest role in tech right now
It's the 10x engineer concept applied to a growth/marketing person
A couple things I’ve noticed on X + a million $ opportunity for someone
1. People share everything
If you need anything from specific playbooks to valuable tips to creative growth hacks, you’ll find people sharing it
Sounds like secret sauce, why share?
Because it helps build distribution and smart accounts know that is far more valuable than keeping these tips a secret
2. If you express even a little interest in these topics, your FYP becomes flooded with it
Say you click on a few SEO/AEO articles, next thing your FYP is nothing but those posts
Which means that the previous topic you were interested in gets suppressed, so you can’t easily get the best new content across a range of specific topics
And here’s the big opportunity
Figure out how to aggregate all these best practices, playbooks, tips and hacks that people share
And train a team of agents on it, constantly learning from the best new content to stay on the leading edge, then sell agentic services
But wait, why can’t you just ask Grok for this content?
Go try it - I promise what you get is more generic and not the cutting edge stuff you see on these topics
Maybe because it’s new content and Grok gives more weight to mature content on certain topics
But whatever the reason, the real sauce is on the recent timeline and the only way to consume it is via a well curated timeline
Figure out how to harness this sauce and you can build a good business on it


















