JC (@shiftj) header
JC (@shiftj) profile picture

YC Founder • Advisor • Investor • Builder • SMB Owner

Earth
Joined July 2025
This heatmap might be worth $10M+. 10 years ago, capabilities like this were reserved for companies like Starbucks. Today, a small business owner can build them over a weekend. That’s what I find most interesting about AI. It’s not changing what’s possible. It’s changing who gets access.
I helped an SMB owner make $1M this weekend.  here’s the story 👇 A small business owner was looking to expand to 4 more locations. if successful, this would net him $1m/ year. The problem? If it fails he’s millions in debt and he was relying on: > gut feel > personal experience > randomly clicking thru google maps So I helped build an AI agent that scouts locations. We combined: > his 10+ years of industry knowledge > data across population, demographic, income, competitor, comparable businesses types > 30+ factors from a decade of experience. Then ran it across every subdivision in the United States. The result? We overshot. By a lot.. We ended up finding 50 exact-match expansion opportunities. The exact profile he was looking for. He was stunned. 4x locations makes him one million/yr this list is closer to $10M. But here’s my favorite part: These capabilities used to be reserved for large companies but are now rapidly becoming available to those with less resources. This kind of sophistication you’d normally expect from a company like Starbucks. Now available to a small business owner. Giving opportunities to people who didn’t think they had them. The type of scale he didn’t think was possible. Pretty damn cool.
3
5
47
12,174
What should you do if you’re young and ambitious but don’t know what to work on? Meet as many business owners as you can. Every industry, company type, and business model. Find one that interests you. Find one you can see yourself waking up for. Diving into. Where you’d be so pumped to win that you can’t lose. Where losing would still be a win because you can’t imagine a better way to spend your time. Expose yourself. You’ll probably find it.
1
2
19
499
Companies are using 10-20% of what's technically possible right now. Not a tech problem. An implementation problem. Most AI today handles basic, isolated tasks: • Drafting docs • Meeting summaries • Chatbots The real opportunity is integrated workflows. Stringing together multiple decisions, pulling context from systems, automating entire processes. That's where the value is. The gap between capability and usage is massive. Whoever figures out how to close it wins. And there’s so much surface area to tackle. — How to find AI implementation gaps: 1/ Pick a lane 
Find a high TAM industry with budget. Try to choose one that’s generally open to adopting tech and have budgets. If you find a more traditional industry which has those characteristics, even better - they’re probably hungry for solutions and underserved. 2/ Talk to experts
 Find people who know how the industry works and have control of budget. Ask about their job, how they get it done, and what roles they hire for. Dive deeper into those roles, rinse and repeat the questions. 
The job is to see all the moving pieces and how they all keep the business running. 3/ Find the workflow 
Once you see all the pieces, find something that's mission critical. It should either happen at high volume, expensive when it breaks, or leaves money on the table. You’re mapping an ROI statement, trying to find the area with the highest possible impact, if done better.
 That's where companies will pay to automate. 
Warning: You’ll be tempted to make big claims like “well what if I could replace your whole sales force with an AI avatar”. Find something impactful, but realistic. You should be able to build it inside of a month. Personally? I aim for 1-2 week builds to prove the concept. 4/ Land one contract 
Prove it works with a real customer. Get them to pay. Then get the results you promised. At this point you’e not only built it, but you’ve got the proof point to back it. 5/ Use that to sell everyone else 
One reference customer in an industry opens doors to the rest. They all have the same problems. 
The opportunity isn't building better models. It's finding which workflows to automate first.
2
2
1
25
3,152
Here's what I've got so far. Still rough around the edges, but it works. -- p.s. If you're building something similar, DM me. Still working through kinks.
12
1
1
145
13,520
Replying to @levelsio
No way, I love my UI. Even built one for my openclaw agents. living purely in chats kind of sucks
11
871
Replying to @mrmagan_
I created a control panel for my openclaw agents (vid attached) - this seems like a perfect fit tbh you have a skill for it?
1
4
165
Replying to @joaquintenest
Built out a full control panel for all my agents. Use Cases: > Check my calendar, prep for my calls > Monitor X, give me updates > SEO research, draft weekly articles > Learn my X, suggest posts daily > Check github, summarize releases > ... building out a lot more too ...
2
6
478
Replying to @CasJam
I spent wayyy to long on my multi-agent setup. Workin pretty well now tho!
1
179
I caved yesterday. bought a mac studio. a beefy one to run local inference. had to see for myself what all the hype was about. Result: I've never seen the future so clearly before.. cancelled my calls, didn't sleep, automated most of my morning rituals - and then some. > daily standup > kpi round up for sales > customer slack summaries > daily ideas for twitter posts > weekly tee times to book > weekly restaurants to try still hungry for more use cases. u have any?
3
11
1,718
Founders these days..
1
172