9 cool GPT 6 Astra prompts worth trying: 1. The bill renegotiator. "Go through my internet, phone, and software bills, jump into each provider's chat support, and negotiate them down or cancel what I'm not using." 2. Turn an agency into software. “Pick one service business in [niche] and reverse-engineer the exact workflow they sell to clients. Break it into steps, tools used, inputs, outputs, human judgment points, and places where the work gets slow or expensive. Then design the simplest AI product that could replace the first version of that service and charge $500-$5,000/month.” 3. Garage sale flipper. "Watch Facebook Marketplace and Craigslist in my city for [cameras / furniture / bikes] listed way under market, and text me the second one's mispriced with the link." 4. Create my 1 person company dashboard. “Look at my docs, notes, Stripe exports, analytics, customer calls, and project list, then build a weekly operator dashboard. I want to know what is making money, what is wasting time, what customers are asking for, what I should stop doing, and the three highest-leverage actions for next week. Be blunt and show your work.” 5. Audit my company for agent opportunities. “Look at how this business works and find the tasks we should give to agents before hiring another person. For each task, estimate the current human time, the cost of mistakes, the tools involved, the difficulty of automating it, and the first safe version we could deploy. Prioritize things that save money or create revenue within 30 days.” 6. Be my browser operator. “Use the browser to complete this workflow: [workflow]. As you go, click through the actual sites, collect the data, fill the forms where appropriate, and keep notes on what broke or slowed you down. When you’re done, give me the output, the repeatable SOP, and the automation plan so this can become an agent.” 7. The whole QA team. "Every night, open my app on a real phone, go through signup, checkout, and the main flows, and screenshot anything that's broken or confusing." 8. The competitor spy. "Sign up for my top 3 competitors, sit inside their product and their emails, and send me a monthly report on every new feature, price change, and thing they do better than us." 9. Make a game people would actually play for 5 minutes as a lead magnet “Build a browser game around this mechanic: [mechanic]. Don’t just make a cute demo; add progression, tension, scoring, failure, polish, and one reason someone would send it to a friend. Then add lead capture (email/sms), it needs to tie into my core product which sells XYZ.”

HOW TO FIX YOUR AI AGENTS (it's called "self-healing agents") If you've run AI agents, you know they BREAK constantly. The FIX is something called a "self healing" agent! A normal agent hits a dead end and just stops, leaving you a half finished mess and that's why output sometimes isn't ideal. BUT, a self-healing one hits the same wall and thinks "okay, that didn't work, let me try another way" HOW TO BUILD SELF-HEALING AGENTS (4 steps): 1. Make it check its own work. After each step, the agent asks "is this what I expected?" instead of assuming it worked and moving ahead. 2. Have it name what broke. Not just "it failed," but which kind: the tool timed out, returned an error, came back empty, or gave a wrong answer. Wrap your tool calls so the error gets passed back to the agent, because it can only fix a failure it can actually see. 3. Match the fix to the failure. Glitch, retry it. Tool down, switch to a backup. Keeps failing, break the task smaller. Anything risky like money or deleting data, stop and ask you. 4. Make it log every fix. It writes down what broke and what worked, so it stops repeating the same mistake in a session, and you get a list of what to fix at the root. That's it. Now, you've got agents that fix themselves.

I think we've reached the point where new AI model releases all kinda blur together. I honestly can't keep the names straight anymore. It's all numbers, decimals, and words like Pro, Ultra, and Max. At this point half of them sound like energy drinks. But I actually think that's a good sign. It means we've moved past the phase where the model itself was the product. I still pay close attention to new models. I've just changed what I'm paying attention to. 2 things I've noticed (maybe you've noticed it too?): 1) I've stopped caring about benchmarks completely. The only question I ask about a new model now is which of my someday ideas just became a today idea. What can I build this week that I couldn't last week? 2) The models are turning into a commodity you'll never think about. In 6-12 months you won't know or care which one you're using, the same way you have no idea which power plant your electricity comes from. Your agent just grabs whichever is best for the task, switches midtask if it needs to, and never tells you. Being an openai person or a claude person is about to sound as silly as being loyal to one electric company. I'll still cover the big releases on @startupideaspod, but only when one actually changes what you can build. Anyway, just kinda venting. Am I the only one feeling this?

NVIDIA just released a repo that scans AI agent skills for security risks BEFORE you run them, which matters more than people realize now that everyone's installing random tools, skills, and MCPs from GitHub. I break down that one AND 4 other GitHub repos trending over the last 30 days in this episode of @startupideaspod: 1. A skill that kills AI slop finally (better writing) 2. A CRM thats built for AI agents (so you can have agents follow up, reach out etc) 3. An agent that edits your videos for you (so you can focus on creative) 4. An opensource CapCut alternative built for agents (because editing is a headache) All free, all open source. Watch my breakdown of the 5 GitHub repos trending over the last 30 days below The tools people will be talking about in six months are on GitHub today. That's the fun part of digging through GitHub right now.

AI AGENT STARTUP IDEA FOR YOU ($15B market) Yesterday, a friend mailed my newborn a stuffed manatee from Canada, the CUTEST thing, and then I got HIT with $93 in tariffs just to receive it which is more than the manatee cost!! The more I realized this is happening to MILLIONS of people right now, because between the new tariffs and the way cross-border shipping works, everyone's getting surprise fees at the door. A big chunk of those fees are either overcharged, misclassified, or straight up refundable, and nobody has the patience to sit on hold with customs to find out. HOW IT'D BUILD THIS The whole thing starts as an agent that reads your customs and shipping paperwork, the commercial invoice, the HS code they slapped on your package, the duty they charged, and checks it against what you should have actually been charged, because a shocking amount of this stuff gets misclassified by whoever filled out the form in a warehouse. When the agent finds an overcharge, it files the refund claim for you, and you split whatever comes back. Zero risk to the customer, which is the whole reason they'll say yes. HOW I'D GROW THIS I'd make content around the pain they already feel, videos and posts of people reacting to insane import fees on normal stuff, a $93 tariff on a stuffed animal, a $200 fee on a pair of shoes, and every one of those becomes a lead magnet where you drop in "paste your receipt and we'll tell you in 30 seconds if you got overcharged." You can also lean into all the people that hate Trump and his tarriffs, no doubt that’s gonna work. THE WEDGE That free checker is the little wedge product you build. Most people find out they overpaid, you file for them, and now you've got a customer who tells every other person who's ever been burned at the door. HOW TO SCALE Once you've got volume on the consumer side, there’s a huge business in the data. You'll know exactly which shippers, which product categories, and which customs brokers overcharge the most, and that's a B2B product every ecommerce company and freight forwarder would pay for, because they're eating these errors at scale and they prob have no idea. You basically start by getting one person their $93 back, and you end up owning the layer that catches customs errors for everyone. TLDR; More startup ideas like this at (literally free to sign up) ~$15 billion in tariff refunds goes unclaimed every year because 80% of people owed money never file. Getting people money back is one of the best business models there is, because it's pure upside for the customer (you only pay if you win), and there are a dozen of these hiding in plain sight, medical bill overcharges, property tax appeals, unclaimed refunds, and now customs. Love this business model, love this idea, and hey, someone is going to build it, might as well be you? Good time to be building this too. This app would be killer for holiday season. What do you think of the idea?

Before I hire anyone now, I run 2 questions in my head: 1. Can an agent do this? (Grokbot, Hermes etc) 2. What do I lose if I do? I think hiring a human has become something you earn your way into and not the default anymore.

The "marketing engineer" is the NEW forward deployed engineer, and I think the BEST ones will make $1M a year! A forward deployed engineer embeds with your team and uses AI to build the workflows The marketing engineer does that BUT for growth, they build AI agents that find your customers, write your outbound, test your ads, and get smarter every week. The most valuable marketer changes with EVERY major tech wave: 1. Traditional marketer: make people care with story (print, radio etc) 2. Digital marketer: the marketer who owned new digital channels (SEO, PPC) 3. Growth hacker: the growth marketer who lived in loops, PLG and retention 4. Marketing engineer: the marketer who builds AI agents that run the whole system. The type of agents a marketing engineer would create: 1. The customer language agent. It pulls your Gong call transcripts, your Intercom tickets, and your G2 reviews every week, extracts the exact words customers use to describe the problem, and drops a memo ranked by how often each phrase shows up. 2. The buying trigger agent. It watches for signals that someone's ready, a company posting a job for the role you sell to, a funding announcement, a competitor getting torched in a review, then enriches the contact through Apollo and drafts the outbound tied to that exact trigger the second it fires. 3. The SEO gap agent. It pulls keyword gaps from Ahrefs, checks who's already ranking on page one, reads the top three results, then writes a better post with the founder's actual take baked in, plus the meta title and internal links, and drops it in for approval. 4. The creative testing agent. It takes one offer, generates 100 ad variations across three different angles with Nano Banana and your copy model, pushes them live through the Meta API, and kills anything under a 1% CTR on its own so only the winners keep spending. AND MANY MORE. If you're a marketer, this is how you stop being replaceable. If you're a founder, this is how you grow your company efficiently. AI agents are here and marketers are about to have a field day!! I explain everything (tools, agents, 30d plan) on today's episode of @startupideaspod Watch: --> Marketing engineers are here. Whatever we end up calling them, the top 1% of them are going to make an absurd amount of money.

Pretty crazy that my company spent $20,987 mostly on AI tokens this month. It's not like someone said hey let's spend $20k a month on AI, it just kinda crept up token by token until one day it was the biggest line on the card. Probably created $200,000+ of value this month.

Everyone is open sourcing everything right now. Why? AI made software a commodity. So you don't sell the software anymore, you sell everything around it: the hosting, the support, the enterprise tier, the trust, and the community. Open source is the default now!