Ryan Stax

I build AI systems that save peole real time. Not theory. Not hype. Real setups you can steal and use Monday morning. 2,100+ people read The AI Handbook every week for the same reason: I only write about things I've tested myself. If I haven't built it, broken it, and rebuilt it, you won't find it here.

It does appear Claude is trying to cut token usage for it's users. Agent runs on a cheaper model, but when it hits a hard decision, it calls Opus for backup. Better results, lower cost.

@IAmAaronWill These are solid, easy to sell because they save money or make money for the client.

10 AI systems you can sell: Setter agent: $1-3k Order status bot: $1-2k Speed-to-lead bot: $1-3k Patient intake agent: $2-5k Claims intake system: $3-6k Recruitment screener: $2-5k Inventory reorder alert: $1-2k Lease renewal sequence: $2-4k Compliance doc reviewer: $3-8k Proposal follow-up system: $1-2k Pick a lane. One system. One niche. Unlimited opportunity right now.

Not everything in AI is a conspiracy or failure.. Claude Mythos wasn't shelved because it doesn't work. It works too well in areas where 'too well' is dangerous. The reflex to dismiss anything you can't download is costing you the ability to read where the industry is going.

@aryanlabde More like 20 different products earning $0

Nobody talks about the cost of vibe coding. Claude, cursor, vercel, supabase. You’re paying $200/month in tools to build a product making $0.

@gregisenberg I find a lot of what I see in this space is scaling cold outreach, which is going to result in endless spam

The best thing ANY engineer/programmer can do right now is learn how to become a top 1% marketer For 20 years, the engineer was the most important person in the room. They had the rarest skill. They could build the thing. Everyone else had to wait for them. Claude Mythos and the models coming after it are ending that era The new scarcity is the person who can look at a human being and understand exactly what they need to hear to take action. What makes someone click buy at 11pm. What makes someone tell a friend. What makes a stranger feel like a product was built specifically for them That is a completely different muscle than writing code or architecting systems Study why TBPN built a brand silicon valley is obsessed with. Learn why the headline is 80 cents of every dollar. Figure out why one subject line gets 40% open rates and the next one gets ignored Most engineers have never trained this muscle. They are world class at clearly defined problems. Marketing is the opposite. Fuzzy. Emotional. Irrational. The engineer who trains it becomes the most dangerous person in any room The CTO/CMO combo is the most valuable human in tech right now and almost nobody has both Computer Science school in 2026 should basically be part technical knowledge/part marketing knowledge I really think that... The best thing any engineer can do right now is learn how to become a top 1% marketer

@tomfgoodwin I heard the only accounts left on ChatGPT all end in @ Hotmail. com

Pretty sure everyone I know stopped using ChatGPT a while ago Everyone uses Gemini or Claude or Grok now. It changed fast. ChatGPT is for the olds.

@StrengthNPurpos Strong men take actions to better themselves, not for bragging rights.

Sign of a weak man: Wakes up at 4:30am to impress strangers on the internet.

@NoahEpstein_ The new Managed Agents are only available on Anthropic API, they aren't part of a subscription plan They blocked OpenClaw from subscriptions but can be run on API. Makes sense for Anthropic to block third party agents from OAuth if this was their plan.

anthropic cut openclaw access then 4 days later dropped managed agents you have to respect the play even if it makes you sick this is what they’re doing: openclaw proved there was real demand for agent infrastructure power users showed everyone what was possible then anthropic closed the door and shipped the cleaned up version themselves that is an unbelievable play (as annoying as it is) I do need to try it when it properly drops but won’t be suprised if it doesn’t work, because at the state opus is at it could be an issue. claude is still the best llm i’ve used my dad of all people is on claude now that alone tells you where this is going managed agents is massive for the consumer market because most people do not want to manage infra they do not want to set up sandboxes they do not want to think about orchestration they want the thing to work and for enterprise this is even bigger because now anthropic can say: you want agents? cool we’ve got the model we’ve got the infra we’ve got the managed layer just plug in and go that’s the whole stack so yeah this is an incredible move it’s just also incredibly annoying if you are an openclaw user because they basically watched the ecosystem prove the market cut off the most committed users then dropped their own version days later anthropic had my heart until this openclaw beef i’m taking that to the grave this will probably win the market But i still have generational beef with them

@elonmusk Every Moon Man colonizers dream... a lawn to cut

A friend suggested this but on the Moon

@coreyganim How do you manage unpredictable token costs? Even with clients API?

This is a business in a box. Claude just made it possible to build and deploy AI agents at scale with managed infrastructure. Here's how to sell it: 1. Pick a niche (real estate, e-com, coaching, agencies) 2. Build one agent that solves their #1 repetitive problem 3. Deploy it on Claude's managed platform (they handle the infrastructure) 4. Charge $1,500-3,000 setup + $500/month per client 5. The agent runs 24/7. You maintain it for an hour a month. 10 clients = $5,000/month recurring before you even think about scaling. The "prototype to launch in days" part is key. You can have a client's agent live before the second sales call.

@coreyganim makes sense.

@ryanstax agreed 100%, I usually refer to them as an "assessment" but for the sake of the post most people would understand audit right away

Easiest entry point into AI consulting: AI audits. Here's the framework: 1. Map every repeatable process in their business (takes 45 min on a call) 2. Identify which ones are costing the most time or money 3. Recommend the AI tools and workflows to fix them Charge $1,000 per audit. Takes about 2 hours to deliver. You're not even building anything. You're delivering a structured assessment and a prioritized action plan. Do 4 audits a month and you're at $48K/year from a service you can deliver in a single afternoon each week. Then upsell the implementation.

@SavinoDrip Probably could, I just have it scheduled on cowork as part of a few ither chained morning research agents

@ryanstax What about ccode? Or only works with cowork

How to build your own social listening system for free: 1. Give Claude browser access 2. Have it open X and talk to Grok naturally 3. Weight results by recency, authority, velocity 4. Add a fourth signal that matters to YOUR business 5. Schedule it to run at 9am daily 6. Read the summary over coffee That's it. That's the $800/month tool.

@godofprompt get deeper billionaire insight and just ask to give you advice based on everything it knows about you.

🚨 BREAKING: You can now turn Claude into any billionaire coach you want. Naval for leverage. Hormozi for offers. Bezos for customers. It adapts its thinking to whoever you choose and applies it to your situation. Here’s how to activate it:

@SavinoDrip If you try to do it in Chat, it will refuse. You need to use Cowork and tell it to go to X and navigate to Grok

@ryanstax Have it talk to Grok… 🤔

@RyanHanley_Com @SahilBloom When I held my first-born I could actually feel my brain rewiring, it was an incredibly odd sensation.

@SahilBloom My boys rewired my entire operating system. Before kids I optimized for achievement. After kids I realized achievement without someone to share dinner with is just expensive loneliness. They don't make you soft. They make you clear.

I spent most of my 20s thinking I didn’t want kids. Thought it was a distraction from achieving “success” (whatever that meant). Well, last night, I was tucking my son into bed and he looked at me and said, “Dada, you’re my hero.” It was the best moment of my life. I can’t imagine not experiencing this. I’m not sure how anything will ever measure up to the feeling I had in that moment. Purest joy I’ve ever felt. I’m glad my definition of success changed, because this version is much better.

What this update is in plain English: • Managed Agents is giving Claude a job and letting it go work on its own until it's done. • The agent can search the web, write files, run code, and execute commands without you guiding every step. • You can interrupt it mid-task if you want to redirect it. • It runs in a secure cloud container that Anthropic hosts for you. • It's API-only. You need to write code to use it. There's no chat interface for it. • You pay per usage (tokens consumed), not through a subscription.

@bazfurby multitasking

Someone explain this to me like I’m 9 years old pls

When someone says they used ChatGPT I just assume they have a hotmail account

@boringmarketer I am getting tired of Claude asking me for smoke breaks every 25 minutes.

I get tired of Claude Code telling me "honestly it's been a massive session, want me to save this for next session and wrap up?" no keep going, make no mistakes