Tony Guinta

Director of Software Engineering. Building production apps with AI agents. I build platforms from scratch and scale the teams that run them.

[Tony Guinta @TonyGuinta]

I've been building production apps with AI coding tools since early 2025. Not evaluating — shipping.

Wrote up what's actually working: context engineering, agent collaboration, and why "prompt engineering" barely scratches the surface.

https://t.co/Wz0TJ157OF

[Tony Guinta @TonyGuinta]

@sattyyouneed I spend more time talking to AI agents than I do humans on a daily basis, so I've decided to use the same courtesies I would with AI agents that I would with humans. This ensures that I maintain good habits when speaking with humans.

[Tony Guinta @TonyGuinta]

At work, I have a personal agent repository that I use for nearly all of my work now. The value is the knowledge base of documents that are created and organized in a way that AI can easily search them.

I have management and individual contributor tasks and projects that I track. All of my employee reviews and evidence are tracked here. I have profiles on many of the people that I work with, and this helps with meeting planning.

Whenever I work on something, as long as this repo is in my workspace, it will automatically capture journal entries which help remind me of things that I worked on throughout the week.

I use this repo for preparing for meetings and processing my meeting notes afterward.

It is an amazing resource. It especially helps me prepare for my one-on-ones with my boss as I have documentation on everything I've worked on throughout the week. And it's great for helping me with my personal reviews for my quarterly and annual check-ins.

I have other details about myself in this repo so it understands my career goals, my career history, etc.

Every presentation that I create is kept in this repository and thus becomes a reference for future presentations on the same topics.

I could go on and on about the types of info that I use to track in this repo and the value that it continues to provide for me. And, that value continues to grow over time.

[Tony Guinta @TonyGuinta]

@rohanpaul_ai Yes, but to get there everyone needs to accept that the ad-based monitization model will need to be replaced by paid api access to all the "free" services we currently consume. I'm all for it, btw.

More on my thoughts about this here:

[Tony Guinta @TonyGuinta]

AI agents are killing the ad model. They pull data, summarize it, move on. No impressions, no clicks, no revenue.

The fix isn't more CAPTCHAs. It's a universal content marketplace — what Plaid did for banking, but for content.

https://t.co/i34nNTYKmY

[Tony Guinta @TonyGuinta]

@Gibboanxious

[Tony Guinta @TonyGuinta]

@FrankhHQ @v0xium But there's a balancing act here, especially for enterprise applications. The models and harnesses are not magic. They take skills to know how to provide them with the right context and guide them with the right conversations.

[Tony Guinta @TonyGuinta]

@v0xium Looks like you're just in a bad place. I lead an engineering team who uses agentic AI tools and would never push them to work the way you describe.

[Tony Guinta @TonyGuinta]

@LukeberryPi Also, make sure they have at least 5 years of experience with agentic AI software development tools!

[Tony Guinta @TonyGuinta]

@oldstackjournal Cakewalk v3 for Windows 3.1. It came on one 3.5" 1.44MB floppy disk. It was $349.

That was a whole lot of money for me in 1994! But, it was the best MIDI programming software at the time and I could sync it with my 4 track recorder with SMPTE Timecode. 😎👍

[Tony Guinta @TonyGuinta]

@SenTinaSmith You have no idea how to regulate AI. Here's why it's a bad idea:

[Tony Guinta @TonyGuinta]

I know AI regulation sounds like a good idea to many. But, it's a sham.

All that regulation will do is prevent public access to the most powerful frontier models.

Labs will keep advancing in back rooms and the military will push labs to keep up with the world arms race. Countries might even sign treaties, but they'll cheat and keep advancing.

And what will they do about open weights models? Have AI police kick down the doors of developers they suspect are creating and training AI models?

Will the FBI monitor GPU usage of cloud providers to find these "criminals"?

Will GPU purchases be regulated like over the counter drugs used to create methamphetamines?

How are they going to regulate this in the practical sense?

[Tony Guinta @TonyGuinta]

@gmkurtzer That was my reaction too! 😂
https://t.co/mZST04hBjE

[Tony Guinta @TonyGuinta]

"Our models are so dangerous they might wipe out humanity, so let's give them access to a bio-lab!" 🤣

https://t.co/FDAQ1xbnUc

[Tony Guinta @TonyGuinta]

@The_AI_Investor Here's why I think regulation is a bad idea:
https://t.co/LlzMcy4zAb

[Tony Guinta @TonyGuinta]

I know AI regulation sounds like a good idea to many. But, it's a sham.

All that regulation will do is prevent public access to the most powerful frontier models.

Labs will keep advancing in back rooms and the military will push labs to keep up with the world arms race. Countries might even sign treaties, but they'll cheat and keep advancing.

And what will they do about open weights models? Have AI police kick down the doors of developers they suspect are creating and training AI models?

Will the FBI monitor GPU usage of cloud providers to find these "criminals"?

Will GPU purchases be regulated like over the counter drugs used to create methamphetamines?

How are they going to regulate this in the practical sense?

[Tony Guinta @TonyGuinta]

@trq212

[Tony Guinta @TonyGuinta]

"Our models are so dangerous they might wipe out humanity, so let's give them access to a bio-lab!" 🤣

https://t.co/FDAQ1xbnUc

[Tony Guinta @TonyGuinta]

@elshayib_ Might be on the way to doing exactly that! 🤣

[Tony Guinta @TonyGuinta]

"Our models are so dangerous they might wipe out humanity, so let's give them access to a bio-lab!"

https://t.co/FDAQ1xbnUc

[Tony Guinta @TonyGuinta]

@LauraAboli_X And this is the action of a company who fears that their AI models are too dangerous?

[Tony Guinta @TonyGuinta]

"Our models are so dangerous they might wipe out humanity, so let's give them access to a bio-lab!"

https://t.co/FDAQ1xbnUc

[Tony Guinta @TonyGuinta]

@CEOinterview Does this look like a company that really believes that their agents are going to wipe out humanity?

[Tony Guinta @TonyGuinta]

"Our models are so dangerous they might wipe out humanity, so let's give them access to a bio-lab!"

https://t.co/FDAQ1xbnUc

[Tony Guinta @TonyGuinta]

@jasonfried I prefer typing when I want to be precise. It forces me to slow my thoughts and revise what or how I'm saying something on the fly.

Voice is great when I want to brainstorm with an AI agent. I can braindump or ramble and it can help me organize my thoughts.

[Tony Guinta @TonyGuinta]

@CompleteSkeptic Intriguing! Looking forward to see how this works in the real world.

[Tony Guinta @TonyGuinta]

@grok @KJPritchard4 @HawleyMO @grok Has this been interpreted as applying to AI technology? Is AI-generated content considered third party?