Steve Huynh

YouTuber | Podcaster | Author

[Steve Huynh @ALEngineered]

@TorontoLinda @TrungTPhan Thanks!

[Steve Huynh @ALEngineered]

If your product gets better as AI models get better, you've got a great business that's resilient to the future. If your product gets worse as the models get better, time to pivot before you get subsumed.

[Steve Huynh @ALEngineered]

As a Seattle-area sports fan I’m getting sick and tired of turnovers.

[Steve Huynh @ALEngineered]

Thinking about getting your first dog and live in the Seattle area? Meet Kahlua.

She’s an older Spitz mix, about 40 lbs, and the chillest dog you’ll ever meet. She's house-trained, crate-trained, knows her commands, not interested in fetch, and very interested in a long walk and a nap.

She's good with kids, fine with other dogs, (and unbothered by squirrels.)

Don't get a puppy. You want an older, calm, well-mannered dog that just wants to be loved.

https://t.co/GsvK1gyqlH

[Steve Huynh @ALEngineered]

With AI I go back and forth between feeling more powerful than ever and feeling really obsolete (today). I realized that I may never write another line of code by hand ever again. I feel like I need to have a funeral for something that was such a big part of my life.

[Steve Huynh @ALEngineered]

If app companies were smart, instead of trying to prevent decompilation, they would lean into it and made it easy for agents to customize and extend their software, but in a way they could still make money.

[Steve Huynh @ALEngineered]

Me when I found out coding is solved.

[Yuta @ylevi0]

Looks like Rachel Zegler had one line too many before the Met Gala.

[Steve Huynh @ALEngineered]

Wait a second, why does the AI never choose Java?

[Steve Huynh @ALEngineered]

I think it's wild that AI has turned software developers into QA testers.

[Andrej Karpathy @karpathy]

We'll be spending a lot more time trying to understand the outputs of language models. A few thoughts, tips & tricks:

Writing. Something I've had success with: Ask your LLM to explain something in ASD-STE100, it's a controlled language specification originally developed for aerospace maintenance documentation. LLMs well-versed in this language and it comes with heavy constraints on clean writing style that I often find a lot more readable. Sometimes I've tried to soften it a bit e.g. ask for "80% of the way to ASD-STE100" because the spec is quite stringent. But even better:

Diagrams / images. Instead of writing, ask your LLM to create a diagram. These can be a lot easier to process, parse, and understand. But even better:

Web pages. Ask for output "in HTML" to get a beautiful, interactive webpage. LLMs are getting really good at frontend and can create beautiful experiences, animations, etc. But even better:

Explainer videos. The output format I am most bullish on is fully custom / bespoke explainer videos generated on any arbitrary topic. Experiment with things like "Create a 3b1b style video explainer on X. Use my ElevenLabs API key for audio narration". (you'd need an API key for the latter or you can ask your LLM to find you decent free alternatives that use your local compute). This is actually starting to work!

In summary:
- As LLMs get better, they will do more and more of the legwork autonomously, and a lot more of our work will rise up the abstractions into oversight and understanding.
- Luckily, LLMs can help here too because as intelligence and code are increasingly abundant, you can ask for large, custom, discardable software artifacts (e.g. web apps, video explainers) that would have never made sense to create before. Push the boundaries here and you'll be surprised.

[Steve Huynh @ALEngineered]

The ability to explain why we shouldn't build something is becoming a legitimate technical skill. It probably always was.

[Steve Huynh @ALEngineered]

It’s clear that the best people to direct agents are people who know how to code and build products. No agent is skilled at recursive self-improvement (make this software/product better independently) and I don’t see anybody making progress on this front.

[Peter Yang @petergyang]

Today, I'm launching a new AI course so you can copy the exact AI skills and workflows that I use to run my work and life.

Since going independent, the most useful thing that I've done with AI isn't trying every new tool or vibe coding apps. It's building an AI system that proactively takes work off my plate and can work across any model or harness.

Across 25+ bite-sized lessons, you'll:

1. Build an AI chief of staff to manage your inbox, calendar, and other tasks.

2. Set up an AI advisor to make better decisions about your life and career.

3. Automate workflows by chaining together reusable AI skills.

4. Make AI proactive with my best scheduled tasks for work and life.

I've included 40+ copy/paste prompts for you to set all this up in an afternoon.

In addition to the self-paced course, you'll also get:

- $600+ in AI tool credits for my favorite tools
- 16 ready-to-install AI skills
- Monthly live workshops with me in a private Slack community of 400+ builders

Get instant access to all of the above by signing up as a paid subscriber to my newsletter: https://t.co/sWbx5TfU2A

Join by October 7 to lock in $150/year. After that, the price goes up to $200/year.

Learn more at https://t.co/62XSo9NsDX and thanks for your support!

[Steve Huynh @ALEngineered]

To be right a lot you need to have the courage to change your mind when the new data says you were wrong.

I'll go against the grain here and say the company is doing the right thing by hiring people back. But I'll also say that I hope they are paying those that boomerangs a lot more to compensate them for this churn.

[Eugene Kim @eugenekim222]

New: Amazon wants former employees back — including some it laid off. In one email, a recruiter called the push the “Boomerang Reengagement Initiative.” In another, a recruiter asked whether Amazon’s RTO policy had driven the ex-employee away.

https://t.co/6yasgAeIGj

[Steve Huynh @ALEngineered]

Companies are downsizing aggressively but the people they retain are more important than ever because they know where all of the dead bodies are buried. They will have to give outsized rewards to desperately keep people in bad situations if companies cut too far. If you’re worried about being laid off, it's a good time to be the only person that knows how the critical systems work.

[Steve Huynh @ALEngineered]

You can’t say coding is solved but then have a flaky product. Coding will be solved when there are zero bugs, and your system has 100% uptime.

[Steve Huynh @ALEngineered]

My first thoughts about Jev:
1. The value is in the near-real-time generation of discrete options with associated distributions. You could probably do better on your own for specific domains, but doing it on the fly and at what seems to be arbitrary high quantity is truly amazing, if the quality is there. I suspect a lot of load is going to shift from agent to Jev.
2. A new tool chain will emerge to combine with LLMs. Like other tooling in AI, the first ones will be wonky, people will piece together something that works, then the frontier labs will incorporate this new hybrid architecture in their own products. I suspect Jev will be a hot acquisition target if what they have announced is true. I believe the frontier labs can develop similar on their own but I suspect it will take on the order of a year to do so. In AI time that’s a decade.
3. I am curious what type of bench marking and qualitative measures will surface. This is a big unknown-unknown but really exciting.
4. A good decision maker that won’t blow up scope (as my buddy @kunchenguid puts it with current LLMs “I ask it to make me a meal and it remodels my kitchen”) gives me real line of sight to getting humans out of the loop for recursive self-improvement. Currently, there’s no credible way to point a swarm of agents to make a system better without supervision. Good judgement is the critical piece to taste. If Jev is for real, that moat humans have may disappear.

I truly believe this marks a paradigm shift. Buckle in folks.

[Diogo Almeida @CompleteSkeptic]

After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI?

I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev

• 20-200x faster
• 40-400x cheaper (w/ output tokens free)
• Frontier composable intelligence optimized for decisions

AFAICT the shortest path to AI-based economic revolution

[Steve Huynh @ALEngineered]

I don’t know whether we're going to generate technical debt at a record pace or whether it's going to go extinct.

[Steve Huynh @ALEngineered]

The value of knowing how to build something is tanking while the value of knowing what shouldn’t be built is skyrocketing.

[Steve Huynh @ALEngineered]

The funniest possible outcome is that senior engineers become more valuable because we see what happens when unskilled people direct agents.