> I'm an agentic engineer with an emphasis in software engineering
This is odd. You wouldn't consider yourself a software engineer with an emphasis in...agentic engineering, or something like that? What is your background?
The extent to which AI is performing essentially all of my code generation and a large portion of the actual software engineering is astounding. My role has largely evolved into perfecting the process by which the agents are guided into well-architected solutions, manually testing their outputs, judging the artifacts that they present as evidence of clean code and robust implementation, and understanding business workflows well enough to guide the agents to a good working app that can be used by people.
AI is so integral to everything that I do professionally and in the software engineering space, it's outrageous. And I can apply those skills with extreme effectiveness to other areas outside of software engineering, such as tax preparation, legal document design, and my associates to CAD, energy flow analysis, structural engineering, etc.
I believe it is a very useful way to set ourselves up for our current reality and the future of work that is increasingly upon us.
I typed "define engineer" into Google just now and it's AI result says: "An engineer is a professional who applies science, math, and technology to design, build, and solve practical real-world problems"
Is that not very close to what I and many others are doing with AI now? The agents are the technology that I'm using to design, build, and solve practical real-world problems.
We need to adopt the agent as the core engineering tool now to be most effective moving forward, so I believe that we should all consider ourselves agentic engineers or developers with an emphasis in our previously acquired skillsets.
Typically, when you say “X engineer”, X refers to the kinds of problems you solve. Civil engineer, mechanical engineer, electrical engineer, software engineer.
Unless you’re actually engineering agents, “agentic engineering” is not the correct term for what you’re doing.
Yes, one way to think of it that I think is very useful is that all software engineers who have embraced AI agents as the core tool through which they design, build, and solve practical real-world problems are agentic engineers...with an emphasis in software engineering.
The background in software engineering is incredibly useful, currently, especially as a means to judge the evidence that the agents present attesting to the correctness of the solution.
Do I feel bad about misusing the engineering term? Maybe...but am I misusing it? How do you define engineering? If using science, math, logic, and technology to solve practical real-world problems is engineering...then?
I believe that software engineering will be considered solved for the vast majority of current software development sometime in 2027.
I'm setting myself up to take advantage of the most useful tool ever created, the AI agent, to create in whatever way I want and is necessary in the future. Software engineering may not be necessary in the future, at least not for a lot of normal software application development we see today.
To me it's very much a development is dead, long live development type of situation.
Sure development is "solved" if you define it as producing code. I think it's solved already. It's incredible in a way but then I notice nothing changed: my boss still doesn't know anything. Stuff doesn't get done until I get on it. Responsibilities haven't moved, accountability is the same, if there are issues I'm sure some AI could take a look at it but who instructs the AI? My non-tech colleagues don't even know that's a question they could ask. The entire problem domain of what used to be my work is completely opened up yet at the same it's closed off as ever.
It reminds me of mathematics in a way. Sure, I have the most powerful machines in the history of the world assisting me but I can't make heads or tails of even (relatively) basic math no matter how much sci-fi tech you're giving me. I either need to A) defer to the machines completely B) have a mathematician on call.
I know AI's could perform B) technically, but that would be A) but I'm absolutely not sure we'll ever fully move on to A) until a lot of progress has been made / society is fundamentally restructured.
I'm curious when you started "focusing" on software engineering and how much XP you have in the field especially commercially.
You're not wrong if software development for you only means putting shiny things on the web. Luckily it's only a tip of the ...
> take advantage of the most useful tool ever created, the AI agent,
It's just cringe, One Direction young female groupie esq...
I have nothing against AI/Agent (use it and love), but some people, including myself, should recognise their AI psychosis before subjecting others to their delusions.
I think this is very interesting, but what are you specifying when you build a website?
I feel like this is the new "test driven development", which is totally a good idea, but also not totally useful for the actual process of building web apps, where most of your failure modes are weird CSS, JS, or server issues.
I don't know yet, but with Lasm I'll be spending a lot of effort finding out. And hopefully I'll get some help.
For me Lean is the key that has unlocked the idea of formal verification for general-purpose software engineering...before it was just an amazing technology that I couldn't practically touch in my profession and personal projects. That's changed now.
Thank you! Many more example apps with many more proofs are coming.
Also everything you've seen so far in Lasm at the Lean source code level has had almost no human honing from me. I'm a beginner to Lean, but I'll be honing the Conduit app and also bringing more example apps in.
The extent to which we can formally verify traditional web frontend and backend code is something I'm very excited to find out. My personal first goal is to be able to delete much of my test suites and replace them with Lean theorems.
Oh boy! My favorite thing about vibe coding is when agents write tests that assert whether certain strings exist in the codebase. Now we can formally verify that they do!
> I'm an agentic engineer with an emphasis in software engineering
This is odd. You wouldn't consider yourself a software engineer with an emphasis in...agentic engineering, or something like that? What is your background?
Hi. Well, sure I could consider myself a software engineer with an emphasis in agentic engineering...but that's not how I see the world now.
Check out this episode from my podcast to understand my thinking more: https://millillion.org/podcast/001-we-are-all-agentic-engine...
The extent to which AI is performing essentially all of my code generation and a large portion of the actual software engineering is astounding. My role has largely evolved into perfecting the process by which the agents are guided into well-architected solutions, manually testing their outputs, judging the artifacts that they present as evidence of clean code and robust implementation, and understanding business workflows well enough to guide the agents to a good working app that can be used by people.
AI is so integral to everything that I do professionally and in the software engineering space, it's outrageous. And I can apply those skills with extreme effectiveness to other areas outside of software engineering, such as tax preparation, legal document design, and my associates to CAD, energy flow analysis, structural engineering, etc.
I believe it is a very useful way to set ourselves up for our current reality and the future of work that is increasingly upon us.
I typed "define engineer" into Google just now and it's AI result says: "An engineer is a professional who applies science, math, and technology to design, build, and solve practical real-world problems"
Is that not very close to what I and many others are doing with AI now? The agents are the technology that I'm using to design, build, and solve practical real-world problems.
We need to adopt the agent as the core engineering tool now to be most effective moving forward, so I believe that we should all consider ourselves agentic engineers or developers with an emphasis in our previously acquired skillsets.
So were you a vim engineer or a sublime text engineer before AI?
Typically, when you say “X engineer”, X refers to the kinds of problems you solve. Civil engineer, mechanical engineer, electrical engineer, software engineer.
Unless you’re actually engineering agents, “agentic engineering” is not the correct term for what you’re doing.
what is this even supposed to mean?
Is every software dev who instructs agents regularly now an agentic engineer?
Don't people feel bad about misusing the engineering term so badly?
Yes, one way to think of it that I think is very useful is that all software engineers who have embraced AI agents as the core tool through which they design, build, and solve practical real-world problems are agentic engineers...with an emphasis in software engineering.
The background in software engineering is incredibly useful, currently, especially as a means to judge the evidence that the agents present attesting to the correctness of the solution.
Do I feel bad about misusing the engineering term? Maybe...but am I misusing it? How do you define engineering? If using science, math, logic, and technology to solve practical real-world problems is engineering...then?
It's even supposed to mean this: https://news.ycombinator.com/item?id=50302599
Looks more streamlined on the CV when agentic engineers pivot to an emphasis im plumbing, roofing, etc. once software is solved
I believe that software engineering will be considered solved for the vast majority of current software development sometime in 2027.
I'm setting myself up to take advantage of the most useful tool ever created, the AI agent, to create in whatever way I want and is necessary in the future. Software engineering may not be necessary in the future, at least not for a lot of normal software application development we see today.
To me it's very much a development is dead, long live development type of situation.
Sure development is "solved" if you define it as producing code. I think it's solved already. It's incredible in a way but then I notice nothing changed: my boss still doesn't know anything. Stuff doesn't get done until I get on it. Responsibilities haven't moved, accountability is the same, if there are issues I'm sure some AI could take a look at it but who instructs the AI? My non-tech colleagues don't even know that's a question they could ask. The entire problem domain of what used to be my work is completely opened up yet at the same it's closed off as ever.
It reminds me of mathematics in a way. Sure, I have the most powerful machines in the history of the world assisting me but I can't make heads or tails of even (relatively) basic math no matter how much sci-fi tech you're giving me. I either need to A) defer to the machines completely B) have a mathematician on call.
I know AI's could perform B) technically, but that would be A) but I'm absolutely not sure we'll ever fully move on to A) until a lot of progress has been made / society is fundamentally restructured.
I'm curious when you started "focusing" on software engineering and how much XP you have in the field especially commercially.
You're not wrong if software development for you only means putting shiny things on the web. Luckily it's only a tip of the ...
> take advantage of the most useful tool ever created, the AI agent, It's just cringe, One Direction young female groupie esq...
I have nothing against AI/Agent (use it and love), but some people, including myself, should recognise their AI psychosis before subjecting others to their delusions.
I think this is very interesting, but what are you specifying when you build a website?
I feel like this is the new "test driven development", which is totally a good idea, but also not totally useful for the actual process of building web apps, where most of your failure modes are weird CSS, JS, or server issues.
I would be happy to be proven wrong though
I don't know yet, but with Lasm I'll be spending a lot of effort finding out. And hopefully I'll get some help.
For me Lean is the key that has unlocked the idea of formal verification for general-purpose software engineering...before it was just an amazing technology that I couldn't practically touch in my profession and personal projects. That's changed now.
Interesting! Would've loved to see more frontend proofs, for example you have proofs about behavior but none on layout or design.
Thank you! Many more example apps with many more proofs are coming.
Also everything you've seen so far in Lasm at the Lean source code level has had almost no human honing from me. I'm a beginner to Lean, but I'll be honing the Conduit app and also bringing more example apps in.
The extent to which we can formally verify traditional web frontend and backend code is something I'm very excited to find out. My personal first goal is to be able to delete much of my test suites and replace them with Lean theorems.
Oh boy! My favorite thing about vibe coding is when agents write tests that assert whether certain strings exist in the codebase. Now we can formally verify that they do!
same! i noticed this a while ago and can't stop seeing it everywhere - https://github.com/legluondunet/SpiderMan2k1Fix/blob/7c6e032...
This is wild, I had no idea Lean could be used for things like this.
That's what hooked me on Lean. Lean is a general-purpose software development language, not just a formal verification language. This is powerful.
> AI agents: this README is complete on its own. llms.txt is the short index, and docs/AI.md lists the rules that save you from wrong turns.
The fall of technical writing.