Hacker News Reader

Comment by mpeg | Hacker News Reader

Honest question: what do you do with them? I would be so fascinated to see a video of this kind of workflow… I feel like I use LLMs as much as I can while still being productive (because the code they generate has a lot of slop) and still barely use the agentic CLIs, mostly just tab completion through windsurf, and Claude for specific questions by steering the context manually pasting the relevant stuff

View on HN

I focus more on reading code & prompting claude to write code for me at a high level. I also experiment a lot. I don't write code anymore by hand except in very rare cases. I ask claude for questions about the code to build understanding. I have it produce documentation, which is then consumed into other prompts. Often, claude code will need several minutes on a task so I start another task. My coding throughput on a day to day basis is now the equivalent of about 2-3 people.

I also use gemini to try out trading ideas. For example, the other day I had gemini process google's latest quarterly report to create a market value given the total sum of all it's businesses. It valued google at $215. Then I bought long call options on google. Literally vibe day trading.

I use chat gpt sora to experiment with art. I've always been fascinated with frank lloyd wright and o4 has gotten good enough to not munge the squares around in the coonley playhouse image so that's been a lot of fun to mess with.

I use cheaper models & rag to automate categorizing of my transactions in Tiller. Claude code does the devops/python scripting to set up anything google cloud related so I can connect directly to my budget spreadsheet in google sheets. Then I use llama via openrouter + a complex RAG system to analyze my historical credit card data & come up with accurate categorizations for new transactions.

This is only scratching the surface. I now use claude for devops, frontend, backend, fixing issues with embedder models in huggingface candle. The list is endless.

Can you share some code? I work with a guy like this who claims this level of output but in reality he consumes massive amounts of other devs time in PR review.

Are you doing a lot of broad throwaway tasks? I’ve had similar feelings when writing custom code for my editor, one off scripts, etc but it’s nothing I would ever put my professional reputation behind.

Sorry, most of my code is proprietary. However, I have a stock exchange project on my github I plan to rewrite in rust. I'm pretty busy now at work but I'll do that using claude code.

If your friend is consuming massive amounts of other dev time in PR reviews, maybe he has other issues. I'm willing to bet even without agentic coding, he would still be problem for your coworkers.

Sometimes I do broad throwaway tasks. For example I needed a rust lambda function that would do appsync event authorization for jwt tokens. All it needed to do was connect to aws secrets, load up the keys & check inbound requests. I basically had claude-code do everything from cdk to building/testing the rust function & deploying to staging. It worked great! However, I've certainly had my fair share of f-ups like I recently tried doing some work on the frontend with claude code and didn't realize it was doing useEffect everywhere!! Whoops. So I had to adapt and manage 2-3x claude code instances extremely closely to prevent that from happening again.

As a follow-up, I've gotten much much faster at modeling code in my mind and directly translating it into prompts. It really changes how you code! For each task, I'm extremely specific about what I want and depending on how closely claude does what I want, I change my specificity. Sometimes like with the lambda function, I can be high level and with my react.js codebase, due to it's lack of types (I know...) needs extra attention.

To be effective with agentic coding, you have to know when to go high level and low level. And have to accept that sometimes agentic coders need a lot of help! It all depends on how much context you give it.

I wouldn't blame AI. A terrible developer using AI becomes a bad developer. A good developer using AI becomes a great developer.

Do you use it to make any actual products that makes money?

yes. https://app.grantpuma.com/

I can’t read any text on homepage with iOS Safari as first few characters/words of each sentence is cut off. Text justification is different on different pages. Vertical spacing and images are uneven across the site. FAQ accordion animation is jittery and the hamburger menu doesn’t open actual menu from any other page other than homepage. Going back to homepage from FAQ page renders homepage in previous state, such as with sidebar open, momentarily before resetting to expected state. Focusing on input on sign in page zooms in on email field, but half of it is off page.

I’m assuming this is work in progress and currently it’s been vibe coded to MVP stage?

Reply on news.ycombinator.com