I manually wrote a "bad" spec, asked it for feedback, improved spec until the problem, the solution and overall implementation design were clear and had a very high level of detail and were trying to do exactly what I needed. The lots of thinking, reading and manual editing helped me understand the problem way better than where I began from.
New session: Fed the entire spec, asked to build generic scaffolding only.
New session: Fed the entire spec, asked to build generic TEST scaffolding.
New session: Extract features to implement out of spec doc into .md files
New session: Perform research on codebase with the problem statement "in mind", write results to another .md. Performed manual review of every .md.
New session(s): Fed research and feature .md and asked for ONE task at a time, ensuring tests were written as per spec and keep iterating until they passed. Code reviewed beginning with test assertions, and asked for modifications if required. Before commit, asked to update progress on .md.
Ended up with very solid large project including a technology I wasn't an expert on but familiar, that I would feel confident evolving without an agent if I had to, learned a lot in the process. It would've taken me at least 2 weeks to read docs about it and at least another 3 to implement by hand; I was done in 2 total.
I explore writing ideas using various LLM tools,
This helps me build the text structure and gather ideas and information in one place.
I review everything, of course, and this is not just an automated process.
Same for this post,
I am combining my own natural tone while keeping some segguestion, like this line, that has a super "Marketing feel," and for that, the "GPT tone" is noticeable.
But I think this is nice information to put in this post, so I kept it
All my replies are manually typed, and I use Grammarly to fix typos and improve sentence structure.
"detect unauthorized interference with the Mobile Banking application"
I wonder if this has become a feasible avenue for scammers to interfere via other apps they could convince someone to install on rooted phones. Or if they are worried about skilled people being able to debug/MITM and find vulnerabilities on the banks.
Though from that statement alone, sounds more of a measure to protect banks than customers.
I’ve been running these sorts of training exercises w gpt5 and it’s been quite insightful. Not for mgmt but general senior-staff level communication. That even has flexibility to explain better context on my specific situation and role.
You nailed the trade-off. LLMs are incredible for open sparring, but they often work best if you already know the underlying principles to guide the roleplay.
I view this tool as the 'Drills' to learn those heuristics, so you can then go to GPT and practice them with your specific context. Appreciate the honest signal on pricing.
New session: Fed the entire spec, asked to build generic scaffolding only. New session: Fed the entire spec, asked to build generic TEST scaffolding. New session: Extract features to implement out of spec doc into .md files New session: Perform research on codebase with the problem statement "in mind", write results to another .md. Performed manual review of every .md. New session(s): Fed research and feature .md and asked for ONE task at a time, ensuring tests were written as per spec and keep iterating until they passed. Code reviewed beginning with test assertions, and asked for modifications if required. Before commit, asked to update progress on .md.
Ended up with very solid large project including a technology I wasn't an expert on but familiar, that I would feel confident evolving without an agent if I had to, learned a lot in the process. It would've taken me at least 2 weeks to read docs about it and at least another 3 to implement by hand; I was done in 2 total.
reply