Category
5 articles· All content in this category
We spent a month using four of the top AI coding assistants on production work. Here's what each one is actually good at, where they break, and which to pick for your team.
Local models finally caught up to something useful. Here's a real-world walkthrough of getting Llama 3, Qwen, and others running on consumer hardware — and when it's worth doing versus staying in the cloud.
We ran the same 50 queries through each of them. The results are messier — and more interesting — than the marketing suggests.
Reasoning models changed the rules. A lot of the old advice ('be very specific,' 'use few-shot examples') is now outdated or actively harmful. Here's what we've learned.
Strip the hype away and 'agents' are LLMs in a loop with tool access. That's powerful — but only in specific shapes of problem. Here's our real-world take.