Technical deep-dives on agentic QA, AI testing, and running a company on 40 agents.
QAWave has no employees. One human, 40 specialized AI agents. Each business function — content, code review, competitive intelligence, QA delivery — has a dedicated agent with its own eval harness. Here's what works, what doesn't, and why the eval harness is the actual product.
Read moreWe swapped the NetworkGraph for a live 8-agent CI terminal, ran a 3-agent copy review that cut 6 banned words, and signed our first design partner LOI with a team in Brno running 12% flake rate. Numbers, lessons, and what's next.
Read moreThird iteration of the comparison table — this one frames each row as a customer question, not a feature dump. ROI calculator landed and 89 users ran the numbers in week one. We also deleted two homepage sections. Honest metrics inside.
Read moreTheridion moved to closed beta with 7 external testers — and they found keyboard-shortcut inconsistencies in 30 minutes that we'd missed for months. Agentic Moat became our top-traffic page with 4,200 reads in week one. And Applitools quietly raised seat pricing 22%.
Read more