The Counting House — Steve Yegge

kept by eddie oct 27

The Counting House pattern applies AI to high-volume customer service via stateless, policy-governed model invocations evaluated against human baselines using confusion matrices.

AWS in 2026: The Year of Proving They Still Know How to Operate - Last Week in AWS Blog
AWS is fine but faces real challenges: the AI gap is closing, talent attrition threatens operational excellence, and 2026 will test execution.
also on Bedrock, #ai
The age of vertical models is here (via @eoghan)
Intercom released a new vertical model for its Fin customer service agent, trained by its AI Group, marking its biggest technological advance in three years.
also on #customer-service, #enterprise, #ai
Be There for Every Customer With Meta Business Agent
Meta launches Business Agent AI for WhatsApp, Messenger, and Instagram, automating customer responses, sales, and operations for businesses of all sizes.
also on #customer-service, #ai-agents, #ai
How to Eval AI Agents — The 2026 Guide
Evaluating AI agents requires floor-raising error analysis, code-aware offline tests, production monitoring, and a tight feedback loop rather than benchmark-maxxing.
also on #evaluation, #ai-agents
OpenAI’s AGI number came from a harness, not the model
OpenAI’s 99.9% AGI score for GPT‑6 Astra relied on its custom Provider Adapter harness, while the benchmark’s standard harness yields only 62.7% for the same model.
also on #evaluation, #ai