Blog

Claude Haiku 5.5 explained: Anthropic's fastest, cheapest model, its specs, effort settings, tiered pricing, what changed from Haiku 4.5, and how to start using it.

Kodif AI vs Gorgias AI Agent compared: where each agent lives, what each one bills, what it really costs to run Kodif on top of Gorgias, and which teams should pick which.

A Kodif AI review from public docs, 16 case studies and every readable G2 review: strong actions and cancel saves, but setup and edits lean on Kodif's own engineers.

Kodif pricing is a flat annual rate: 24,000 conversations at $1 on Starter, down to $0.75 on Enterprise. Here's what that works out to per month and per resolution.

My Claude Dashboards review: how the SQL-per-chart design works, which plans get it, what sharing really does to your numbers, and when to keep your BI tool.

EmbeddingGemma 2 puts text, code, images, audio and video in one search index on a phone. What changed from v1, what the benchmarks hide, and when to use it.

Cactus Whistle is a free 16.9 MB speech-to-text model that runs on a phone CPU. Here's where it beats Whisper base, where it loses, and what to build with it.

My Mistral Large 4 review: a 4x jump over Large 3, a mid-pack score next to rivals, a 41.9% hallucination rate, and two niches where it is the right pick.

Google's Gemini agent plans work, spins up sub-agents, and can join your team with its own email address. Here is how it works, what it costs, and where it stops.