Frontier AI just got cheap enough for the busywork

August brought a wave of new models — and a price war. The bottleneck for midsize businesses is no longer the AI. It's the wiring.

August was one of the busiest months for AI model releases in a long time — more than a dozen new or updated models in a few weeks, from Google, Anthropic and a crowd of challengers. The headline isn't any single model. It's the pricing.

Google's new Gemini 3.7 Flash launched with introductory pricing around 75 cents per million input tokens, aimed squarely at document work and business automation. Anthropic's latest Claude update pitched near-flagship performance at half the cost. Across the board, the price of “good enough to trust with real work” keeps falling.

For a midsize business, that changes the maths. A year ago, automating something like invoice follow-ups or inbox triage with a capable model was a line item worth debating. Today the model cost is usually the smallest part of the project.

The bottleneck has moved. It's no longer “can we afford the AI?” — it's connecting it to the systems your business already runs on, making it reliable, and making sure your team actually uses it. The model is the easy part now.

That's the part we do. If there's a repetitive job you've been meaning to get off someone's plate, the economics have probably never been better — and a discovery call will tell you quickly whether it stacks up.