Our AI practice has always argued for model-agnostic architecture in client work, so it made sense to hold ourselves to the same standard internally. Anthropic's Claude models are now a first-class option in our stack, evaluated and swapped in per project based on cost, latency, and task fit rather than default habit.
In practice this means client engagements get the model that fits the job: one provider for long-context document work, another for tool-calling agents, chosen after evaluation rather than convention.
This is one of several partner additions this quarter as we formalize the vendor relationships behind projects we already ship.