What we build in the Claude API
Our usual choice for long-document work, structured extraction and anything where the output has to be reliably shaped. Used narrowly. Most of a good automation is deterministic, with a model doing one specific job inside it.
The Claude API is our usual choice for long-document work, structured extraction and anything where the output has to be reliably shaped. Long context windows suit contract, tender and specification review; structured output support makes typed, validated extraction reliable; and it is comparatively good at declining to answer when the information is not present. We use it narrowly, with the model doing one specific job inside an otherwise deterministic workflow, and only when the input is unbounded natural language.
We do not reach for it when a sufficiently detailed rule would handle the job, when a user is waiting synchronously, because latency is seconds rather than milliseconds, or where it would be verifying its own output. It is priced per token, input and output, varying by model and paid to Anthropic directly with no margin from us. A triage agent handling a few hundred messages a day usually lands in the tens of dollars a month.
What it is, and where it sits
Anthropic's Claude API is one of two AI providers we build with. We choose per task on measured accuracy and cost rather than preference, and we build so the model can be swapped without rewriting the workflow around it.
Where Claude tends to earn its place: long context windows suit contract, tender and specification review; structured output support makes typed, validated extraction reliable; and it is comparatively good at declining to answer when the information is not present, which reduces invented answers.
What we build in it
- Long-document review against your checklist, with clauses quoted and located rather than summarized away
- Structured extraction from unstructured documents into typed, validated fields
- Inbound message classification and detail extraction for triage
- Drafted replies grounded in your own documentation, queued for human approval
- Internal knowledge assistants answering from your content with citations
- Meeting and call summarization into decisions, owners and dated actions
How it is priced
Priced per token, input and output, varying by model. You pay Anthropic directly; we take no margin.
| Model | What you pay for | Best when |
|---|---|---|
| Per token | Tokens in and tokens out, at a rate that varies by model. | Always. This is the whole bill, and model selection per task is the lever that moves it. |
| A smaller model | The same per-token billing at a lower rate. | Classification and extraction, which often run well on it. |
| A larger model | A higher per-token rate, and more of them on a long input. | Genuine reasoning over long context. Long-document review is the expensive case, because a 60-page tender is a lot of input tokens. |
For a typical SMB workload the numbers are smaller than people expect. A triage agent handling a few hundred messages a day usually lands in the tens of dollars a month. Long-document review is worth modeling if you process tenders at volume.
When to choose it, and when to think twice
Same test as always: could a sufficiently detailed rule handle this? If yes, use a rule. Reach for a model only when the input is unbounded natural language. See agent or workflow?
Structured output with validation, confidence thresholds routing uncertain cases to a person, an evaluation set built from your real historical data and measured before launch, full logging of input, output and confidence, and human approval on anything consequential. The same guardrails every time; all six are on the AI agents page.
We hold no reseller agreement with this vendor, take no affiliate commission, and do not resell licenses. You buy your own subscription directly. It is the only arrangement under which our recommendation means anything.
Want the cost model run on your numbers?
Tell us your workflows and volumes and we will show you the three-year cost across every option.
You will talk to Fayez or Kareem. Nobody else.