Use cases · Documents
One question does not need the whole file
A contract, a report, a policy — sent in full so the model can answer one line of it. You pay for every page it skims past.
What you hand over, what comes back
- Contracts
- Reports
- Policies
- Transcripts
- Your question
Scores every passage against your question, keeps what the answer needs, forwards one call on your key.
- The answer
- Standard JSON
- Streaming
- Token counts
How much goes depends on what you ask
"What is the notice period?"
A narrow question needs a narrow slice. This is where the numbers are largest.
"Everything about termination"
More survives, because more is genuinely relevant. Still a saving, just a smaller one.
"Summarise the whole thing"
Almost nothing is safe to drop — so almost nothing is. It declines rather than guess. See the measurements →
Connect it
Change the base URL. Keep your SDK, your model, your key.
# Python — OpenAI SDK client = OpenAI( base_url="https://compreslm.com/v1", api_key="YOUR_COMPRESLM_TOKEN", default_headers={"X-LLM-API-Key": "YOUR_PROVIDER_KEY"}, ) # The document goes in `context`, the question stays the message — # that separation is what makes the selection accurate. resp = client.chat.completions.create( model="gpt-4o-mini", messages=[{"role": "user", "content": "What is the termination notice period?"}], extra_body={"context": contract}, )
Nothing may be dropped? Pick the exact profile for that call — see profiles.
Run this exact case
A whole company handbook, one HR question, both answers side by side.
Talk to us
- Connect it — we'll get it running against your traffic and check the numbers with you.
- Fit it to you — closed network, unusual documents, a shape we don't handle yet.
- Build it — no product to connect yet? Our partners build it, we make it cheap to run.
Usually a reply the same working day.