Use cases · Support

Every ticket pays for the whole help centre

Nobody knows in advance which article answers the question, so the assistant is handed all of them. Every single time.

−75% Of a help centre, refund answer still exact
15.4k→ 3.9k Tokens reaching the model

What you hand over, what comes back

It works with your retrieval, or instead of it

Already using search?

Send the chunks it returned. Retrieval picks candidates; we drop the ones this question does not need.

No search yet?

Send the whole knowledge base and skip building an index — the question does the selecting.

Replies get shorter too

Turn on shorter answers and support replies stop rambling — customers read them, and you stop paying for padding.

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"},
)

# Articles go in `context`, the customer's words stay the message —
# that is what everything gets measured against.
resp = client.chat.completions.create(
    model="gpt-4o-mini",
    messages=[{"role": "user", "content": ticket_text}],
    extra_body={"context": knowledge_base, "compress_output": True},
)

compress_output is what keeps replies short. See the profiles page.

Run this exact case

A whole help centre, one customer asking about a refund, both answers side by side.

Try it — no account

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.