Skip to content
Metatext Decide — typed decisions for your agents, with a probability on every answerGet early access
Metatext
Sign Up
Sales & SDR

Classify every reply and score every lead — in code, with probabilities you can weight.

Reply triage is a closed set of categories; lead fit is four ordinal scores you weight in code. Both are what a System One model does natively, and neither needs a paragraph of generated reasoning.

Get early access See the benchmark

In development — the benchmark numbers on this page are measured; the Decide-1 model is in training.

The problem

Replies are sorted by hand

Or by an LLM that invents a fifth category.

Scores without numbers

“Strong fit” from a chat model is not weightable.

Nobody trusts the score

Because nothing measures whether a 0.8 was right eight times in ten.

How it works with Decide
  1. 1

    One choice per reply

    Over your categories; route removal requests and questions immediately.

  2. 2

    Four scores per lead

    industry fit, company size, stated pain, buying signal — each 0–3, weighted in code.

  3. 3

    Calibrate on outcomes

    Closed-won is the label. A temperature per question type keeps the scores honest.

r = client.system_one(state=reply, questions={
  "category": {"type": "choice", "instructions": "How did they reply?",
               "criteria": {"interested": "...", "not_interested": "...", "wrong_person": "...",
                            "out_of_office": "...", "remove": "...", "question": "..."}},
  "buying_signal": {"type": "score", "instructions": "Buying intent", "criteria": ["none","curious","evaluating","ready"]},
})
  • Choice — pick one option, a probability on each
  • Score — a position on your ordered rubric
Something you can check

Pattern from the maximem.ai decision-seams map (sales / SDR section). Calibration matters most here: the benchmark’s open encoder went from ECE 0.194 to 0.020 with one fitted temperature per type on 1,000 held-out judgments.

Numbers about Jev are its authors' or the cited source's; numbers from the benchmark are ours, on one harness, with the raw JSON published. Full table.

Get early access — and tell us which decision you'd hand over first

We're building Metatext Decide with the first teams that use it. Leave your email, answer as much as you like, and say yes to a short interview if you want a say in what ships.

  • Early access to Decide as it opens up
  • A 20-minute interview if you want one — research, not a pitch
  • The benchmark harness and raw results, today
  • No spam — early-access updates only
Where do decisions happen in your stack today?
What kind of decisions?
Have you tried Jev?
What stops you using a hosted decision API?
Would you pay?
Open to a 20-minute interview?

Research, not a pitch — we want to hear how you decide today.

Early-access updates only. No spam, unsubscribe any time.

Questions

Can I add a category later?
Yes — the options are in the request, not in the weights.

Related