Use cases

A question you ask.
A model you train.

Start with past examples and an outcome you can measure. Here are six ways to frame a decision task.

See the training workflow
01 / SalesProbability

Which deals will close?

Learn which patterns in past opportunities were followed by a closed deal.

Inputs
Call notes, deal stage, and customer history
Known outcome
Closed or didn’t close
A question to train for

“Will this deal close in the next 7 days?”

02 / Customer supportProbability or category

Who had a good experience?

Connect the support experience to the rating a customer eventually leaves.

Inputs
Onboarding context, questions, complaints, and resolution notes
Known outcome
The customer’s recorded review or rating
A question to train for

“Will this customer leave a 5-star review?”

03 / OperationsCategory

How severe is this incident?

Use past failures and their confirmed severity to help triage a new report.

Inputs
Error reports, affected services, and incident context
Known outcome
Confirmed SEV-0, SEV-1, SEV-2, or SEV-3
A question to train for

“Which severity category fits this incident?”

04 / RecruitingProbability

What leads to an offer?

Study past recruiting outcomes using interview evidence and work samples. Keep individual hiring decisions with your team.

Inputs
Interview transcripts and relevant work evidence
Known outcome
Recorded hiring outcome
A question to train for

“Which patterns in past interviews were followed by an offer?”

05 / Sports researchProbability or category

How might this matchup end?

Compare pre-game context with historical results to model a defined match outcome.

Inputs
Who played, previous matchups, and home or away context
Known outcome
The recorded game result
A question to train for

“How likely is the home team to win this baseball game?”

06 / Market researchProbability or score

Which signals preceded outperformance?

Explore how historical business and market conditions relate to a clearly defined outcome over a fixed period.

Inputs
Macro trends, past performance, and management changes known at the time
Known outcome
Performance against a defined benchmark
A question to train for

“Which companies outperformed their industry over the following quarter?”

The setup matters.

Use only information that would have been available before the outcome. Set aside independent test data and check the model on examples it hasn’t learned from.

These are illustrative decision tasks, not claims of measured performance. Available connectors are Attio, Pylon, and Fireflies. Talk to us about sourcing data for your use case.

Bring your own question.

Talk to us