# AI use cases for Awesome Decision Models (AnotiaWang/awesome-decision-models)

Repo: [AnotiaWang/awesome-decision-models](https://github.com/AnotiaWang/awesome-decision-models) | Use cases: 4 | Page: https://dedupedai.com/use-cases/anotiawang-awesome-decision-models/

A curated, community-maintained list of decision models (models that return option probabilities instead of generated text), with the hosted APIs, open-weight models, runtimes, SDKs, benchmarks, and papers around them.

This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL.

- Route support tickets with a hosted decision API: https://dedupedai.com/use-cases/anotiawang-awesome-decision-models/route-support-tickets-with-hosted-decision-api/
- Run a small open decision model locally: https://dedupedai.com/use-cases/anotiawang-awesome-decision-models/run-small-open-decision-model-locally/
- Check agent tool calls before they run: https://dedupedai.com/use-cases/anotiawang-awesome-decision-models/check-agent-tool-calls-before-they-run/
- Flag risky behavior in AI traces with Span-01: https://dedupedai.com/use-cases/anotiawang-awesome-decision-models/flag-risky-behavior-in-ai-traces/

## How to route support tickets with a hosted decision API with Awesome Decision Models

Repo: [AnotiaWang/awesome-decision-models](https://github.com/AnotiaWang/awesome-decision-models) | Difficulty: Intermediate | Time: 1 hour | Basis: From the README

Send a ticket and your label set to a hosted decision model and get a probability for each label.

**Who it is for:** A support engineer or ops lead who wants a triage step that assigns each incoming ticket to one queue without writing a classifier from scratch.

**The problem:** Manual triage is slow, and keyword rules miss ambiguous tickets. Generated text from a chat model is hard to turn into a reliable label.

**The outcome:** Each ticket gets a probability for every queue you define, so you can route the top label and send low-confidence tickets to a person.

### Steps

1. Open the Hosted APIs section of the repo and pick a model; Jev and Microsoft-Decision-1 both describe a Choice question over text.
2. Create an API key in the provider console (for Jev, the TypeSafe console linked from the repo).
3. Write your queue names as the Choice options and the ticket text as the state.
4. Send one request to the provider's /v1/systemone endpoint and read back the probability for each option.
5. Route tickets whose top probability is above your threshold, and send the rest to a human queue.

### Prompt to paste into your agent

```text
Using the Hosted APIs section of github.com/AnotiaWang/awesome-decision-models, pick one hosted decision model and call its /v1/systemone endpoint with the ticket text in <ticket-file> and these Choice options: billing, bug, account, other. Expected output: one probability per option that sums to about 1. Verify that the response contains a probability for every option you sent.
```

### Caveat

The list is community-maintained and not affiliated with any provider, and the README says benchmark numbers are self-reported on each project's own chosen benchmark, so test on your own tickets before relying on any model.

Risk, credentials: The hosted APIs need a provider API key, so store it in an environment variable or secret manager and do not paste it into prompts or shared files.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/anotiawang-awesome-decision-models/route-support-tickets-with-hosted-decision-api/

## How to run a small open decision model locally with Awesome Decision Models

Repo: [AnotiaWang/awesome-decision-models](https://github.com/AnotiaWang/awesome-decision-models) | Difficulty: Intermediate | Time: 1 hour | Basis: From the README

Pull a small open-weight decision model into Ollama and score your own options on your own machine.

**Who it is for:** A developer who wants to keep input data on a local machine and avoid per-request API costs while testing decision questions.

**The problem:** Hosted decision APIs send your state text to a provider, and you may need a local option for privacy or offline testing.

**The outcome:** A local model that answers Choice, Score, or Noul questions and returns probabilities, served from your own Ollama install.

### Steps

1. Open the Open Models section and pick a model that the README says runs in Ollama, such as Nimble (listed as nimble) or Tev1 (listed as tev1).
2. Pull the model with `ollama pull nimble` after checking the model name in the README. (inferred, check the README)
3. Read the model card on Hugging Face or GitHub for the expected request format and any license terms before using it.
4. Send a state and your options to the local model and compare its probabilities against a few cases you already know the answer to.

### Prompt to paste into your agent

```text
Using the Open Models section of github.com/AnotiaWang/awesome-decision-models, pick one model that the README lists as available in Ollama and run it locally on the sample state in <sample-file> with the options in <options-file>. Expected output: one probability per option. Verify the license of the chosen model in its model card before you use the output in a product.
```

### Caveat

Scores in the Open Models section are as reported by each project on the benchmark it chose, so they are not comparable across entries and the local model may behave differently on your data.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/anotiawang-awesome-decision-models/run-small-open-decision-model-locally/

## How to check agent tool calls before they run with Awesome Decision Models

Repo: [AnotiaWang/awesome-decision-models](https://github.com/AnotiaWang/awesome-decision-models) | Difficulty: Advanced | Time: 2 hours | Basis: From the README

Score a proposed tool call against your own allowed options before the agent executes it.

**Who it is for:** An engineer building a Strands agent who wants a fast check on each tool call, using a model that runs on a local GPU or CPU.

**The problem:** Agents can call the wrong tool or run a risky action, and a second full model call on every step adds latency and cost.

**The outcome:** A probability for each allowed action on each proposed tool call, so the agent can block or ask for approval on low-probability actions.

### Steps

1. Open the Open Models section and read the Strands Decider 2B entry, which describes checking Strands agent tool calls before they run.
2. Clone the project from the strands-labs/strands-decider GitHub link and read its setup instructions.
3. Define your allowed actions as the options and the pending tool call as the state.
4. Add a check in your agent loop that reads the option probabilities and blocks calls below your threshold.
5. Log blocked and allowed calls and review them against real runs.

### Prompt to paste into your agent

```text
Using github.com/strands-labs/strands-decider linked from github.com/AnotiaWang/awesome-decision-models, set up the Strands Decider 2B model and score the pending tool call in <tool-call-file> against the allowed actions in <allowed-actions-file>. Expected output: a probability per allowed action and a block or allow decision. Verify that a call outside the allowed list gets a low probability on every option.
```

### Caveat

The README describes this model's tool-call use only through the launch post, so confirm the setup steps in the strands-decider repo itself before wiring it into an agent.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/anotiawang-awesome-decision-models/check-agent-tool-calls-before-they-run/

## How to flag risky behavior in AI traces with Span-01 with Awesome Decision Models

Repo: [AnotiaWang/awesome-decision-models](https://github.com/AnotiaWang/awesome-decision-models) | Difficulty: Intermediate | Time: 45 minutes | Basis: From the README

Label each AI trace for behaviors you define, such as prompt injection or agent loops.

**Who it is for:** An AI product engineer reviewing agent traces who needs to find prompt injection, hallucination, or loops without reading every trace by hand.

**The problem:** Reading traces manually does not scale, and keyword search misses many behaviors described in plain language.

**The outcome:** Each trace gets present, absent, or not observable for every behavior you define, with a list of traces to review first.

### Steps

1. Open the Hosted APIs section and read the Span-01 entry, which classifies AI traces for behaviors you define in plain language.
2. Get an OpenRouter API key and find the Span-01 page at openrouter.ai/respan/span-01 linked from the README.
3. Write one plain-language definition for each behavior you want to flag, such as prompt injection, hallucination, or agent loops.
4. Run the classifier over a sample of traces and save each result with its trace ID.
5. Review every trace marked present, and spot-check a sample marked absent.

### Prompt to paste into your agent

```text
Using the Span-01 entry in github.com/AnotiaWang/awesome-decision-models, classify the traces in <traces-folder> for prompt injection and agent loops through the OpenRouter route the README links. Expected output: one present, absent, or not observable value per behavior per trace. Verify by reading three traces marked present and confirming the behavior is actually there.
```

### Caveat

The README gives Span-01 no accuracy numbers, so check its labels against traces you have already reviewed before trusting it on a larger set.

Risk, credentials: The OpenRouter key grants access to your account, so keep it in an environment variable and do not commit it to the trace review repo.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/anotiawang-awesome-decision-models/flag-risky-behavior-in-ai-traces/