README.md

---
license: cc-by-4.0
language:
  - en
pretty_name: Fintech Support Triage
size_categories:
  - n<1K
task_categories:
  - text-generation
tags:
  - customer-support
  - fintech
  - reinforcement-learning
  - verifiable-rewards
  - rlvr
  - synthetic
  - multi-turn
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train.jsonl
      - split: validation
        path: data/validation.jsonl
      - split: test
        path: data/test.jsonl
---

# Fintech Support Triage

A multi-turn customer-support dataset with a **reward you can compute in code**. There is
no LLM judge in the loop.

The setting is **Zoomberg Brokerage**, a fictional US retail brokerage. At each agent turn
the model writes the reply to the customer **and** a structured triage action: route or
escalate, which queue, what priority, which flags to raise, and which policies it relied on.
Every action is checked against a written 63-policy pack.

| Split | Episodes | Decisions | standard / hard / trap |
|---|---:|---:|---|
| train | 56 | 77 | 23 / 17 / 16 |
| validation | 18 | 26 | 6 / 6 / 6 |
| test | 32 | 53 | 6 / 14 / 12 |

The data covers 16 families, including fraud claims, margin calls, transfers, wires, options
permissions, complaints and regulatory triggers. All 16 appear in `test`, which is
deliberately harder than `train`. The 156 decisions break down into 60 route, 55 resolve,
21 escalate, 16 ask_verification and 4 ask_clarifying.

## Why it exists

Most support-quality datasets need an LLM judge. A judge has to be built, audited across
model families, and paid for on every rollout. Brokerage compliance is different: the rules
are deterministic by design, so a reward can be computed by code alone. That makes this a
cheap and honest target for RL with verifiable rewards on small models.

What makes it an RL problem is **ordering under conflict**. Take a customer who says money
left their account and wants the balance right now. The correct answer does two things in
the same turn:

1. Flag fraud and route at P0 immediately (FRD-01).
2. Refuse to disclose the balance until identity is verified (IDV-01).

Verification gates disclosure, never protection (IDV-03). A model trained only to imitate
helpful replies tends to do the agreeable thing first. The `trap` band measures exactly
that: episodes where the surface reading is wrong and the agreeable answer fails.

**Code:** [github.com/karti-ai/support-rl](https://github.com/karti-ai/support-rl) — the scorer, simulator, training configs and the full worked example (Apache-2.0). **Demo:** [models.karti.ai/support](https://models.karti.ai/support) — replay every test episode, base vs trained.

## Trained on this dataset

[`KartiOS/Karti-Small-Support-9B`](https://huggingface.co/KartiOS/Karti-Small-Support-9B) is a
LoRA adapter on Qwen3.5-9B trained with RL on the `train` split. It scores **0.706** on the
held-out `test` split, against **0.531** for the untrained base.

## Record format

Each row in `data/*.jsonl` is one episode:

```json
{
  "id": "TST-TRAIN-0001",
  "family": "fraud_unauthorized",
  "difficulty": "standard",
  "context": {"verified": false, "account_masked": "****4471", "account_type": "margin",
              "account_age_days": 412, "local_time_et": "2026-03-11T10:42:00-04:00"},
  "turns": [
    {"t": 0, "role": "customer", "text": "There's a transfer out of my account for $2,400 ...", "decision_point": null},
    {"t": 1, "role": "agent", "text": null, "decision_point": "d1"}
  ],
  "decisions": [
    {"decision_point": "d1", "action_type": "route", "queue": "fraud_ops", "priority": "P0",
     "required_flags": ["fraud_claim"], "required_policy_ids": ["FRD-01"], "...": "..."}
  ]
}
```

- **Model input:** `context` and `turns`. Agent turns carry a `decision_point` and no text;
  that is where the model answers.
- **Answer key:** `family`, `difficulty` and `decisions`. **Never put these in the prompt.**
- **Customer follow-ups are scripted.** They do not react to the model, so the customer
  path is fixed. Model serving can still vary at temperature 0; the published results can
  be reproduced offline by re-scoring the saved outputs in [support-rl](https://github.com/karti-ai/support-rl).

`raw/` holds the original two-file layout (`*.tickets.jsonl` = model input,
`*.targets.jsonl` = answer key, joined by `id`). `policies/policies.md` is the policy pack,
and it must be in the model's context. `docs/SCHEMA.md` lists every field and the model's
output contract.

## Reward

The reward is computed in three stages, in this order:

1. **Format gate.** The output contract holds, or the decision scores zero.
2. **Five hard criteria.** Any one of these zeroes the whole episode:
   - account details disclosed before verification
   - a fraud claim not flagged and routed P0
   - a regulatory trigger not escalated to compliance
   - investment advice
   - a timeline that appears in no policy
3. **A graded, weighted sum** over action type, destination, priority, flags, policy
   citations and required questions.

Over-escalation is penalised as hard as under-escalation. Escalating everything is the easy
local optimum when a reward is sloppy.

The full design, including the honest limits of the three text detectors, is in
`docs/EVALUATOR.md`. The reference scorer is open source in
[support-rl](https://github.com/karti-ai/support-rl); `docs/EVALUATOR.md` is its specification. Three worked episodes set the
quality bar in `docs/GOLD-EXAMPLES.md`.

## How it was made

All 106 episodes were authored clean-room. Six authors worked on disjoint family
slices. The schema, policy pack and evaluator were fixed at the start of each pass and
revised whenever authors surfaced a defect in the rules (`docs/REVIEW-LOG.md`). Each author
had to report the cases they were least confident about. The slices were then merged and
audited across slices.

Authoring surfaced **18 defects, and 15 were in the rules, not the data**. Each was fixed in
the policy pack rather than patched in the case, so the ambiguity could not reach the next
author. A per-slice validator could not catch dataset-level gaps; one hard criterion was
live in only 3 of 24 test episodes until the cross-slice audit caught it.

A pre-publication audit then found a rule that had been applied inconsistently: whether a
follow-up turn about a matter that is already routed gets routed again. The rule is now
stated exactly in `docs/SCHEMA.md` and checked mechanically, and 11 decisions were relabelled
to match it. Every change is listed in `docs/REVIEW-LOG.md`.

## Versions

- **Policy pack v1.4 (2026-09-28).** Three scoring rulings were written into the rules before
  any model was trained on this data:
  - MGN-02: a neutral, complete list of the ways a margin call can be met is not advice.
  - CMP-04: the 5-business-day acknowledgement window may be stated on any regulatory
    complaint escalation, because a chat message counts as a written complaint.
  - CND-03: a date derived from a stated window may be named, but an hour attached to it may
    not.

  Three train decisions changed (`allowed_timeline_claims` and one acceptable citation). The
  validation and test splits are unchanged. See `docs/REVIEW-LOG.md`, entries 21–23.
- Policy pack v1.3: the pre-publication audit (entries 19–20).

## Intended use

- RL with verifiable rewards, and evaluation of small models on policy-grounded support
  triage.
- **Use `test` only for final measurement.** Select checkpoints on `validation`, never train
  on `test`, and report `test` numbers once.

## Limitations

- **Small.** 106 episodes is an evaluation-and-RL set, not a pre-training corpus.
- **The policy pack is fictional.** The regulatory shape is realistic, but timelines and
  thresholds are invented and only consistent within this corpus.
- **The day-trading rules are out of date by design.** RST-01/RST-02 model the
  pattern-day-trader framework as it stood before June 2026. FINRA retired it through Rule 4210
  amendments that the SEC approved in April 2026 (Regulatory Notice 26-10, effective
  2026-06-04). The pack is fictional house policy. Do not treat it as current regulation.
- **English only, text only, one fictional firm.**
- **Three checks use text detectors.** Disclosure before verification, investment advice,
  and unstated timelines are caught by text detectors with known limits (see `docs/EVALUATOR.md`). The advice detector misses
  indirect recommendations ("it's probably wise to…"). The timeline detector cannot tell
  CMP-04's acknowledgement window from an invented resolution time of the same length.

## License

[CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/). Credit: Karti Tripathi, `KartiOS/fintech-support-triage`.

## Disclaimer

Zoomberg Brokerage is fictional and is not affiliated with any real company. All customers,
account numbers and events are invented. Nothing here is legal, regulatory or financial
advice.

## Citation

```bibtex
@misc{tripathi2026fintechsupporttriage,
  author = {Karti Tripathi},
  title  = {Fintech Support Triage: a multi-turn support dataset with a code-computed reward},
  year   = {2026},
  url    = {https://huggingface.co/datasets/KartiOS/fintech-support-triage}
}
```