> ## Documentation Index
> Fetch the complete documentation index at: https://docs.fincelo.app/llms.txt
> Use this file to discover all available pages before exploring further.

# Contract Intelligence

> How Fincelo's AI extracts, learns from, and manages your customer contracts.

## Overview

Fincelo reads every contract you upload, extracts all commercial terms with a confidence score per field, and gets smarter with every correction your team makes. No templates required — it works with any contract format.

***

## The 14 Extracted Fields

| Field              | Example               | Why It Matters            |
| ------------------ | --------------------- | ------------------------- |
| Start Date         | 1 April 2026          | Revenue recognition start |
| End Date           | 31 March 2027         | Renewal trigger date      |
| Contract Value     | ₹12,00,000            | ARR calculation           |
| Billing Frequency  | Quarterly             | Invoice schedule          |
| Payment Terms      | Net 30                | Dunning trigger           |
| Currency           | INR                   | FX treatment              |
| Discount %         | 15%                   | Net revenue calculation   |
| Auto-Renewal       | Yes — 30-day notice   | Renewal agent trigger     |
| Price Per Seat     | ₹5,000/seat           | Expansion billing base    |
| Notice Period      | 30 days               | Termination tracking      |
| Renewal Terms      | Annual, same terms    | Renewal playbook          |
| Termination Clause | Convenience, 30 days  | Churn risk flag           |
| SLA Terms          | 99.9% uptime          | Compliance monitoring     |
| Usage Limit        | 1,00,000 API calls/mo | Usage alert threshold     |

***

## Uploading Contracts

1. Fincelo → **Contracts → Upload**
2. Upload PDF (any format — scanned or digital)
3. AI extracts all 14 fields with confidence scores
4. Review each field — correct any misses
5. Click **Confirm** → contract activated

<Info>
  Start with your top 10 customers by ARR. The AI learns fastest when it sees corrections on real contracts early.
</Info>

***

## Confidence Scoring

Every field gets a blended confidence score:

```
Confidence = 60% Claude model certainty + 40% historical field accuracy
```

* **≥ 80%** — High confidence, shown pre-selected
* **30–79%** — Medium confidence, shown for review
* **\< 30%** — Low confidence, flagged for manual entry

***

## Self-Improving AI

Every correction you make teaches the AI for next time:

1. AI extracts field incorrectly
2. You correct it
3. Fincelo stores the correction as a few-shot example
4. Next similar contract — correction is injected into the AI prompt
5. Field accuracy improves permanently

No ML retraining. Prompt augmentation only. Accuracy compounds over time.

***

## Ratchet Clauses & Special Terms

Fincelo detects:

* **Ratchet clauses** — minimum revenue guarantees that prevent contraction
* **Floor quantities** — minimum seat/usage commitments
* **Price escalation clauses** — annual price increases built into the contract
* **Most-favoured nation (MFN)** clauses

These are flagged in the contract record and fed to the Revenue Anomaly Agent.

***

## Contract States

| State               | Meaning                     |
| ------------------- | --------------------------- |
| Draft               | Uploaded, not yet confirmed |
| Active              | Confirmed, billing running  |
| Renewal In Progress | Within renewal window       |
| Grace Period        | Expired, renewal expected   |
| Terminated          | Ended — churn recorded      |
| Suspended           | Billing paused              |