# Vision & Positioning

## Why Edge exists

Banks and regulated enterprises that ship AI features face a hard requirement before any model touches a customer: **prove the model's quality, reproducibly, against a frozen golden dataset, with an audit trail**.

The market answer is usually one of:

* A SaaS evaluation tool that the bank's GRC team will not approve (data leaves the perimeter).
* An internal hack built on Jupyter notebooks (not reviewable by auditors).
* A research framework (DeepEval, ragas, etc.) wired to a homegrown UI (no workflow, no RBAC, no audit log).

**Edge** is the production-grade alternative: a single deployable that lets a business analyst build golden datasets, an ML engineer run evaluations with 24 metrics, and a GRC reviewer pull a signed evidence pack — all on-prem when needed, all open by design.

## Positioning

| Dimension        | Edge                                             | Cloud eval tools   | DIY notebooks         |
| ---------------- | ------------------------------------------------ | ------------------ | --------------------- |
| **Deployment**   | SaaS, on-prem, signed delivery                   | SaaS only          | Anywhere, but fragile |
| **Auditability** | Frozen datasets, item-level lifecycle, audit log | Limited            | None                  |
| **Workflow**     | Visual flow builder, RBAC, scheduled runs        | Code-only          | Code-only             |
| **Compliance**   | DORA / GDPR / ISO 27001 / EU AI Act mapped       | Provider-dependent | Manual                |
| **Cost model**   | One container; bring your own LLM                | Per-eval pricing   | Free, but TCO is high |

## Who Edge is for

* **Banks and insurers** deploying Gen-AI to customer-facing or revenue-critical surfaces.
* **GRC and Internal Audit** functions who need to read the evidence themselves.
* **Platform teams** who want one tool to evaluate every model their org ships.

## What Edge is not

* Not a model training platform — bring your own model or use an enterprise LLM gateway.
* Not a vector store — bring your own retrieval; Edge evaluates whatever you point it at.
* Not a logging system — Langfuse integration is opt-in; Edge does not replace observability.

## Roadmap signal

* **v1.1.0** — bank-ready baseline (current). See [Banking Readiness](/banking-readiness/overview.md).
* **v1.2.0** (planned, T+2 weeks) — `pb_migrations` refactor closes the last binary-in-git audit finding.
* **v2.0** — multi-tenant SaaS control plane on Cloudflare Workers + Durable Objects.


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.edge.nyami.fr/getting-started/vision.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.