From c60a3d84fd81ab4fcd279e28a7cbf1f02f8f2ef9 Mon Sep 17 00:00:00 2001
From: Sid Mohan <61345237+sidmohan0@users.noreply.github.com>
Date: Thu, 2 Jul 2026 11:33:54 -0700
Subject: [PATCH] Reapply "Merge pull request #151 from
 DataFog/docs/v5-roadmap"

This reverts commit 2119918023f49d88f54d9bdcf338fd13ea6bb54f.
---
 .gitignore |  1 -
 ROADMAP.md | 37 +++++++++++++++++++++++++++++++++++++
 2 files changed, 37 insertions(+), 1 deletion(-)
 create mode 100644 ROADMAP.md

diff --git a/.gitignore b/.gitignore
index 1316a1f3..cf11a42c 100644
--- a/.gitignore
+++ b/.gitignore
@@ -69,5 +69,4 @@ docs/*
 */**/__pycache__/
 
 notes/benchmarking_notes.md
-Roadmap.md
 notes/*
diff --git a/ROADMAP.md b/ROADMAP.md
new file mode 100644
index 00000000..1f10a519
--- /dev/null
+++ b/ROADMAP.md
@@ -0,0 +1,37 @@
+# DataFog Roadmap
+
+_Last updated: July 2026 — following the [v4.5.0 stable release](https://github.com/DataFog/datafog-python/releases/tag/v4.5.0)._
+
+DataFog's direction for v5: **the fastest, easiest offline PII firewall for AI apps, logs, and datasets** — no network calls, no model downloads on import, one obvious API.
+
+Independent benchmarking ranks DataFog fastest among Python PII libraries by orders of magnitude. The v5 cycle invests that speed advantage into precision, EU coverage, and a leaner package.
+
+## v5.0.0 themes
+
+### 1. Precision: validators, confidence scores, strictness presets
+
+Speed means nothing if you can't trust the findings. v5 adds a zero-dependency validation pass on every hit (Luhn for credit cards, IBAN mod-97, SSN structure rules, IP plausibility), confidence scores on findings, and strictness presets (`strict` / `balanced` / `lenient`) so you pick the precision/recall tradeoff instead of hand-filtering. We'll publish precision/recall benchmarks alongside the speed benchmarks.
+
+### 2. EU language & entity coverage
+
+v4.5.0 introduced German locale support. v5 generalizes it into a locale pack system — new languages become data plus tests, not code changes — starting with locale-independent EU entities (IBAN, VAT IDs, national ID formats) and expanding across FR/ES/IT/NL/PL through the v5.x line.
+
+### 3. A leaner package
+
+The core install stays pydantic-only and tiny. The heavyweight paths are being cut or spun out: the Donut/transformers OCR path goes away (pytesseract remains), the PySpark wrapper becomes a documented recipe, and legacy duplicate modules are deleted. CI enforces wheel-size and import-time budgets.
+
+### 4. Built for pipelines, not just scripts
+
+The core `scan`/`redact` functions are pure, stateless, and thread-safe, with batch/iterator APIs for high-throughput use. Instead of shipping transport connectors, we're publishing recipes for embedding DataFog in Kafka consumers, Vector/Fluent Bit transforms, and OpenTelemetry collector processors.
+
+### 5. Vault-friendly anonymization
+
+Deterministic tokenization with exportable mappings and format-preserving pseudonymization, so DataFog slots into vault-and-token architectures rather than competing with them.
+
+## API direction
+
+`scan()`, `redact()`, and `protect()` — shipped as previews in v4.5.0 — become the primary documented API in v5.0.0. The legacy `detect()`/`process()` functions keep working as compatibility shims throughout the v5.x line.
+
+## Feedback
+
+Roadmap priorities are shaped by user feedback — open a [GitHub issue](https://github.com/DataFog/datafog-python/issues) or join the [Discord](https://discord.gg/bzDth394R4).