Anna AI-Native 2026 · Developer Tools

DocDrift

Your docs lie the moment code changes. DocDrift catches them — and drafts the fix.

Anna App · schema 2 AES-GCM-256 Human-in-the-loop

✦

Slide 02 // The Problem

Documentation rots in the dark

A function gets renamed. An endpoint is deleted. A parameter changes. Nobody updates the README — because nothing watches it. Every reader after that is misled.

Silent bugs

Devs code against docs that describe a function that no longer exists.

Onboarding drag

New hires lose days trusting stale guides.

Can't use cloud AI

Auditing proprietary code means shipping IP to someone else's model.

Slide 03 // The Solution

An AI that keeps code & docs honest

①Scans locally. An Executa plugin walks the repo, extracts live code symbols.
②AI reconciles. The Anna host LLM classifies every doc mention as fresh, stale, or deleted.
③You approve. Drift lands in a review queue — you accept, edit, or skip every fix.

Snippets are AES-GCM-256 encrypted locally — raw IP never leaves the sandbox in plaintext.

Slide 04 // Core Product Flow

Scan → Reconcile → Review → Ship

project.scan

symbols + SHA-256

➜

docs.crossref

LLM: fresh / stale / deleted

➜

review queue

drift_viewer · human approval

➜

docs.patchgen

unified patch → R2 link

Bonus tools: project.semantic_search (embeddings), project.generate_diagram (image gen), project.history (APS KV).

Slide 05 // Technical Architecture

Native Anna app — not a chatbot wrapper

App UI (iframe)

Multi-view: main + drift_viewer
tools.invoke → real Executa calls
chat.append_artifact SVG cards
upload → R2 download links

Executa (Python, stdio JSON-RPC)

Reverse-RPC sampling/createMessage
storage get/set/list/delete (APS KV)
files/* durable APS objects
embeddings · image/generate

anna-app validate --strict passes · boots in anna-app dev · AES-GCM-256 local crypto.

Slide 06 // Live Demo Highlights

What the judges see

1

Point it at a repo

Real directory walk, symbols extracted live.

2

Drift cards appear

LLM-classified; side-by-side code↔doc viewer.

3

Approve → patch

Accepted fixes become a downloadable patch (R2).

4

Audit card in chat

An SVG summary artifact posts back to the Anna timeline.

No Anna account? It also runs standalone in a browser with seeded data.

Slide 07 // Fit with Anna

We use the platform, deeply & for real

Executa stdio tools sampling/createMessage storage get/set/list/delete files/* (APS) upload → R2 embeddings image/generate multi-view windows chat.append_artifact

Plus a sandboxed Developer Console that exercises the Host-API surface live — routing to the real platform inside Anna, returning clearly-labeled mock responses standalone. Transparent by design.

Slide 08 // Who needs this

Every team that ships code & docs

100M+

developers worldwide — all of them maintain drifting docs.

DevEx

A booming category: docs quality is a top onboarding KPI.

Private

Regulated/IP-sensitive teams that can't paste code into cloud chat.

Slide 09 // Competitive Edge

Why DocDrift wins

VS

manual review — automatic and continuous, not a once-a-quarter chore.

VS

cloud AI tools — air-gapped AES-GCM-256; your source never leaves in plaintext.

VS

linters — semantic embeddings catch drift that string/regex match misses.

And it's delightful. A cute mascot + buttery UI means devs actually run it — adoption is the whole game.

Slide 10 // Roadmap

30 / 60 / 90

30 DAYS

Watch mode

Re-scan on save; post drift deltas to chat.

60 DAYS

GitHub PRs

Scan PR diffs, comment inline on drift.

90 DAYS

Durable history

Browse & diff past audits from APS Files.

Slide 11 // Team

Built solo, shipped clean

11 plugin tests (+54 subtests) passing
anna-app validate --strict green
Real reverse-RPC, multi-view, APS persistence
Original animated mascot & brand system

Docs you can trust.

DocDrift makes documentation honest again — privately, delightfully, and natively on Anna.

Useful Working demo Real AI Native to Anna Delightful