Beta Program — Open

Beta Tester Signup & Feedback

Help shape Tactical AI DuckDB for ATAK-CIV 5.8. Test the offline DuckDB spatial engine, on-device AI, live radar tracks, SAR planning, and ATC-45 damage assessment on real hardware — then tell us what works in the field and what to fix before release.

Beta Testers Wanted — Help test the new ATAK plugin Tactical AI DuckDB. Know TAK & geospatial data? Join the beta.

What Beta Testers Evaluate

The beta focuses on the verified v1.3.0 core loop and the feature domains evaluators ask about first. You'll exercise real workflows with the network off and report back.

🌎

Spatial Engine

DuckDB spatial SQL plus the in-app OGC API Features/Tiles server with GL-styled raster rendering.

🤖

On-Device AI

Gemma 3N text analysis, document RAG, AI Agent chat, and MobileSAM segmentation — no cloud tether.

🚢

Live Radar Tracks

ASTERIX CAT-048/021/062 ingest, CPA analytics, emergency-squawk alerting, and DuckDB history.

📅 How the Beta Works

Sign up — complete the form below so we know your device, ATAK version, and focus areas.
Get the build — we'll send the beta APK plus setup instructions and the ATAK-CIV SDK prerequisite.
Run real workflows — try the Mission Demo loop, SAR planning, damage assessment, and radar ingest offline.
Send feedback — use the same form's feedback section to report bugs, friction, and feature gaps.
Shape the release — your input directly prioritizes what ships in the next verified version.

⚠️ Prerequisites

Beta testing requires an ARM64 Android 9.0+ (API 28) device with ATAK-CIV 5.8 installed and 4GB+ RAM (6GB+ recommended for AI models). You'll need to sideload the beta APK and stage AI models on device. The plugin runs fully offline once staged.

Sign Up & Share Feedback

Complete the form to join the beta and send us your feedback. Loading may take a few seconds — if it doesn't appear, open it directly in a new tab.

Form not loading? Open the signup & feedback form in a new tab →

What We're Listening For

The most useful feedback tells us where the plugin helps — and where it gets in the way. Specifically:

Field usability — gloved-hand operation, sunlight readability, and one-handed reach on a tactical device
Offline reliability — does the core loop hold up with the network fully off for an extended mission?
Performance — DuckDB query latency, AI inference time, and radar ingest load on your hardware
Workflow fit — SAR, ATC-45 damage assessment, radar air picture, and AI agent tool calls in real use
Bugs & crashes — reproduction steps, device model, ATAK version, and logs when you have them

Honesty-Gated by Design

If a feature's runtime dependency is missing during your test — a model not staged, a decoder not wired — the plugin reports that the capability is unavailable rather than fabricating output. That fail-closed behavior is intentional and one of the things we most want feedback on: did the plugin tell you clearly enough what was wrong?

← Back to Features Request a Demo →