Overview — Project 3: Signal Room

AskDocs handled one modality: text. Signal Room reconstructs a software incident from logs, screenshots, distributed traces, a deployment diff, and an on-call voice note. Its result is not one confident paragraph. It is a timeline of claims where every claim points to evidence and uncertainty remains visible.

What you will produce

signal-room/run-01/
├── evidence-manifest.json
├── provenance.json
├── aligned-timeline.json
├── hypotheses.json
├── incident-report.md
└── evaluation.json

The four days

Day	Build	Rust pressure	AI pressure
10	four-modality intake	enums, validated constructors, hashes	preserve modality and provenance
11	screenshot and voice-note evidence	bytes, structs, fallible transforms	OCR/transcription are observations
12	one incident clock	integer time, ordering, reversible maps	cross-modal correlation
13	cited hypothesis graph	graphs, state, serialization	uncertainty and evidence grading

Run the pieces that become Signal Room:

cd rust/rust-ai-engineering
cargo run -q -p mosaic-evidence --example evidence_envelopes
cargo run -q -p mosaic-evidence --example image_pipeline
cargo run -q -p mosaic-evidence --example audio_pipeline
cargo run -q -p mosaic-evidence --example alignment_pipeline

The product grows inside reusable mosaic-evidence types because Cutroom and Mosaic need the same acquisition, provenance, and alignment rules.

Start with Day 10 — Evidence Envelopes →.