← Blog
ProductJan 28, 2026·6 min read

Introducing SNOW AI v2.1

Our AI engine gets smarter — Pulse briefings, staleness detection, automatic thread linking, and contributor analysis. SNOW now understands not just what your team knows, but how that knowledge evolves over time.

S
SLEDS Team

SNOW — the Semantic Network of Observations and Work — is the intelligence layer that powers everything in SLEDS. Today we're shipping v2.1, and it's the biggest upgrade since launch.

Pulse Briefings

Every sled now has a Pulse: a narrative briefing generated by Frost that synthesizes everything happening across your threads, assets, and contributors. It's not a list of updates. It's a story — what's progressing, what's stalled, what questions are open, and what needs attention.

Pulse briefings are designed to replace the "catch me up" conversation that happens every Monday morning. Instead of spending thirty minutes getting context from teammates, you read a two-minute briefing that captures the state of the project with full provenance.

Staleness Detection

SNOW now tracks how long since each thread and asset was last touched. When something goes stale — an open question nobody answered, a spec that hasn't been updated since a major decision changed — Frost flags it. This is particularly valuable for larger teams where things can slip through the cracks without anyone noticing.

Automatic Thread Linking

When SNOW detects semantic overlap between threads, it creates links automatically. If your Claude conversation about database schema mentions the same concepts as a Cursor session about migrations, SNOW connects them. This builds a knowledge graph organically — no manual tagging required.

Contributor Analysis

Pulse briefings now include contributor context: who's been active, what tools they're using, and what areas they're focused on. This helps team leads understand workload distribution and ensures that knowledge isn't siloed in one person's conversations.

Under the Hood

V2.1 includes significant infrastructure improvements: pgvector with HNSW indexing for faster semantic search, BullMQ workers for background processing, and a new embeddings pipeline that processes observations in near real-time. The Pulse generation pipeline can now synthesize context from hundreds of thread observations in under five seconds.

SNOW v2.1 is live now for all sleds. Pulse briefings generate automatically and refresh when significant new context arrives.

← Previous
Frost Dispatch: AI That Works While You Sleep
Next →
Shared Memory Gives Everyone Equal Footing

Ready to try shared AI memory?

Free to start. Connect your AI tools in minutes.

Get Started Free →
sleds — Shared AI memory for teams