Overview
Roadmap & Changelog
Up Next
- ๐งท Count and scan primitives โ filter-count mode, snapshot truncation removal, route renames (RFC 0019, #67)
- ๐ Indexing failure-mode E2E runbook โ Aerospike stop-writes + Postgres pressure (#55)
- ๐งฌ Embedding UDF writeback via row re-upsert (#52)
- ๐ฑ Namespace init UDF for first-time embed population
- ๐ฆ Snapshot-aware ready signal โ
layer.is_stablehonors UDF state (#46) - ๐จ Full dashboard redesign โ 6-tab layout from the prototype (#11)
Later
- ๐ RBAC: scoped API keys and entitlements as a Layer primitive (#8)
- โป๏ธ Soft delete with TTL + restore (#7)
- ๐ชข Hybrid text search via RRF (#18)
- โจ๏ธ Typeahead via Turbopuffer regex index (#19)
- ๐ฐ๏ธ Temporal queries โ
as_ofselector for/query,/scans,/fetch, and/snapshots(RFC 0020, #68) - ๐ฟ
copy_from_with_filterโ time travel + subset branching (#20) - ๐ Exact kNN result cache keyed by consistency watermark (#21)
- ๐งช A/B variant indexes with operator-controlled rollout (#23)
- ๐ฆ Per-query observability with LLM-judged Tail Quality (#41)
- ๐๏ธ Pipeline crash recovery via source replay + deterministic IDs (#43)
- โ ๏ธ Paginated UDF dead-letter list (#44)
- ๐๏ธ Narrow cluster topology defaults (#45)
- ๐ฃ Write amplification baselines (#15)
- ๐ฎ
layer pushโ Python UDF dev experience via Depot (#64)
0.1 Release (UAT)
API hardening
- ๐งฉ Scaling CRD consolidation โ
Pipeline,UDF,InfraRules(RFC 0012) - ๐๏ธ
IndexCRD redesign (RFC 0013) - ๐ธ Snapshot scan naming conventions (RFC 0014)
- ๐งน Remove unused APIs
Lifecycle and operability
- ๐๏ธ Autoscaling compute for pipelines and UDFs
- ๐๏ธ Document cache endpoint for building multi-stage pipelines
- ๐ธ Index snapshot history
- ๐งจ Coordinated delete
- โต Helm and Terraform install scripts
Surfaces
- ๐ช Dashboard MVP โ basic CRD management and observability
- ๐
layerCLI MVP โ kube-style commands for Layer - ๐ Official Python SDK
Search
- ๐ฏ Strongly consistent queries during heavy writes via
_hevlayer_upserted_at - ๐งฎ Result count over FTS/vector queries via scatter/gather
- ๐ Precomputed facet listings in snapshots
- ๐ช Precomputed facet counts in snapshots
- ๐ช Scans for filter IDs and filter counts not available in a snapshot
- ๐ Search by id via document-cached vector
- ๐ฐ Search history saved to S3
- ๐๏ธ Enhanced namespace metadata