From zero to savings in four steps
Artifact access
Under a definitive agreement or data-room grant, the buyer receives the surfclaw-core tree (and related materials). Dependencies install from the supplied requirements into your environment — there is no advertised public PyPI package for turnkey acquisition of the full asset.
Graph is Built
Nodes, edges, Leiden clusters, confidence scores. Your selected code/document corpus becomes a queryable graph — 8x–35x fewer tokens per query (independently benchmarked, conservative estimate).
Query Anything
Ask natural language questions. SurfClaw traverses the graph and returns only the relevant context for your existing LLM call — no raw file dump.
Embed In-House
The acquiring enterprise runs the middleware inside its own private cloud or closed-network gateway. All traffic and LLM spend remain buyer-owned.
Enterprise-grade core. Buyer-owned deployment.
Large code and docs. One graph.
SurfClaw converts selected codebases and document corpora into a traversable knowledge graph, then extracts only the context relevant to each LLM query.
Serverless. Zero fixed cost.
Runs on Modal — you pay only per invocation, never for idle time. Your infrastructure bill is literally $0 when no one is calling the API.
Incremental. Never rebuilds.
Change one file, patch one subgraph. SurfClaw tracks SHA256 content hashes so unchanged nodes are never re-extracted. Enterprise corpora update in under 1 second.
All LLMs. One interface.
OpenAI, Anthropic, Google, Meta — every model family is supported. Savings are calculated at the exact per-token price of whichever model your team uses.
Your LLM token spend is too high.
We fixed it.
SurfClaw Core is prepared for in-house embedding as a buyer-owned middleware asset.