Open infrastructure for Decision Graphs
Reasoning Layer infrastructure for enterprise AI agents needs a shared substrate - an agreed-on shape for what a decision is, how decisions reference each other, and how an agent queries the chain. Naboo is publishing that substrate, open and free, under Apache 2.0.
Two artifacts, one principle. An open schema specification anyone can implement against, and a reference MCP server that plugs straight into Claude Desktop, Cursor, and any MCP-compatible agent runtime. Build your own backend, fork the reference, or run Naboo as the managed implementation - the choice stays yours.
What's shipping
decision-graph-spec
The open schema. Node types, edge types, attributes, state machine, GraphQL SDL, MCP protocol contract, worked examples. Implementations welcome. Apache 2.0.
github.com/nabooai/decision-graph-spec@naboo/mcp-server
A working MCP server that exposes the spec's tools (list_decisions, get_blockers, trace_decision_chain). Plugs into Claude Desktop, Cursor, and any MCP-compatible agent. Apache 2.0.
npx @naboo/mcp-servergithub.com/nabooai/mcp-serverWhy this is open
Naboo's reasoning is simple. The category needs a shared shape before any single vendor can win it - otherwise every agent runtime ends up integrated against a different proprietary surface, and the whole stack rots into glue code.
We have a strong opinion on what that shape should be, because we've been deploying Decision Graphs inside large organizations (Global-E, Melio) for a while. So we're publishing the shape as a spec, and a reference implementation that works today, and inviting the rest of the ecosystem to ship alternatives.
The schema is the contract. The reference is the proof. The hosted Naboo product - the Forward Deployed Agent who sits with your team and encodes your organization's decisions into the graph - is the thing we sell. Those are two cleanly separable layers. We're betting the category is bigger than any one vendor.
How to get involved
Read the spec
Start with spec/0001-introduction.md in the repo.
Run the MCP server
Five-minute install. Comes with an Acme Co sandbox backend you can query immediately.
Comment on RFCs
All schema discussions live in GitHub Discussions. Issues are for bugs only.
Governance
v0.x is BDFL-governed by Naboo while the shape settles. From v1.0 onward, schema changes require an RFC and rough consensus from active implementers. We'll publish the process in CONTRIBUTING.md alongside the spec.
Related reading
Reasoning Layer for Enterprise AI Agents
Definition, architecture, and the two tiers - Topic Graph and Decision Graph.
Read moreDefinitionWhat is a Decision Graph for AI Agents?
Decisions as first-class nodes - owners, triggers, blockers, evidence. The primitive AI agents need to act.
Read moreHow-toHow to Build a Decision Graph
Seven concrete steps from elicitation to a queryable graph. Two to four weeks via Forward Deployed Agent.
Read moreCFO briefHow to Reduce LLM Token Costs
Don't meter the waste, cut the cause. Reasoning Layer vs observability and caching, compared.
Read moreGuideImprove AI Agent Accuracy
Accuracy is upstream of evals. Four causes of enterprise AI inaccuracy and how a Reasoning Layer fixes them.
Read moreArchitectureConnect Enterprise Data Sources
Live joins vs stale copies. Warehouse, ETL, knowledge graphs, and Reasoning Layer compared.
Read moreGuideOvercome GenAI Hallucinations
Hallucinations are a context-handoff problem, not a model problem. Four causes, one upstream fix.
Read moreROIHow Naboo Saves Cost
Five places Naboo cuts cost in enterprise AI deployments. Four-minute explainer video.
Read moreHubCompare Naboo
Every category enterprise AI buyers weigh against the Reasoning Layer - in one place.
Read moreComparisonNaboo vs Helicone
Reasoning Layer cuts the cause; Helicone measures the waste. Composable.
Read moreComparisonNaboo vs Langfuse
Different layers. Langfuse versions + traces; Naboo grounds the agent.
Read moreComparisonNaboo vs LlamaIndex
RAG framework vs Reasoning Layer. When to use each.
Read moreComparisonNaboo vs LangChain
Orchestration vs substrate. Compose them.
Read moreComparisonNaboo vs Cognee
Open-source agent memory vs enterprise Reasoning Layer. Different primitives, different jobs.
Read moreComparisonNaboo vs Hyperspell
Cloud 'Company Brain' API vs enterprise Reasoning Layer with on-prem, RBAC, and FDA.
Read moreComparisonNaboo vs Modern Relay
Git-style graph DB primitive vs a complete Reasoning Layer shipped end-to-end.
Read moreBackgroundWhy retrieval was the wrong foundation
How enterprise AI agents got built on RAG, why it falls short, and what a reasoning layer fixes.
Read moreComparisonNaboo vs RAG
Retrieval vs reasoning - head-to-head benchmarks, architecture, and when to use each.
Read moreComparisonNaboo vs Glean
Enterprise search vs reasoning layer - when each fits.
Read moreConceptAI Search vs Reasoning Layer
Search returns links; the reasoning layer returns the chain. When to use which.
Read moreCategoryAgent Memory vs Reasoning Layer
Memory recalls what the agent saw. A Reasoning Layer returns what the company decided. Different primitives, different jobs.
Read moreCase studyGlobal-E case study
How Global-E (NASDAQ: GLBE) gave AI agents secure access to customer data.
Read moreComparisonCompare alternatives
Naboo vs other enterprise AI agent infrastructure platforms.
Read moreTwo ways to engage Naboo
The schema and the reference MCP server are yours to fork, extend, and ship against. Or talk to us about the hosted Reasoning Layer infrastructure - a Forward Deployed Agent who encodes your organization's Decision Graph alongside your team.