Agent graph
Many small agents. One graph. Nothing unverified.
Ask the rail a statutory question. A coordinator plans the run, a retriever pulls the provisions, a grader decides whether they are good enough, and only then does the filing agent draft. A separate quality agent checks every citation back to a retrieved passage before the answer reaches the desk — and every step is timed, counted and costed.
Multi-agent graph
No single prompt does everything. A coordinator routes work to a retriever, a grader, a rewriter, a filing drafter and a quality assurance agent, each with one job and one contract.
Corrective RAG
Retrieved statute is graded before it is used. Weak retrieval rewrites the search; a second weak pass escalates to the external statutory index. The drafter never guesses.
LCEL nodes
Every node is a composed prompt-to-model chain consumed as a stream, so long generations flow instead of stalling, and nodes run asynchronously inside the graph.
Observation layer
Each span carries latency, token counts and cost. Runs export to JSON and CSV, and mirror to LangSmith when a workspace key is present.
Graph intake
StateGraph topology
Corrective RAG: when the grader marks retrieval partial or poor, the coordinator sends the question back through the rewriter, and only escalates to the external statutory index if a rewritten search still falls short. The drafter never sees ungraded context.
Managed deployment and graph testing
The graph is declared, deployed and scaled as managed infrastructure, and it is regression tested like code: run the suite below and every case is executed against the deployed graph, scored on the structured filing packet it returns, and exported for the record.
Managed deployment — LangGraph Platform
graph “statutory” · langgraph.json
Runtime
LangGraph Platform — managed task queue, Postgres checkpointer, horizontal autoscale
Scaling
Queue-backed workers scale on pending-run depth; long statutory drafts run as background runs and stream tokens back to the desk
Persistence
Thread-scoped checkpoints after every node, so a run resumes at the failed span rather than from intake
Observability
LangSmith project tracing on every managed run; the same spans render in the desk trace console
Published assistants on the deployment
Statutory determination
asst_statutory_full
Full run: coordinate, retrieve, grade, draft the filing packet, quality-check every citation.
recursion_limit 24 · streaming on · corrective retries 2
Corrective retrieval probe
asst_retrieval_only
Retriever + grader + rewriter only — used to test the corpus without spending draft tokens.
interrupt_after ['grader'] · streaming on
Filing QA gate
asst_qa_gate
Re-runs the quality agent against an existing draft before a filing leaves the desk.
entry ['qa'] · deterministic temperature 0
The factory at src/lib/agents/platform-graph.server.ts:makeDeployedGraph is the same builder the desk runs in-process, so a managed run and a desk run execute an identical topology.
Graph regression suite — 4 cases against the deployed graph
Not run in this session
How an agent answer reaches a filing
Treasury desk — 3D process simulation
How a treasurer's action becomes a supervised, rate-stamped record.
Live USD/INR
₹86.9565 ▬
Tape status
Syncing
Print sequence
#0
Paper today → rail
24 days → 9 min
Who acts
Treasury
Rule 7 headroom
What happens at this stage
Existing commitments and headroom are read from the ledger, not rebuilt in a spreadsheet.
Elapsed 30s of 130s on the rail · stage cost 30s
Qwen · reasoning
Statutory reasoning model reads the rule and the filing together, and writes the conclusion with the rule quoted.
LangChain · orchestration
Chains the retrieval, tool calls and checks in a fixed order so every filing is examined the same way.
Alibaba Cloud · compute
Runs the heavy document and screening workloads in an India-resident region.
Federated ledger
Each result is hash-chained into a block every regulator node holds a copy of.
Human desk
A named officer signs this off. The machine only prepares the file.