GraphSentinel：Agent 驱动的欺诈调查系统

GraphSentinel: Building an Agentic Fraud Investigation System on TigerGraph Fraud investigation is not only a classification problem. An analyst needs to understand why a transaction is risky, gather the right evidence, follow policy, choose an action, document approvals, and preserve the investigation for the next case. GraphSentinel was built to explore that complete workflow. It is an agentic fraud-investigation and next-best-action system for the TigerGraph Agentic Fraud Investigation challenge. The system starts from a risk alert, customer report, or analyst request and produces a traceable investigation record containing graph evidence, model belief, policy citations, evidence requests, actions, approvals, SAR decisions, and case memory. The central idea is: Use the graph to gather structured evidence, use a model to estimate risk, use policy to constrain actions, and use an agent to decide what investigation should happen next. What we built GraphSentinel combines five main ideas: A temporal graph investigation layer for transactions, customers, devices, cards, addresses, prior cases, and relationships. A risk model trained on closed investigations rather than relying only on a hard-coded fraud score. GraphRAG retrieval for policies, fraud patterns, and regulatory references. A policy engine that determines what is permitted, what requires approval, and when customer contact is restricted. A bounded agent workflow that can select follow-up tools, request evidence, update its belief, and produce an auditable explanation. The interesting part is that the workflow records a next-best action before requesting additional evidence. If the case is still uncertain, the system chooses an allowed evidence request using value-of-information scoring. After the response, it updates its belief and records another next-best action. This makes the effect of evidence visible instead of hiding everything inside a final classification. The application supports: Local graph store │ ├── Offline development │ ├── TigerGraph REST │ └── TigerGraph MCP Enter fullscreen mode Exit fullscreen mode The local graph implementation follows the same query contract as the TigerGraph backends, allowing the investigation workflow to be tested without requiring a live TigerGraph instance. Architecture At a high level, an investigation follows this path: Trigger │ ▼ Intake │ ▼ Baseline Graph Evidence │ ▼ Agent-selected Follow-up Queries │ ├───────────────┐ ▼ ▼ Similar Cases GraphRAG │ │ └───────┬───────┘ ▼ Risk Model + Fraud Classification │ ▼ Policy Decision │ ▼ NBA Before Evidence │ ▼ Is the case uncertain? / \ No Yes │ │ │ ▼ │ Select Evidence │ │ │ ▼ │ Apply Response │ │ │ ▼ │ Update Belief │ │ │ ▼ │ NBA After Evidence │ │ └──────┬───────┘ ▼ Actions / Approvals / SAR │ ▼ Explanation │ ▼ Graph Write-back Enter fullscreen mode Exit fullscreen mode The main runtime is assembled by services/runtime.py. It loads the dataset, graph store, policy configuration, pattern library, risk model, likelihood tables, case repository, evidence provider, and optional CrewAI client. The agent/workflow.py module builds the LangGraph state machine with explicit nodes for: intake baseline evidence follow-ups memory / RAG assessment decision evidence finalization Enter fullscreen mode Exit fullscreen mode This explicit state-machine approach makes the investigation path easier to test and reason about than putting the entire workflow inside a single agent prompt. Separation of powers One of the most important architectural decisions was deliberately separating responsibilities. ┌───────────────────────────┐ │ TigerGraph │ │ │ │ Evidence + Relationships │ └─────────────┬─────────────┘ │ ▼ ┌───────────────────────────┐ │ Risk Model │ │ │ │ Fraud probability │ └─────────────┬─────────────┘ │ ▼ ┌───────────────────────────┐ │ Policy Engine │ │ │ │ Permissions + Approvals │ └─────────────┬─────────────┘ │ ▼ ┌───────────────────────────┐ │ Agent / LLM │ │ │ │ Follow-up + Explanation │ └─────────────┬─────────────┘ │ ▼ ┌───────────────────────────┐ │ Action Gateway │ │ │ │ Approved actions only │ └───────────────────────────┘ Enter fullscreen mode Exit fullscreen mode The graph supplies evidence. The risk model estimates fraud probability. The policy engine determines permissions and approval routes. The LLM proposes optional follow-up work and generates language. The action gateway executes only policy-approved actions. The LLM is never allowed to authorize or execute a protective action. Its JSON output is validated, unknown tool names are discarded, citations are filtered against retrieved policy clauses, and failures fall back to deterministic templates. The investigation workflow 1. Baseline evidence Every investigation begins with a focal transaction. The system gathers a bounded set of temporal graph queries: Transaction details + owner │ ├── Customer history ├── Card activity ├── Shared devices ├── Address peers ├── Customer case history └── Linked cases Enter fullscreen mode Exit fullscreen mode Representative queries include: gs_txn_detail gs_customer_history gs_device_customers gs_address_peers gs_linked_cases Enter fullscreen mode Exit fullscreen mode These queries are transformed into signals such as: signals = { "amount_ratio": amount_ratio, "device_novelty": device_novelty, "account_age": account_age, "card_velocity": card_velocity, "email_mismatch": email_mismatch, "linked_confirmed_cases": linked_confirmed_cases, "graph_fraud_proximity": fraud_proximity, "address_cluster_size": address_cluster_size, } Enter fullscreen mode Exit fullscreen mode A critical constraint is that behavioral queries are evaluated strictly before the focal event. If a transaction occurred on January 10, information that only became available on January 20 cannot influence the January 10 decision. A simplified interface therefore looks like: def get_customer_history( customer_id: str, as_of: datetime, ): return graph.run_query( "gs_customer_history", { "customer_id": customer_id, "as_of": as_of.isoformat(), }, ) Enter fullscreen mode Exit fullscreen mode This as_of boundary is part of the graph-access layer rather than an assumption made by the analyst. 2. Follow-up planning Baseline evidence is not always enough. The agent can select additional graph investigations, for example: Device ring expansion Card activity Address-cluster transactions Community statistics Spending trajectory Enter fullscreen mode Exit fullscreen mode The agent can select up to three additional tools. CrewAI acts as a bounded investigation assistant for this stage. It receives the available signals and a menu of installed tools and returns structured output such as: { "tools": [ { "name": "device_ring", "reason": "Device is shared with multiple high-risk customers." }, { "name": "address_cluster", "reason": "Several recently created accounts share this address." } ] } Enter fullscreen mode Exit fullscreen mode The important part is that the model doesn't receive arbitrary database access. The returned tool names are validated against the installed tool registry: TOOLS = { "device_ring": investigate_device_ring, "card_activity": investigate_card_activity, "address_cluster": investigate_address_cluster, "community_stats": investigate_community, "spending_trajectory": investigate_spending, } def validate_tools(requested): return [ tool for tool in requested if tool["name"] in TOOLS ] Enter fullscreen mode Exit fullscreen mode Unknown tools are discarded. If the model produces malformed output or fails completely, the workflow falls back to deterministic planning. This creates a controlled boundary: LLM │ │ proposes ▼ Tool Registry │ │ validates ▼ Installed Graph Queries Enter fullscreen mode Exit fullscreen mode rather than: LLM ─────────────► arbitrary database access Enter fullscreen mode Exit fullscreen mode 3. Memory and GraphRAG Fraud investigation requires more than transaction data. The agent may need to understand: previous investigations, internal policies, known fraud patterns, regulatory requirements. GraphSentinel therefore uses GraphRAG. The case is embedded and compared with previous cases using both vector similarity and structural relationships such as shared devices and cards. Policy documents are parsed into a document graph containing: DocumentChunk │ ├── PolicyClause ├── Pattern └── Regulation Enter fullscreen mode Exit fullscreen mode Retrieval combines vector search with graph expansion: User / Case Question │ ▼ Vector Search │ ▼ Relevant Document Chunks │ ▼ Graph Expansion │ ▼ Policy / Pattern / Regulation Context │ ▼ Cited Investigation Context Enter fullscreen mode Exit fullscreen mode Representative queries include: gs_similar_cases_vec gs_doc_search_vec gs_policy_context Enter fullscreen mode Exit fullscreen mode This allows the system to attach policy citations to evidence requests, actions, explanations, and SAR decisions. 4. Similar-case retrieval Vector similarity alone is not enough for fraud investigations. Two cases may have similar descriptions but completely different graph structures. Therefore case retrieval combines: Vector similarity + Shared devices + Shared cards + Structural relationships + Previous case outcomes Enter fullscreen mode Exit fullscreen mode Conceptually: similar_cases = retrieve_similar_cases( embedding=current_case_embedding, graph_links=[ customer_id, *device_ids, *card_ids, ], ) Enter fullscreen mode Exit fullscreen mode This allows the risk model to use historical cases without reducing the investigation to a simple nearest-neighbor search. 5. Assessment After graph evidence, follow-up investigation, and memory retrieval, the system estimates fraud probability. The risk model is a regularized logistic model. Conceptually: features = [ bank_risk_score, amount_ratio, device_novelty, account_age, card_velocity, graph_fraud_proximity, confirmed_case_links, similar_case_outcomes, pattern_strength, trigger_type, ] fraud_probability = model.predict_proba( [features] )[0, 1] Enter fullscreen mode Exit fullscreen mode The resulting belief is separated into fraud hypotheses: Legitimate Third-party fraud First-party fraud Enter fullscreen mode Exit fullscreen mode Evidence can later update those hypotheses. For example, the supplied likelihood tables can make a failed step-up authentication increase the probability of third-party fraud, while a passed authentication shifts belief in the other direction. 6. Policy decision and next-best action This is one of the most important parts of the architecture. The policy engine applies configurable thresholds. Conceptually: if probability ACTION_THRESHOLD: decision = "action" else: decision = "uncertain" Enter fullscreen mode Exit fullscreen mode For uncertain cases, the system evaluates potential evidence requests. But policy is applied before optimization. Suppose we have: Customer validation Step-up authentication Analyst review Device investigation Enter fullscreen mode Exit fullscreen mode A naive implementation might calculate information gain first. GraphSentinel instead does: Candidate Evidence │ ▼ Policy Filter │ ▼ Allowed Evidence │ ▼ Value of Information │ ▼ Selected Evidence Enter fullscreen mode Exit fullscreen mode Conceptually: allowed = [] for request in evidence_requests: policy_result = policy.check( case=case, request=request, ) if policy_result.allowed: allowed.append(request) best_request = max( allowed, key=lambda request: expected_information_gain(request) - request_cost(request) ) Enter fullscreen mode Exit fullscreen mode This matters because a request can be statistically useful while still being impermissible. For example: customer validation can be blocked for customer-reported cases; contact can be restricted when first-party fraud is likely; requests that could create SAR tipping-off risk can be excluded. The optimizer never sees those prohibited requests. 7. NBA before evidence Before any additional evidence is requested, the system records the current next-best action. For example: nba_before = { "decision": "escalate", "probability": 0.71, "risk_level": "high", "fraud_class": "third_party", "action": "hold_transaction", "permission": "approval", "approval_route": "fraud_analyst", "policy_clause": "POL-2.1", } Enter fullscreen mode Exit fullscreen mode This creates an important property: We know what the system would have done before seeing the additional evidence. Each NBA records information such as: decision probability risk level fraud class selected evidence request excluded requests policy reasons actions permission type approval route policy clause received evidence Enter fullscreen mode Exit fullscreen mode 8. Evidence request and belief update If the case remains uncertain, the selected evidence request is executed. For example: evidence = { "type": "step_up_auth", "result": "failed", } Enter fullscreen mode Exit fullscreen mode The result updates the fraud hypotheses. Conceptually: posterior = bayesian_update( prior=belief, evidence=evidence, likelihoods=likelihood_tables, ) Enter fullscreen mode Exit fullscreen mode The important architectural property is that evidence is represented as an explicit transition: Belief Before │ ▼ NBA Before │ ▼ Evidence Request │ ▼ Evidence Response │ ▼ Belief Update │ ▼ NBA After Enter fullscreen mode Exit fullscreen mode This makes it possible to inspect exactly how new evidence changed the investigation. 9. NBA after evidence After updating the belief, the system records a second next-best action. nba_after = { "decision": "protect", "probability": posterior.fraud_probability, "fraud_class": posterior.fraud_class, "action": "block_transaction", "permission": "approval", "approval_route": "fraud_analyst", "policy_clause": "POL-2.1", } Enter fullscreen mode Exit fullscreen mode The case now contains a complete decision timeline: Initial Evidence │ ▼ Initial Belief │ ▼ NBA Before Evidence │ ▼ Evidence Request │ ▼ Evidence Response │ ▼ Updated Belief │ ▼ NBA After Evidence Enter fullscreen mode Exit fullscreen mode This is more informative than simply returning: { "fraud": true } Enter fullscreen mode Exit fullscreen mode 10. LangGraph workflow The complete investigation is represented as a state machine. A simplified version looks like: from langgraph.graph import StateGraph, END workflow = StateGraph(InvestigationState) workflow.add_node("intake", intake) workflow.add_node("baseline", baseline_evidence) workflow.add_node("followups", followup_planning) workflow.add_node("memory", retrieve_memory) workflow.add_node("assessment", assess_risk) workflow.add_node("decision", policy_decision) workflow.add_node("evidence", request_evidence) workflow.add_node("finalize", finalize_case) workflow.set_entry_point("intake") workflow.add_edge("intake", "baseline") workflow.add_edge("baseline", "followups") workflow.add_edge("followups", "memory") workflow.add_edge("memory", "assessment") workflow.add_edge("assessment", "decision") workflow.add_conditional_edges( "decision", route_after_decision, { "evidence": "evidence", "finalize": "finalize", }, ) workflow.add_edge("evidence", "assessment") workflow.add_edge("finalize", END) app = workflow.compile() Enter fullscreen mode Exit fullscreen mode The important property is that an evidence response can send the case back through assessment. The system is therefore: Trigger ↓ Investigate ↓ Assess ↓ Decide ↓ Need evidence? ├── No ───────► Finalize │ └── Yes ↓ Evidence ↓ Reassess ↓ Decide Enter fullscreen mode Exit fullscreen mode This is the core agentic loop. How TigerGraph is used TigerGraph is the graph system of record for investigation evidence and case memory. Graph model The graph contains vertices for: Transaction Customer Device Card Address FraudCase Finding Action DocumentChunk PolicyClause Pattern Regulation Enter fullscreen mode Exit fullscreen mode Edges represent relationships such as: Customer ──owns──────► Card Customer ──uses──────► Device Customer ──lives_at──► Address Transaction ──belongs_to──► Customer Case ──has_finding──► Finding Case ──has_action────► Action Case ──similar_to────► Case Document ──references─► PolicyClause Enter fullscreen mode Exit fullscreen mode This turns the fraud investigation into a connected evidence problem rather than a flat feature table. Installed GSQL query contract Every graph read is a named installed query. The contract in graph/contract.py defines query names, parameters, and result parsing. Conceptually: QUERY_CONTRACT = { "gs_txn_detail": [ "txn_id", ], "gs_customer_history": [ "cust", "as_of", "max_rows", ], "gs_address_peers": [ "cust", "window_sec", "as_of", "min_first_seen", ], "gs_similar_cases_vec": [ "query_vec", "k", ], } Enter fullscreen mode Exit fullscreen mode The local graph store, TigerGraph REST store, and MCP store all implement the same interface. This gives us two major advantages: Investigation code isn't coupled to one TigerGraph client mechanism. Local, REST, and MCP implementations can be tested against the same query names and response shapes. Representative installed queries include: gs_txn_detail gs_customer_history gs_device_customers gs_address_peers gs_linked_cases gs_similar_cases_vec gs_doc_search_vec gs_policy_context Enter fullscreen mode Exit fullscreen mode Graph algorithms The repository also includes: Weakly Connected Components Louvain Communities Personalized PageRank Enter fullscreen mode Exit fullscreen mode Personalized PageRank is seeded from confirmed-fraud customers to create a graph-based fraud-proximity feature. The setup also computes: device degrees customer links connected components communities fraud proximity Enter fullscreen mode Exit fullscreen mode Highly connected hub devices are excluded from useful proximity signals because shared corporate or public devices can otherwise create misleading fraud relationships. TigerGraph MCP The agent-plane MCP adapter exposes four operations: run_installed_query add_nodes add_edges get_node Enter fullscreen mode Exit fullscreen mode The architecture becomes: LangGraph Agent │ ▼ TigerGraph MCP │ ├── run_installed_query ├── get_node ├── add_nodes └── add_edges │ ▼ TigerGraph Enter fullscreen mode Exit fullscreen mode The MCP server is launched over stdio using the same TigerGraph configuration. The repository also contains an MCP emulator backed by the local graph store. This made it possible to test the complete MCP path without requiring a live TigerGraph deployment. Precomputation Before investigations run, graph-derived features are precomputed. Conceptually: graph.compute_device_degrees() graph.compute_customer_links() graph.compute_wcc() graph.compute_louvain() graph.compute_fraud_pagerank() Enter fullscreen mode Exit fullscreen mode These values can then be used during investigation instead of repeatedly traversing the entire graph. Temporal correctness One of the less obvious challenges was preventing future information from leaking into the investigation. Consider: January 10 │ └── suspicious transaction January 15 │ └── investigation starts January 20 │ └── case confirmed as fraud Enter fullscreen mode Exit fullscreen mode The January 20 outcome must not become a feature for the January 10 transaction. Therefore graph queries use: as_of = case_opened_at Enter fullscreen mode Exit fullscreen mode and only retrieve information available before the relevant event. The same principle applies to customer history, account age, devices, cards, and linked cases. Left-censored accounts also need special handling. If the available dataset starts after the account was created, we shouldn't automatically classify that account as "new." These rules belong in the graph access layer and tests, not only in analyst convention. Case memory The investigation does not disappear after the final API response. The case is written back to graph memory as a FraudCase. Conceptually: graph.add_node( "FraudCase", { "id": case.id, "status": case.status, "embedding": case.embedding, }, ) Enter fullscreen mode Exit fullscreen mode Findings and actions are then connected: graph.add_node( "Finding", { "id": finding.id, "type": finding.type, "confidence": finding.confidence, }, ) graph.add_edge( "HAS_FINDING", case.id, finding.id, ) Enter fullscreen mode Exit fullscreen mode The case can be connected to: Customer Focal Transaction Related Transactions Findings Actions Similar Cases Enter fullscreen mode Exit fullscreen mode This creates a continuous memory loop: Past Investigations │ ▼ Case Memory │ ▼ New Investigation │ ▼ New Findings │ ▼ Updated Memory Enter fullscreen mode Exit fullscreen mode But agent outcomes and analyst-confirmed outcomes are deliberately kept separate. An agent prediction should not automatically become training ground truth. Only analyst-confirmed outcomes should become authoritative learning data. SAR handling SAR eligibility is evaluated by policy. The system considers factors such as: Posterior probability Aggregate amount Suspect identification Money-laundering indicators Applicable thresholds Enter fullscreen mode Exit fullscreen mode When a SAR is required, the agent can draft the narrative: if policy.requires_sar(case): sar = draft_sar( case=case, evidence=evidence, citations=policy_context, ) approval_queue.submit( sar, role="bsa_officer", ) Enter fullscreen mode Exit fullscreen mode The LLM can help write the narrative, but it does not independently authorize the SAR. The policy engine controls eligibility and the required approval route. Agentic capabilities The system is agentic in a constrained, auditable sense. It can: choose which optional graph tools to run; decide that additional evidence is needed; select evidence using expected information gain and cost; update beliefs after evidence responses; produce separate pre-evidence and post-evidence decisions; request human approval through policy-defined routes; write structured explanations; preserve the complete investigation timeline; discover residual patterns in closed cases and flag undocumented patterns for policy review. The important design choice is that agency is bounded by contracts and policy. The agent can explore and explain. It cannot silently bypass an approval route or turn an uncertain case into an automatic protective action. Undocumented pattern discovery GraphSentinel also contains a discovery loop for closed investigations. The idea is: Closed Cases │ ▼ Residual Analysis │ ▼ Unexpected Pattern │ ▼ Candidate Rule │ ▼ Policy Review Enter fullscreen mode Exit fullscreen mode For example, the discovery system might identify an unexplained cluster of newly created accounts sharing the same address. The important part is that discovery does not automatically become policy. The candidate is flagged for review: candidate_pattern = { "pattern": pattern, "documented": False, "requires_policy_review": True, } Enter fullscreen mode Exit fullscreen mode This keeps pattern discovery separate from authorization. Finalization Once the investigation is complete: Automatic actions │ ▼ Mock Action Gateway Approval actions │ ▼ Required Approval Route SAR required │ ▼ BSA Officer Approval All paths │ ▼ Explanation │ ▼ Graph Memory Enter fullscreen mode Exit fullscreen mode Automatic actions are sent to the mock gateway. Approval actions are queued for the required role. SAR eligibility is evaluated using the policy rules. Finally, the complete case is written back to graph memory. Running the system locally The project can run without TigerGraph for the initial development loop. pip install -e ".[dev]" graphsentinel synth graphsentinel build graphsentinel run-benchmark graphsentinel eval graphsentinel serve pytest -q Enter fullscreen mode Exit fullscreen mode The local graph store follows the same query contract as the TigerGraph implementation. For TigerGraph: cp .env.example .env # Configure: # GS_MODE=tigergraph # TG_* variables graphsentinel tg-setup graphsentinel tg-check graphsentinel serve Enter fullscreen mode Exit fullscreen mode The default agent access path is MCP: GS_TG_ACCESS=mcp Enter fullscreen mode Exit fullscreen mode REST access is also supported: GS_TG_ACCESS=rest Enter fullscreen mode Exit fullscreen mode Administrative operations such as DDL, bulk loading, and algorithm setup use the REST path. Using the real HHGOA dataset The application also supports the actual HHGOA/IEEE-style dataset through configurable column mappings. The dataset directory is configured with: GS_DATA_DIR=/path/to/dataset Enter fullscreen mode Exit fullscreen mode The loader resolves file and column names using: config/dataset_mapping.yaml Enter fullscreen mode Exit fullscreen mode For the provided benchmark case pack: case_pack.csv Enter fullscreen mode Exit fullscreen mode is placed alongside the dataset files. The benchmark can then be run with: GS_DATA_DIR=/path/to/case-pack \ graphsentinel run-benchmark Enter fullscreen mode Exit fullscreen mode The generated cases are written as: cases/ ├── HHG-001.json ├── HHG-002.json ├── ... └── HHG-020.json Enter fullscreen mode Exit fullscreen mode Each investigation record contains the investigation evidence, findings, decisions, actions, graph write-back status, SAR details where applicable, and the next-best action before and after evidence. Evaluation The project contains two different evaluation modes. The first is the generated benchmark. The second is temporal replay. This distinction is important because the benchmark uses synthetic data where the generator deliberately plants patterns. The benchmark therefore demonstrates that the system behaves correctly against the generated ground truth. It should not be interpreted as a production fraud-detection accuracy estimate. The temporal replay is a more realistic test because the model trains on earlier cases and investigates later cases. The repository reports: Training: Months 1–3 69 closed cases Replay: 37 later cases Enter fullscreen mode Exit fullscreen mode The replay produced: 21 cases decided directly 20 correct decisions 16 escalations Precision: 1.0 Enter fullscreen mode Exit fullscreen mode The important limitation is that many escalations came from evidence requests for which no response was recorded in the closed-case data. This means the system sometimes correctly identifies uncertainty but does not have enough historical evidence to resolve it automatically. That is an important difference between: "I don't know" Enter fullscreen mode Exit fullscreen mode and: "I am confident this is legitimate." Enter fullscreen mode Exit fullscreen mode A production system should preserve that distinction. What we learned Graphs are most valuable when they change an action A large graph is not automatically useful. Device, card, and address relationships matter when they: change the posterior; explain a fraud class; justify an action; identify useful evidence. The query contract helped keep graph investigation focused on decisions rather than graph traversal for its own sake. Time correctness is harder than it looks Fraud data contains future outcomes, later cases, and accounts that may predate the available dataset. Every query therefore needs an as_of boundary. Left-censored accounts must also be handled correctly. Otherwise a seemingly good model can quietly learn from information that would not have been available at decision time. Evidence selection needs policy before optimization A request can be statistically informative and still be impermissible. Therefore: Evidence candidates ↓ Policy constraints ↓ Allowed candidates ↓ Value-of-information ↓ Selected evidence Enter fullscreen mode Exit fullscreen mode This ordering prevents an optimizer from selecting a request that creates customer-contact or tipping-off risk. Human outcomes must remain authoritative Agent-confirmed and agent-cleared cases are useful memory. But they are not automatically analyst ground truth. Keeping those outcomes separate prevents feedback loops where the model starts training on its own previous decisions. Deterministic fallbacks are essential LLM calls can fail because of: Credentials Rate limits Provider changes Malformed JSON Network failures Enter fullscreen mode Exit fullscreen mode The investigation should still continue. That's why GraphSentinel has deterministic planning and explanation fallbacks. The LLM is an optional reasoning and language layer, not a single point of failure. What we would improve with more time 1. Live TigerGraph validation Execute every GSQL query against a real TigerGraph Savanna or Community Edition deployment and add stronger deployment/version checks. 2. Real integrations Replace the mock action gateway with authenticated banking, notification, evidence-provider, and e-filing integrations. 3. Production identity and authorization Replace the development X-Role header with an identity-provider integration and enforce role claims at a trusted proxy boundary. 4. Model calibration Train and calibrate thresholds on real closed investigations, monitor drift, and add confidence intervals and champion/challenger evaluation. 5. Richer evidence providers Connect real authentication, customer-validation, and analyst-review systems with asynchronous response handling. 6. Evaluation against the actual case pack Run the complete HHG-001 through HHG-020 case pack and compare decisions with independent review. 7. Frontend modernization The current console is a lightweight static analyst UI. A production version would use richer graph interactions, accessibility improvements, and durable event streaming. 8. Operational observability Add structured traces for: Graph latency Model versions LLM calls Policy decisions Approval turnaround Action outcomes Enter fullscreen mode Exit fullscreen mode These would be essential for production monitoring. Final architecture The entire system can ultimately be reduced to this: ┌───────────────────┐ │ TRIGGER │ │ │ │ Risk alert │ │ Customer report │ │ Analyst request │ └─────────┬─────────┘ │ ▼ ┌───────────────────┐ │ LANGGRAPH │ │ AGENT │ └─────────┬─────────┘ │ ┌─────────────┼─────────────┐ │ │ │ ▼ ▼ ▼ ┌──────────┐ ┌──────────┐ ┌───────────┐ │TigerGraph│ │ GraphRAG │ │ Case │ │ │ │ │ │ Memory │ │ Evidence │ │ Policies │ │ Similar │ │ Relations│ │ Patterns │ │ Cases │ └────┬─────┘ └────┬─────┘ └─────┬─────┘ │ │ │ └─────────────┼──────────────┘ ▼ ┌───────────────────┐ │ RISK MODEL │ │ │ │ Probability │ │ Fraud class │ └─────────┬─────────┘ │ ▼ ┌───────────────────┐ │ POLICY ENGINE │ │ │ │ Permissions │ │ Approval routes │ │ SAR rules │ └─────────┬─────────┘ │ ▼ ┌───────────────────┐ │ NBA BEFORE │ │ EVIDENCE │ └─────────┬─────────┘ │ uncertain? / \ no yes │ │ │ ▼ │ ┌──────────────┐ │ │ EVIDENCE │ │ │ SELECTION │ │ └──────┬───────┘ │ │ │ ▼ │ ┌──────────────┐ │ │ BELIEF UPDATE│ │ └──────┬───────┘ │ │ │ ▼ │ ┌──────────────┐ │ │ NBA AFTER │ │ │ EVIDENCE │ │ └──────┬───────┘ │ │ └──────┬───────┘ ▼ ┌───────────────────┐ │ ACTION / APPROVAL │ │ SAR / EXPLANATION│ └─────────┬─────────┘ │ ▼ ┌───────────────────┐ │ GRAPH MEMORY │ │ │ │ Case │ │ Findings │ │ Actions │ │ Evidence │ └───────────────────┘ Enter fullscreen mode Exit fullscreen mode The final system is not an unconstrained chatbot making banking decisions. It is a traceable investigation workflow in which: Graph → provides evidence Risk Model → estimates risk Policy → controls permissions Agent → chooses useful investigation work Human → provides required approvals Action Gateway → executes approved actions Graph Memory → preserves the investigation Enter fullscreen mode Exit fullscreen mode That separation is the central design principle behind GraphSentinel. The goal is not simply to predict fraud. The goal is to build an investigation system where evidence, reasoning, policy, actions, approvals, and outcomes remain connected and auditable.