Deterministic step budget · 2 MB scored in ~3.4 ms · 947-byte edge bundle, no database · Blue/green rollback
See how it compares to ODMKnowledge Hub
Deep dives into WebAssembly decision runtimes, IBM ODM migration strategies, measured latency methods, and AI-assisted BAL 1.0 rule governance.
A detailed financial and architectural analysis of IBM Operational Decision Manager licensing costs, server overhead, VPC pricing, and how modern WASM edge execution dramatically lowers total cost of ownership.
Dr. Alex Vance
Head of Enterprise Systems Architecture
DecisionManager ships two testing instruments with different purposes. Test Suites are hand-written, per-case, release-gate tools. Scenario Sets are DVS-scale bulk runners with coverage KPIs. Understanding which to use — and when to use both — prevents coverage gaps at the point of promotion.
The Decision Manager compare page produces a deep-linkable, structured diff between any two ruleset versions — changes, structured, and raw views. Reviewers get a permanent URL for a specific comparison so a finding can be dropped into a review thread and open on the same rows.
Decision Manager's Pipelines feature brings rule-aware CI/CD to governance-controlled rulesets. Each pipeline step knows the governance state of its artifact, runs against a specific version, and exposes its status via a permanent run URL for deep-linking from CI systems.
The DecisionManager Admin console ships a webhook engine with at-least-once delivery guarantee, exponential-backoff retry, and an AI-assisted delivery diagnosis panel. Every delivery carries X-DM-Delivery for idempotent deduplication on the receiver.
DecisionManager's Impact Analysis page scans every ruleset in a project for vocabulary binding references, showing exactly which rulesets and versions will break if you rename a vocabulary term. The same tool surfaces ruleset dependency graphs for cross-artifact impact.
The Rule Hotspots panel combines telemetry-ranked cost data with AI-generated explanations of what each expensive rule does. The ranking is never asked of a model — only the narration is. Here's how it works and how to use it to optimize step budget.
A release that errors is caught by every monitor. A release that keeps deciding at the same latency but starts deciding differently is caught by nothing — unless you measure outcome distributions. The DecisionManager Outcome Drift panel does exactly that, using PSI and shape movers.
The RCA panel gives on-call engineers the answer to 'is anything wrong, explain this window, show me the evidence' without opening four consoles. Execution anomaly detection finds statistical outliers in the decision trace; RCA narrates the window in business terms with a SIEM-exportable artifact.
The RuleApp Builder is a six-step AI wizard that generates a complete decision service — vocabulary, rulesets, decision tables, and ruleflow — from a plain-English description. Every artifact is editable before creation. Nothing is committed without human review.
The Blast Radius panel answers 'if I change this, what breaks?' by crossing the structural diff with the dependency graph. Which consumers pin a version vs. float to latest-published? Which are live in production? AI narration explains the behavioural change per consumer.
The Boundary Coverage panel derives boundary values directly from stored rule definitions — both sides of every comparison, every enumerated domain, every unconstrained input. AI generates test cases, each verified by server-side execution before being offered. No token spent on unverified suggestions.
During a progressive rollout at 25%, the operator must decide whether to widen to 50%. The Deployment Risk panel shows how many decisions per day are on the challenger, what the challenger's p95 and error rate are, and what rules changed — before the decision is made.
The Version Regression Explainer replays real recorded decisions through both versions and reports which output keys moved, which rules churned, and whether 'same answer, different rules' counts as equivalence. Then the AI explains what changed in business terms.
The Infrastructure console generates Terraform for AWS Lambda, EKS, GCP Cloud Run, and Azure Container Apps — with a dry-run step and a typed apply guard. The Terraform is checked into your repo, not managed by the platform.
The Migration Runs workspace is the right shape for a real IBM ODM migration: persistent runs, per-artifact review rows, scenario replay against what shipped, and certificates. The direct import path is for evaluation — the runs path is for migration.
The Audit Brief panel answers 'what changed here, and what should I look at?' across an append-only audit trail where every service writes its own detail payload. The model is asked to read — counting and classification stay with the console.
A delivery failure row carries two fields nothing in the console can interpret: 'error' (whatever the HTTP client reported) and 'response_body' (the receiver's own words). The AI reads them and turns them into actionable diagnosis — without touching the counts or replay actions the deliveries view already handles.
The Library page is the Decision Center–style entry point: see all decision services across all projects, open an existing service, or create a new one through the guided service wizard. No ODM project folder navigation required.
DecisionManager's global search index covers all rulesets, vocabularies, decision tables, action rules, and test suites across all projects. Search by content, by artifact type, by author, or by governance state.
The Reports Hub in both the Manager and Admin consoles provides structured exports for enterprise documentation: project catalogues, governance trails, analysis reports, member and group rosters, API key inventory, and compliance evidence — all in CSV and JSON.
The Admin console's Permissions Matrix, Limits, and Usage pages complete the enterprise administration story: granular permission overrides beyond the four default tiers, per-workspace execution limits, and real-time usage dashboards with AI spend tracking.
Plans without a public rate card. The live demo runs a real ruleset in the browser without an account. Free trial — no credit card, no auto-conversion.