Whale AI is an intelligence layer woven into every module rather than a chatbot beside them: 50+ production use cases across cost, security, ITSM, operations, migration and governance, each grounded in your live inventory, billing, tickets and findings. It runs in three tiers — Spark, Tide and Abyss — with your choice of model, including fully offline inside the perimeter for sovereign estates.
- 50+
- Use cases, each with declared data dependencies
- Spark · Tide · Abyss
- Three tiers routed by task complexity
- Offline
- In-region and air-gapped models for sovereign deployments
How it is built
Grounded, not generative guesswork
Every use case declares the providers it reads — inventory, cost, tickets, findings — and cites them in the answer; numbers come from the platform, not the model.
Tiered intelligence
Spark answers interactive questions in under three seconds; Tide reasons over migration plans and IaC; Abyss runs multi-step agents for the hardest cross-module work.
In the flow of work
An assistant drawer in every module, streamed answers, exports to PDF and Excel, and prompt caching so repeat questions are instant.
Inside the perimeter
Bring your model or run ours in-region; sovereign deployments run Whale AI fully offline with zero external data exposure.
From question to grounded answer
- 1
Ask, in context
From any screen, ask about what you are looking at — a workload, a bill, an incident.
- 2
Ground
The use case gathers the live facts it declares, scoped to your permissions.
- 3
Answer with citations
The right tier reasons over the facts and streams an answer with the evidence behind it.
Three tiers
Spark — Standard & up
Entry, fast — summaries, inventory Q&A and sizing recommendations in natural language.
Tide — Enterprise & up
Deep reasoning for migration planning, IaC drafting and multi-cloud analysis.
Abyss — Enterprise · Government (in-region, offline)
Premium multi-step agents for the most complex, cross-module workflows — including fully offline inside the perimeter.
Use cases by category
| Category | Use cases |
|---|---|
| FinOps & cost | 12 |
| Security | 10 |
| ITSM & incidents | 8 |
| Cloud operations | 8 |
| Infrastructure | 6 |
| Observability | 5 |
| Executive intelligence | 4 |
| Governance | 4 |
| Migration | 3 |
| Performance | 3 |
Counts as shipped in the September 2026 release; every use case declares the data providers it reads and cites them in the answer.
Grounding providers
- resource_inventory — the live resource model and workloads
- cost_summary and cost_recommendations — billing joined to inventory, rightsizing and commitments
- scanner_findings — security and compliance findings
- open_incidents — ITSM incidents and changes
- tenant_metadata — tenant scope and permissions
Questions security and data teams ask
Does our data leave our environment?
On SaaS, requests are processed in your chosen region; on BYOC they run in your accounts; on sovereign deployments Whale AI runs fully offline on in-region models with no external calls.
Which models are used?
Your choice — hosted models including Anthropic Claude, or models you bring and run inside the perimeter. Use cases are model-agnostic; grounding and citations work the same way.
Can it take actions?
Use cases draft plans, reports and configurations; actions still go through the platform's approvals and audit trail, so Whale AI never bypasses governance.