Tableau Cloud
Tableau Server in AWS
Tableau Server in Azure
Tableau Server in Google Cloud
Tableau Server on-premises
Salesforce Data 360
Snowflake Horizon Catalog
Databricks Unity Catalog
Palantir AIP OntologyOne login for the whole suite.
Sign in once with your own Tableau token and every tool is open to you, carrying the permissions Tableau already gives you.
How the platform worksTabTotal: AI Design, Administration, Orchestration & Governance for Tableau Server & Cloud.
Boreon’s TabTotal reads every warehouse, lakehouse and ontology you run. It answers on Claude, GPT or your own OpenAI-compatible endpoint, and every number comes from a metric you certified.
- Tableau Cloud
- Tableau Server
- Salesforce Data 360
- Databricks Unity Catalog
- Snowflake Horizon Catalog
- Palantir AIP Ontology
- Anthropic
- OpenAI
- Azure OpenAI
Built by certified Tableau Architects and Consultants, for the job we do ourselves.
Build with Agentic AI
Describe the dashboard. Get a real Tableau workbook.
Pick a published datasource and design through conversation. The AI drafts the spec; a deterministic emitter writes the workbook bytes.
A quarter of dashboard work, delivered the same day.
Ten minutes to a published dashboard in the project you chose. Hand it to a Tableau developer for final dressing and it is an afternoon, not a quarter.
Full data validation on every figure it can read. Each zone is re-queried live against your Tableau Cloud datasource and the check is shown, design value beside live value, and any zone that cannot be checked is named rather than quietly passed.
Every figure is read from your own governed datasource on your own session, and checked against live data before anything is published. What arrives is a real .twb tied to that datasource, ready to publish into the project you chose.
Where it sits
One layer over your Enterprise Tableau, Data Warehousing & Ontology Environments
You
Tableau Cloud AdminsTableau Server AdminsAI Tableau CreatorsArchitectural CommitteesSOC Auditors
Boreon’s TabTotal Enterprise Semantic Layer
14 tools, one login, and an AI that runs on a key you own.
- Claude
- OpenAI
Your Tableau Environment
Tableau CloudTableau Server in AWSTableau Server in AzureTableau Server in Google CloudTableau Server on-premises
Your Data Warehouse & Ontology
Salesforce Data 360Snowflake Horizon CatalogDatabricks Unity CatalogPalantir AIP Ontology
Your Data
TablesColumnsCommentsTagsKeysGrantsOwnersDescriptionsClassificationsLineage
Everything your Tableau Environment needs, in one place.
One login above every governed tool. One semantic contract beneath them, and an AI across everything.
A semantic layer over four data platforms.
Inherit Salesforce Data 360, Snowflake Horizon Catalog, Databricks Unity Catalog and the Palantir AIP Ontology into one governed business layer. The same entity in several platforms is one object, and where any two of them disagree, that becomes a finding.
Inside the semantic layerAn AI that knows your Tableau Environment.
Ask about any workbook, datasource or lineage path and get an answer from your own governed model. Anything the Agent proposes arrives as a card your team approves, through the same gates you already trust.
How the AI worksThe Enterprise Semantic Layer
Define a metric once. Get the same number everywhere.
A read-only credential brings in your tables, columns, comments, tags, keys and grants.
Inside the semantic layerConnect a warehouse without giving up write access.
The connectors contain no write method at all, so a read-only credential is all TabTotal ever needs. Whatever your role cannot see, Boreon cannot see either.
Field-level lineage
One dashboard, traced to the columns it actually reads.
How the answer gets there
Boreon’s TabTotal reads Tableau. Claude or GPT does the answering.
Every AI Agent turn runs on a real model, on your own key. No surface hardcodes which one: it asks the running site which provider is live and renders that name.
Claude
- The default provider. With no other key configured, every AI Agent turn on the site answers on Claude.
- Three tiers, matched to the task: Sonnet, Opus and Fable.
- Your own Anthropic API key, vaulted and never echoed back.
OpenAI
- Point it at plain OpenAI, Azure OpenAI, or any OpenAI-compatible endpoint you host yourself.
- The counterpart tiers: GPT-4o and GPT-4.1.
- Switch providers without touching a line of code. Every label on the site relabels itself.
Why Boreon’s TabTotal Orchestration Lake
Only the layer on top can check that they agree.
Everything flows into one governed body, and only the layer on top can see across all of it. Salesforce Data 360, Snowflake, Databricks and Palantir AIP each verify themselves, and TabTotal reads them with your read-only credentials.
A measure is computed one way: the way you defined it.
The identifiers that reach the SQL are the contract’s own copies, never text from the model or the caller. Nothing the model writes has a path to your warehouse, so the guarantee survives a model that gets things wrong.
One table, one object, across every warehouse holding it.
Fields union by name, the description that exists wins, and the merged object remembers every warehouse it came from.
A question that strays across warehouses stops before it runs.
If the names in one question belong to different warehouses, nothing is sent anywhere, and the refusal names the candidates so you can ask again precisely.
Where any two warehouses disagree, the object is named.
Every pair of connected warehouses is compared, and each finding names its own pair: the object one of them holds and the other does not, or the object both hold with different field counts.
When they match, the count reads zero.
Each warehouse defines its metrics its own way, and those objects are excluded on purpose and never counted as drift. Warehouses that agree report nothing, however many you connect.
Divergence lands in your governance score.
Divergence between warehouses joins the same findings total as every other governance job. The score you already read is where it lands.
The suite
14 governed tools, behind one login.
Every tool answers deterministically on its own. AI is yours to switch on per run.
Govern
Prove what changed, who can reach it, and what it exposes.
Orchestrate
Trace what feeds what, watch how it runs, and put the right view in front of the right people.
Design
Describe what you want, or bring a file you already have, and get governed content back.
Evidence
The platform writes its own audit trail as it works.
In the formats your auditor asks for.
| Artefact | What it records | What it never records | Export |
|---|---|---|---|
| Admin audit | Every administrative change: who did what, to whom, and when. | No password, hash or authenticator secret, ever. | XLSX |
| Change ledger | Governed configuration changes in a SHA-256 hash chain, with before and after content hashes. | No secret values. Only their hashes. | CSV · JSON |
| AI log | Every AI call: surface, model, tokens, outcome. | Conversation content is not recorded by default. | JSONL · XLSX |
| Token log | PAT and embed-JWT events: minted, used, expired, refused. | Secret-free by construction, not by scrubbing. | XLSX |
Where it runs
Inside your own network, under your own admins.
The Enterprise license is one container on a Docker host you already run, behind the identity provider and the perimeter you already operate.
Every tool works fully without AI. Turn it on and it runs on your own provider key, or on your own OpenAI-compatible endpoint inside your network, so you decide where a prompt goes. If you would rather not host it yourself, we can run it for you, in the cloud of your choice.
Nothing changes until a person says so.
The AI reads, explains and proposes. Anything it suggests arrives as a card that routes into the same approval gate a person would use, with the same signature and dual-control rules. Publishing always creates a new copy and never overwrites your work.
See it on your own Tableau Environment.
One user, one month, on us. Point it at a real site with your own content.