Create the knowledge.
The knowledge your agents need, built from your existing systems.
ContextOne
Give every agent an understanding of your business.
ContextOne gives your agents shared business context, governed access to live data and tested tools they can reuse across your existing stack.
A shared foundation.
A head start for every agent.
Detect changes.
Source changes are detected and updates prepared to keep shared knowledge current.
Learn through use.
Corrections and resolved exceptions enrich shared knowledge for every connected agent.
Shared context. Built for every agent.
Business meaning connected
to live data.
Connect what your business means with the data your agents need. ContextOne uses shared definitions, relationships and permissions to query connected systems at runtime and return relevant results.
Your agentsYour models. Your orchestration.
MCPAPI
Retrieve contextDiscover and call tools
ContextOne
Agent Ontology
Shared business contextBusiness termsData definitionsSkillsTool definitionsEvaluation criteriaSource references
Typed relationships
Source provenanceTime validityAccess tags
Governed tool access
Discover permitted tools. Authorize and execute calls.Discover permitted tools↓Authorize the call↓Execute the call
Queries ↓ Results ↑
Governed data access
Query connected systems at runtime.
Combine results across sources.
Governed updates ↑
Continuous hydration
Build from sources.
Detect changes. Propose and test updates.
Tool calls ↓ Results ↑
Connected tools + APIs
Call business tools.
Return results to your agents.
Identity across every boundaryOAuth + delegated tokensPermission checks at each service
Databases & warehouses
Business applications
Documents & files
Code & APIs
Connected meaning. Grounded in your sources.
Typed entities and relationships carry source provenance, access tags and validity windows. Existing semantic models, catalogs and business rules contribute to the shared context. Agents retrieve connected context, including the state at a specified point in time. OSI and OKF import and export support interoperability. Your source systems remain authoritative.
Context should change the way code does.
Continuous hydration.
Creating context is the starting point. Keep it dependable as sources, permissions and business rules change. Updates are reviewed, tested and versioned before your agents use them.
New sources + system changesCorrections + agent learning
Context development lifecycle
Detect
Identify source changes and lessons from agent use.
Propose
Prepare updates with source evidence and dependency impact.
Review
Route changes to the responsible domain owners.
Evaluate
Test proposed context against defined cases.
Adopt
Apply approved changes atomically. Retain previous versions.
Publish
Make updated context available to entitled agents.
Published context
Available while the next update takes shape.
Your agents keep using the published version. Each new version is released with its history intact.
Repeatable work.
Ready to run.
Turn calculations, queries and repeatable steps into tested tools your agents can reuse. Your agents choose the tool and supply the inputs. The published code does the work.
Tested once. Reused across agents.
Built to operate.
On your terms.
How evaluation fits your stack
Use familiar work to define test cases and a baseline. Context and reusable tools are developed and tested in an evaluation environment, then you validate the changes in your own agent and framework. Compare accuracy, response time and cost on the same cases with the model held fixed. Measure time to the first validated outcome and how quickly the next agent reaches your standard using shared context and tools. Your agents retain their models, orchestration and execution.
How the knowledge is built
ContextOne reads schemas, queries, procedures, logs and documents in place, resolves entities and drafts source-backed knowledge. Your team controls approval before use, without authoring the knowledge from scratch.
How continuous hydration works
New sources, changes in live systems and lessons from agent use inform source-backed updates. Changes move through review and evaluation before adoption and publication. Published context stays available while updates are prepared; approved changes are adopted atomically, with previous versions retained.
Permissions and traceability
Retrieval uses the user's identity and permissions. Restricted records are excluded; results identify supporting sources. Approval controls govern knowledge updates, with a history linking each change to the creating session and its evidence. Inspect differences and retrieve knowledge at a specified point in time.
Option 1: Your cloud, managed by AI One
Choose bring your own cloud (BYOC) to run ContextOne in your cloud account with AI One managing the deployment. Your models and identity provider stay in place; your source systems remain authoritative.
Option 2: Self-hosted, operated by your team
Alternatively, your team deploys and runs ContextOne in your own environment, in the cloud or on premises. You control the infrastructure and operations.
How does this fit with our data and AI platforms?
Keep the platforms you already use, including Databricks and Snowflake. ContextOne connects business context across your systems and makes it available to your agents through MCP or API. Your source systems remain authoritative, and you control the context and its use.
What would we need to maintain if we built this ourselves?
Beyond the initial connections and definitions, the ongoing work includes tracking source changes, enforcing permissions on each request, evaluating updates, maintaining version history and making approved learning reusable. ContextOne provides that shared capability so your team can focus on its agents and applications.
Use the work you already have
Existing semantic models, catalogs and business rules provide the foundation. Governed definitions are reused, with their source systems remaining authoritative. ContextOne connects operational meaning across systems without replacing your data platform or moving the underlying data.
Interoperability and ownership
ContextOne supports OSI and OKF import and export. Your business knowledge stays yours, with support for exchanging it across compatible systems.
Read in place, without migration
ContextOne reads through the access your systems already expose. There is no migration of your operational systems. The governed knowledge index is distinct from the underlying source data; storage and retention requirements are agreed as part of the deployment.
Start with one agent.
Build a foundation for the next.
Bring your agents
Start with an agent you already run.
Connect its sources
Connect the systems it uses. The business knowledge is created for you.
See the difference
Compare speed, accuracy and cost on work it already does.
Put it to work
Apply the improvements in your existing stack. Reuse the context with your next agent.