What context governance is
Context governance is the set of controls that defines, explicitly and auditably, which sources agents may use, who can access each part of the knowledge and how to prove, afterward, what was consulted. It is what makes enterprise agents safe enough to leave the pilot stage.
Without governance, "giving context to an agent" becomes an open risk: sensitive data can leak through poorly defined scope, unapproved sources can contaminate answers, and any audit is left without evidence. Chatydata treats governance as foundation, not as an optional layer.
Approved sources
Not every company document should feed an agent. Governance begins by curating which sources count: approving trusted origins, classifying by sensitivity and excluding obsolete or unofficial content.
This drastically reduces the risk of the agent grounding answers in the wrong material, and gives the risk team a clear inventory of everything that feeds the agents.
Scope and permissions per workspace and collection
Access to context is controlled in layers: by workspace (the organizational boundary), by collection (the thematic grouping of sources) and by agent (what each specific agent can query). So an HR agent does not see financial data, and an external agent does not access internal content.
These rules travel with the context: when the runtime fetches information, retrieval already respects the scope. Control does not depend on the runtime behaving well — it is enforced at context delivery.
- Workspace: Organizational boundary that isolates environments and teams.
- Collection: Thematic grouping of sources with its own access scope.
- Agent: Defines which collections a specific agent can query.
Audit trail
Every interaction leaves a trace: which sources were consulted, which version was active and which scope applied. When an answer is challenged — internally or by a customer — there is concrete evidence to investigate.
The audit trail also supports quality review: query patterns reveal problematic sources and help refine the context base over time.
Where governance acts in the architecture
Governance is not a report at the end of the process — it is enforced at the context delivery point, between sources and the runtime. Every retrieval passes through scope and permission rules before reaching the agent.
Fontes
Drive, SharePoint, ERP, CRM, PDFs, APIs
Chatydata · Context Engine
Organiza · versiona · governa · observa o contexto
Runtimes
via MCP · API · conectores · pipelines
Risks without governed context
The absence of governance turns each agent into a potential door for incidents. The most common risks:
- Sensitive data leakage. Open scope lets an agent deliver information to someone who should not see it.
- Unapproved sources. Obsolete or unofficial content contaminates answers with the appearance of truth.
- Audit failure. Without a trail of consulted sources, it is impossible to respond to an investigation or legal request.
- Data-protection non-compliance. Lack of access control and traceability makes it hard to demonstrate proper handling of personal data.
Compliance and GDPR
Context governance gives the company the foundations data-protection regimes require in practice: control over who accesses which data, a clear purpose for sources, and traceability of use. By recording consulted sources and restricting scope under the least-privilege principle, it becomes simpler to demonstrate compliance.
Chatydata does not replace the company’s legal work or privacy program, but it provides the technical controls that make that program applicable to agents.
Frequently asked questions
Does governance depend on the runtime behaving correctly?
No. Scope and permission rules are enforced at context delivery, before reaching the runtime. The agent only receives what it is authorized to, regardless of the runtime used.
Can I prove what an agent consulted?
Yes. The audit trail records the consulted sources and the active version in each interaction, allowing you to investigate and demonstrate the origin of any answer.
How does access control work?
In layers: by workspace, by collection and by agent. You define which collections each agent can query, following the least-privilege principle.
Does this guarantee data-protection compliance?
It provides the technical controls — access control, source purpose and traceability — that support a compliance program. Compliance itself also depends on the company’s policies and legal work.
Free assessment: we identify scope and governance gaps before you scale.
Assess the permission risks of your data for agents