User Request
Chat, agent step, extraction task, or repeatable analysis run.Artificial Intelligence
Private Models
Private execution for governed agents and document AI
Clear Ideas gives teams two clear model lanes: Private Models for sensitive, recurring, and cost-aware execution, and Provider Models for tasks that benefit from premium capability.
Model strategy
Two Model Lanes, One Governed Workspace
Private Models and Provider Models are presented clearly inside Clear Ideas so users can choose the right model lane for each chat or agent run.
Use Private Models for sensitive analysis, recurring agents, structured extraction, internal reviews, and high-volume work where fully private Zero Data Retention execution and cost control matter. Use Provider Models when a step benefits from premium reasoning, specialized capability, or a specific model approved by your organization.
Private model path
Private Execution Lane
Private Models sit inside the same governed workspace as Sites, agents, chat, review, and evidence. Model routing follows policy instead of becoming an ad hoc user choice.
Policy Gate
Site permissions, organization model policy, and approved context.Model Routing
Private Models for sensitive work; Provider Models when approved.Governed Run
Scoped execution, source usage, generated outputs, and review state.Retained Evidence
Model lane, sources, tokens, output, and audit trail remain together.Private model routePolicy-controlled execution for sensitive work
Governed execution
Use the Right Model Lane for Each Task
Private Models are most useful when AI tasks move from one-off prompts into repeatable, reviewable operations. The point is to make model choice explicit enough that teams can match sensitivity, cost, and capability to the step being run through an approved model catalog and organization policy.
- Zero Data Retention ExecutionRun sensitive agent steps and document-grounded analysis through fully private Zero Data Retention Private Models when the task calls for the strongest privacy posture.
- Controlled Agent RuntimeKeep model choice tied to scoped run contracts, approved Site context, connector permissions, review gates, continuation state, and structured run events.
- Provider Model ChoiceKeep access to Provider Models for jobs that need premium reasoning, broad multimodal capability, or a specific provider model approved by your organization.
- Organization Policy ControlAdministrators can set permitted-model policy so users see only approved Private Models, Provider Models, aliases, and Intelligent selection options for the workspace.
- Model Catalog VisibilityModel records can define provider, alias, system status, private grouping, UI metadata, and allowed-user visibility so model availability is explicit rather than ad hoc.
- Agent-Ready OperationsUse Private Models for repeatable agent runs where Zero Data Retention, cost, reviewability, and predictable execution matter together.
- Governed EvidenceKeep model choice connected to the run record, source usage, generated files, credits, tokens, review points, and evidence exports.
- Benchmark ChecksUse benchmark checks to test recurring agent work before teams rely on it operationally.
- Cost-Aware RoutingChoose Private Models for high-volume or recurring work, and reserve Provider Models for steps where their additional capability changes the business outcome.
Agent design surfaces
Design With Advanced Tools. Execute in Clear Ideas.
Teams can use tools such as Codex and Claude Code to design, inspect, and improve agents. Agent Designer shows how those definitions run in Clear Ideas against approved Sites, scoped connections, organization policy, benchmark checks, review gates, and evidence.
Cost strategy
A Practical Cost Strategy
The goal is not to make every step use the same model. The goal is to route work to the model lane that matches the task, sensitivity, review burden, and expected volume.
- Use Private Models ForRecurring agents, structured extraction, review workflows, document-heavy analysis, internal summaries, benchmark runs, and high-volume tasks where Zero Data Retention and operating cost are primary decision factors.
- Use Provider Models ForComplex reasoning, specialized writing, broad multimodal work, edge-case analysis, and review steps where a premium provider model materially improves the result.
- Use Both TogetherDesign the process once, then route each step to the right model lane. Many agent runs can use Private Models for routine work and Provider Models for the few steps that need extra capability.
Related reading
Model Strategy, Governed Agents, and External AI Access
Model strategy, governed agents, and external AI access for sensitive work.
- Private Models vs Provider Models for Agent ExecutionA practical comparison for choosing the right model lane for governed agent runs.
- The Best AI Model Is a Governed SystemWhy model choice should be tied to task scope, approved documents, and review requirements.
- AgentsRun repeatable AI work with approved sources, review gates, generated files, and evidence.
- Agent DesignerDesign agents from prompts, Codex, Claude Code, or MCP clients before governed execution.
- Clear Ideas + Codex via MCPUse Codex with governed Clear Ideas context while preserving controlled access.
Frequently Asked Questions
What are Private Models in Clear Ideas?
Private Models are fully private Zero Data Retention model options available inside Clear Ideas for governed AI work. They appear separately from Provider Models so users and administrators can make model choices deliberately.
How are Private Models different from Provider Models?
Provider Models are models supplied by external AI model providers and selected for their capability, specialization, or provider preference. Private Models are selected when the organization wants fully private Zero Data Retention execution, predictable operating costs, and a tighter policy posture for recurring AI work.
Can administrators control which model category is available?
Yes. Organization policy can allow Private Models, Provider Models, or both. Teams can align access with the sensitivity of the workspace and the type of task being performed.
Do Private Models replace premium Provider Models?
No. They give teams another governed execution lane. A practical model strategy often uses Private Models for recurring and high-volume work, while using Provider Models for steps where additional capability is worth it.
How do Private Models fit with Codex and Claude Code?
Authoring tools such as Codex and Claude Code can help teams design, inspect, and refine agent behavior. Agent Designer shows how those designs run against approved Sites, policies, review points, and model choices inside Clear Ideas.
Ready to get started?
Share sensitive information securely with clients, auditors, and partners. Then turn approved content into cited answers, Governed Agents, and measurable engagement.
Start Free
Book a DemoNo credit card required
Need help?
Get personalized assistance
Speak with our sales team to find the perfect plan for your organization.
Technical support & resources
Access support resources, documentation, and help guides.