USE CASES
AI coding, with organisational control.
Different teams bring different requirements to AI-assisted development. BatonX is designed to help developers work with coding agents while giving organisations clearer control over access, policies and activity.
The examples below illustrate intended workflows. Available capabilities depend on the supported agent and release.
Adopt AI coding with clear boundaries
Introduce AI coding agents through a consistent, governed workflow. BatonX is designed to bring approved repository access, managed instructions and runtime controls together while preserving the agent experience developers already know.
- Practical example
- A developer starts an agent session for an authorised repository. The applicable instructions and controls are applied before work begins.
- Intended outcome
- A repeatable approach to AI adoption, with less dependence on each developer configuring controls individually.
Illustrative workflow
A developer starts an agent session for an authorised repository. The applicable instructions and controls are applied before work begins.
- Approved repository
- Coding agent
- Policy controls
Keep agents within the approved workspace
Authorising an agent to work on a project should not mean giving it access to the developer’s entire machine. BatonX is designed to limit filesystem access to the approved workspace through technical isolation.
- Practical example
- An agent can work with files inside the authorised project, while a request to read SSH keys outside that workspace is blocked.
- Intended outcome
- A narrower access boundary around each development session.
Illustrative workflow
An agent can work with files inside the authorised project, while a request to read SSH keys outside that workspace is blocked.
- Project files
- Personal folders / SSH keys
- Blocked outside workspace
Apply shared rules across development teams
Give teams a consistent starting point for agent behaviour and permitted actions. BatonX is designed to combine centrally managed instructions with enforceable controls for commands, network access and approvals.
- Practical example
- An organisation requires approval for package installation. When an agent requests that operation, execution pauses for the required decision.
- Intended outcome
- More consistent policy application and explicit decision points for controlled operations.
Illustrative workflow
An organisation requires approval for package installation. When an agent requests that operation, execution pauses for the required decision.
- Package installation requested
- Organisation policy: approval required
- Awaiting approval
Manage provider access centrally
Reduce reliance on provider credentials distributed across developer machines. BatonX is designed to centralise provider access and protect credentials from developers and agents wherever technically possible.
- Practical example
- An authorised agent request uses the configured provider access path without requiring the developer to copy a shared provider API key into the project.
- Intended outcome
- Clearer ownership of provider access and reduced exposure of shared credentials.
Illustrative workflow
An authorised agent request uses the configured provider access path without requiring the developer to copy a shared provider API key into the project.
- Coding agent
- Managed access · protected provider credential
- Approved provider
Understand activity, usage and cost
Understand how AI coding agents are being used across your organisation. BatonX is designed to bring session activity, security events, model usage and cost information into an administrative view.
- Practical example
- An administrator reviews a project’s recorded sessions, investigates a blocked operation and examines usage and available cost estimates.
- Intended outcome
- Better evidence for operational decisions, investigations and budget discussions.
Illustrative workflow
An administrator reviews a project’s recorded sessions, investigates a blocked operation and examines usage and available cost estimates.
- Recorded sessions
- Blocked operation · policy event
- Model usage / available cost estimates
Keep developers in charge of delivery
Use AI assistance within a workflow that keeps delivery decisions with the developer. BatonX is designed to support incremental work, applicable approvals and a review step before changes enter the team’s delivery process.
- Practical example
- An agent prepares changes within the authorised workspace. The developer reviews the diff, checks the results and creates the merge request.
- Intended outcome
- AI-assisted implementation with explicit developer responsibility for the changes delivered.
Illustrative workflow
An agent prepares changes within the authorised workspace. The developer reviews the diff, checks the results and creates the merge request.
- Agent proposes changes
- Developer reviews diff and checks results
- Developer creates merge request
Find the right controls for your workflow.
Explore the platform behind these scenarios and how its intended security boundaries fit into everyday development.