Proceed
The action may proceed within the authority and conditions recorded for this governance decision.
DGL®
Dynamic Governance LayerDGL controls autonomous AI.In real time.
DGL assesses requests and responses, governs decisions and actions, determines when intervention is required, and records accountability and evidence.
Proceed / Review / Escalate / Withhold
DGL has been successfully deployed into a live agentic pilot.
Governance as AI operates
DGL governs AI requests, responses, decisions and actions within connected workflows as the system operates.
It assesses the request and response, determines whether to proceed, require review, escalate or withhold, and records the governance evidence. Where intervention is required, the record distinguishes the system’s assessment from the accountable person’s decision.
The connected system enforces the governance outcome. Request assessment, response assessment, the final decision and evidence form the control process.
Those actions can affect people, patients, customers, money, operations and assets. They can also create legal or regulatory responsibility and organisational liability.
A governed event may begin with a human, an AI agent, an application, an API, an automated workflow or another system.
How DGL works
Request Governance → Response Governance → Final Governance Decision → Evidence
- 01
Request Governance
Assesses the proposed request or action against the supplied authority, policy and risk context.
- 02
Response Governance
Assesses the AI response, where applicable, before it can be used in the proposed action.
- 03
Final Governance Decision
Determines whether the action can proceed, requires review or escalation, or must be withheld.
- 04
Evidence
Records the assessment and governance decision, with the relevant context and identities. A human decision is recorded where one is made.
Where review or escalation is required, an appropriately authorised person records a decision and rationale. The connected workflow must hold the action until the applicable requirements are met.
Four operational outcomes
Review
The action requires review by an authorised person before it can proceed.
Escalate
The matter requires a decision from someone with the necessary higher or specialist authority.
Withhold
The action cannot proceed on this governance decision. The record explains whether an applied control prohibits it or the assessment could not be completed safely.
Accountability and evidence
DGL connects the proposed action, the assessed controls and the governance outcome in an evidence record.
Where human intervention is required, the record distinguishes the system’s assessment from the person’s decision, including their recorded authority and rationale. Signing in or submitting a request does not by itself establish that a person approved an action.
Evidence supports operational review, audit and examination of how an organisation controls AI. The organisation remains responsible for its policies, authority assignments and the connected system’s enforcement.
Markets
The control problem extends across sectors. The authority, policy and risk context must reflect the organisation and the proposed use.
Insurance
Control over AI actions affecting claims, underwriting, customer communications and delegated authority.
Banking and financial services
Control over AI actions affecting customers, payments, financial decisions and access to sensitive information.
Healthcare — patient safety
As AI becomes more autonomous in healthcare, patient safety depends on retaining effective control over what it is allowed to do. Governance, authority, intervention, accountability and evidence support that purpose. DGL does not guarantee clinical safety.
Government and public services
Control over AI actions affecting people, public services, public resources and statutory responsibilities.
Infrastructure
Control over AI actions affecting operational continuity, assets, maintenance and essential services.
Industry
Control over AI actions affecting production, equipment, supply chains and operational responsibility.
These are areas of application, not claims of deployment or sector-specific validation.
Founder and company
DGL Systems is developed under Jules and Field Limited. Julian Tranter is the founder of DGL Systems.
Read Julian’s publisher profile and his published insurance commentary, dated 14 July 2026.
Patent application filed 21 June 2026 (GB2614319.8).
Discuss your use case
What should AI be allowed to do in your organisation? Where must an action be reviewed, escalated or withheld?
Contact Julian Tranter to discuss DGL, integration or a guided demonstration.
julian@dglsystems.ai