Acuence · agentic AI
AI that works inside the process, not beside it.
Acuence is the agentic AI layer. It reads, drafts, checks and acts as a step in a versioned process, bounded by the carrier’s own rules, subject to the same approval gates as a person, and retaining the same evidence.
- Read application · 14 fields01
- Checked against carrier rule set v902
- Drafted requirement letter03
- Held for underwriter approval04
EVIDENCE RETAINED · PROMPT, SOURCES, DECISION, APPROVER ✓
An agent cannot reach production without passing the same gates
Two step types
AI is a step in the journey, with a version like any other.
A process owner adds an AI step in LifeStudio the same way they add a form, a system call or a rule. It is versioned with the journey, promoted through the carrier’s approval path and effective-dated before it runs.
Step type
Prompt
One bounded task with a defined input and a defined output: summarise a file, classify a request, extract fields from a document, draft a letter for review. The prompt, the sources and the output are retained with the case.
INPUT → PROMPT → OUTPUT → RETAINED
Step type
Agent
A bounded objective with permitted tools: read the record, call a system, check a rule set, prepare the work and stop where a person must decide. The path it took is retained, not reconstructed.
OBJECTIVE → TOOLS → ACTIONS → HANDOFF
The same rules as people
Nothing an agent does escapes the governance around it.
Acuence has no privileged path. It operates on the carrier’s own deployment, under the carrier’s rules, inside processes that were approved before they ran.
Bounded by carrier rules
An agent can act only within the tools, data and decisions the process grants it.
Approved before it runs
An AI step reaches production through the same gates and effective dates as any other change.
Evidence, not reconstruction
Prompt, sources, output, decision and approver are retained beside the case that ran.
Sensitive data stays masked
Personally identifiable information is masked, so an AI step reads what it needs and no more.
Where the advantage actually sits
The model is not the advantage.
Every carrier can reach the same models. What differs is whether the operation can put one to work in a process that examiners, auditors and your own risk function will accept, and then change it next quarter without a project.
That is what Acuence is for. It is not a copilot bolted to the side of a portal. It is an AI step inside a governed process, which means the interesting question stops being “can we trust the model” and becomes “what is this step allowed to do, and what did it do”.
Acuence runs on the same isolated per-carrier deployment as the rest of the platform. See the foundation →
See it under your rules
Bring a step you would not let AI near.
A NIGO review, a suitability check, a claim requirement letter. We will put an agent inside that step under your rules and show you what it is allowed to do, where it stops and what it leaves behind.
Go deeper: how a change is governed · the record it acts on