Plans

Three plans, sized by the work.

Every plan is a custom agent, built in house and kept current, sized by how many systems, how much judgment and how many rules the work carries.

The plans

Relay, Navigator, Sentinel.

Relay

One system, clear rules, no judgment calls

  • Systems
  • Document AI
  • Rule depth
  • Autonomy
Orchestration
Single application, deterministic workflow
Inputs
Structured: forms, sheets, exports
Decisions
Rule based field mapping
Health data
Not included
Verification
Field level read back
Schedule
Scheduled runs
Audit
Run receipts

Navigator

Several systems, health data, some judgment

  • Systems
  • Document AI
  • Rule depth
  • Autonomy
Orchestration
Multisystem orchestration
Inputs
Unstructured too, through AI vision document understanding
Decisions
Confidence gated judgment with human in the loop routing
Health data
PHI aware handling under a BAA
Verification
Field level read back
Schedule
Scheduled runs
Audit
Receipts and decision traces

Sentinel

Complicated tasks, many systems, detailed rulebooks

  • Systems
  • Document AI
  • Rule depth
  • Autonomy
Orchestration
Multisystem state machines
Inputs
High variance documents with cross document checks
Decisions
Deep rulebooks with an exception taxonomy
Health data
PHI aware handling under a BAA
Verification
Read back plus cross system reconciliation
Schedule
Scheduled autonomous runs, around the clock
Audit
Full audit trail, evidence kept on your workstation

Every plan includes

Built, run and kept current, on every plan.

10 engineering hours a month

For upkeep and changes.

Remapping when your software updates

From those hours.

Health checks with alerts

Launches with the alert path.

Connection loss recovery

A weekly 30 minute check in

One point of contact on each side.

Support by severity

The standard is below.

Larger changes scoped first

A new workflow or system is a scoped build, never a surprise invoice.

Support standard

Response targets by severity.

SeverityWhat it meansRelayNavigatorSentinel
CriticalThe agent has stopped and work is not moving2 hours, around the clock1 hour, around the clock30 minutes, around the clock
StandardRunning, but a step needs attentionNext business day4 business hours2 business hours
ChangeA new rule, field or screenScheduled into the monthly hoursScheduled into the monthly hoursScheduled into the monthly hours

Response means a person is working on it, and because the agent stops instead of guessing, a critical issue delays work but never files anything unverified.

What the monthly hours cover

Changes that keep the agent current.

Rule set changes

A guideline changes, the rule set is versioned, and the decision trace follows.

Versioned rule sets · Decision trace

New fields and small new steps

Another field to read, another screen to confirm, inside the same workflow.

Remapping after software updates

Screen state detection is remapped and revalidated before unattended runs resume.

Screen state detection · Adapter revalidation

Report changes

A new column, a new recipient, a new schedule.

Run ledger reviews

Receipts and traces read with you, exceptions explained.

Evidence capture · Audit trail

A new workflow or system

Scoped and agreed first, as its own build.

Every plan starts with a demo.

Book a demo