Controlled automation with targeted AI
Rules you can audit. AI only where the input is messy.
Who gets called, which email gets the reply, and what happens when the carrier changes. Same input, same output, and a condition an auditor can point to.
Book a demoWhere do rules run, and where does AI read?
Rules run
Use the model where there is ambiguity to resolve, not where there is a list to check. Lists, recipient checks, attempt counts, and carrier swaps stay on rules.
AI reads
Reading the body of a carrier reply to extract a driver name and phone number, or interpreting freeform text, is model work.
Where is this already running?
Pre-dial availability check
Already running in production for a maintenance use case. The list is checked, prior calls are skipped, and availability is confirmed over an API before a phone rings. A failed check holds the call.
Read the availability guideEmail reply targeting
The workflow holds the thread identity for a load and checks recipients before replying. The operator chooses the first or the latest message. This works across Gmail and Outlook.
Read the email-thread guideCarrier change mid-load
A carrier change stops the previous run and starts a fresh one. Moneiva reads the updated assignment from the TMS or feed the same way it picks up a new tender.
Read the carrier-change guideNo-answer chase
The chase runs on rules. The rules set how many attempts, how far apart, and when silence becomes an exception. Each attempt, voicemail, and IVR outcome is logged.
Read the check-call guideThe same placement reaches past the phone and the inbox. Moneiva reads the TMS, warehouse systems, customer systems, and financials, and it writes the result back.
What can an auditor follow?
- Same input always yields the same output.
- An explicit condition you can point an auditor to.
- A failure that is loud, on a rule you can read.
- A new edge case is a new branch on the condition.
Those properties belong to the rule-based steps. Triggers, conditions, messaging templates, escalation rules, and integration mappings can be changed to match the operation.
What proof is published?
The breakdown deployment for a Top 5 U.S. Truckload Carrier runs with deep TMS integration, deterministic logic, and real-time escalation paths, 24/7. It handles 18+ distinct breakdown scenarios with scenario-specific follow-up logic. 80-89% of breakdown cases are opened automatically.
What is controlled automation with targeted AI?
Controlled automation with targeted AI runs the structured steps on rules and uses AI only where the input is messy, so a rule-based decision can be audited.
It sits between AI communications and the AI Operator. Communications collect the answer. This tier decides the next step from a rule you can read.
Keep going
Where this sits
FAQ
Frequently asked questions
What is controlled automation with targeted AI?
Controlled automation with targeted AI runs the structured steps on rules and uses AI only where the input is messy, so a rule-based decision can be audited.
When should a logistics process run on rules instead of AI?
Use the model where there is ambiguity to resolve, not where there is a list to check. A rule gives the same output for the same input, names a condition an auditor can point to, and fails loudly on a rule you can read.
Where does AI belong in freight automation?
Reading the body of a carrier reply to extract a driver name and phone number, or interpreting freeform text, is model work. The steps around that read stay on rules.
How does Moneiva decide who to call?
In production for a maintenance use case, the workflow takes the inbound list, checks whether a call was made previously, and confirms availability over an API before it rings a phone. If that check does not succeed, the workflow holds the call and retries rather than dialing blind.
What happens when the carrier changes mid-load?
When Moneiva detects a carrier change on a load, the automation for the previous carrier stops and a fresh run begins for the new one. It reads the updated assignment from the TMS or feed when it arrives.
Can rule-based decisions be audited?
Rule-based decisions can be audited. The same input always yields the same output, the condition is explicit, and a failure points at a rule you can read. Only those rule-based decisions are auditable.
Can the rules be changed to fit our operation?
Yes, all workflows are fully customizable. You can modify triggers, conditions, messaging templates, escalation rules, and integration mappings to match your processes.
Talk through one auditable decision
Map the rules that should always give the same answer, and the few places a model should read.
Book a demo