Operations analyst · fund administrator
Shipped — V0 pilotReconciliation break investigation
One click on a break: the agent investigates match history and pricing, then returns root cause, evidence and a recommended action — fully traced.
The platform's original pilot, delivered end-to-end on the governed V0 runtime. Live Aro data connectors arrive in Beta; the pilot runs on contract-tested, fixture-backed connectors today.
In Aro
The daily reality
The work this replaces
Every morning starts with a queue of unexplained breaks. Each one means digging through match history, price movements and confirmations before judgment can even begin — expert hours spent on retrieval, not decisions. Exception volume scales with markets; analyst headcount does not.
How it works
What the agent actually does
aro.reconciliation-investigator
Invoked from the break screen
The analyst clicks 'Investigate with AI' on a reconciliation break inside Aro. The product calls the platform's Runner API — a bounded task, not an open-ended chat.
Reads the break in context
The agent pulls the break detail, historical match patterns and 30-day price history through governed Aro tools (MCP servers) — each call policy-checked and traced.
Streams its reasoning in business language
Progress streams live — 'checking the break's price history…' — so the analyst follows the investigation, not a spinner.
Returns a structured BreakAnalysis
Probable root cause classified among six classes, a confidence score, links to the evidence, and a recommended action: auto-correct, manual review or escalate. Schema-validated — a malformed result is a hard failure.
The analyst decides
Accept, escalate or annotate. The decision stays with the human; the investigation toil does not.
The outcome
Outcome
Minutes instead of an hour per break — analysts review structured findings instead of digging, and every step of every investigation is traced.
Governed by design
Every run records the exact agent version (catalog SHA), model, ordered tool calls, tokens, cost and duration in an append-only audit log — and the analyst always makes the final call.
FAQ
Questions teams ask about this
Does the agent resolve breaks on its own?
No. The agent investigates and proposes: root cause, confidence score, evidence links and a recommended action (auto-correct, manual review or escalate). The analyst accepts, escalates or annotates — resolution stays a human decision inside Aro.
What exactly is shipped today?
aro.reconciliation-investigator is the platform's original pilot, delivered end-to-end in task mode on the governed V0 runtime — catalog, tracing, audit log and eval CI included. Live Aro data connectors land in Beta; the pilot currently runs against contract-tested, fixture-backed connectors.
Can we audit an investigation after the fact?
Yes. Every run is reconstructable from the append-only audit log: agent version (catalog SHA), model, ordered tool calls with inputs and outputs, tokens, cost and duration. The log is designed for 7-year retention with CSV export for compliance roles.
Go further
Related use cases and concepts
NAV oversight & accounting investigation
In product · unified at GANAV exceptions arrive pre-investigated — hypothesis, evidence, source links — so the close gets faster without losing control.
Compliance breach analysis
In product · unified at GABreaches investigated and explained with traceable evidence at the moment they are flagged — no more manual reconstruction.
Data-quality remediation
In product · unified at GAData-quality errors investigated in the context of the feeds, rules and configuration involved — so remediation becomes review-and-apply.
Join the Early Adopter Program
General availability lands in Q3 2026. The Early Adopter Program is open now — a limited cohort, a one-year platform trial, and three workshop streams: Business ROI, Compliance, and Operational fit.