Reconciliation
Shipped — pilot liveAgents in NeoXam Aro
Every reconciliation break investigated before your analyst opens it — root cause, evidence, recommended action, fully traced.
AI-powered, low-code reconciliation and data lifecycle platform for operations teams at financial institutions.
Embedded agents
What each agent does in Aro
Statuses are exact and published for every capability — shipped means running today, Beta and GA wave are always future tense.
Reconciliation Investigator
Shipped — pilot livearo.reconciliation-investigatortask mode
Investigates a reconciliation break end to end: reads the break detail, checks historical match patterns and 30-day pricing, streams its reasoning in business language, and returns a structured BreakAnalysis — a root cause among six classes, a confidence score and a recommended action (auto-correct, manual review or escalate). Delivered end-to-end as the platform's original pilot; the live Aro data connector lands in Beta.
Status note Aro has shipped AI-powered reconciliation in production since the EZOPS acquisition (2024): break matching, matching-rule suggestions and exception investigation run at clients today and unify on the governed platform at GA.
On the platform
Governed like every NeoXam agent
Aro invokes its agents on NeoXam Agents — one catalog, one runtime, one compliance posture across the suite.
Declared in the Catalog
Every agent is a versioned descriptor in the Git-governed Catalog, reviewed through pull requests. Nothing undeclared can run — the catalog-first integrity contract.
One Runner API
Product screens invoke agents through the same stable Runner API — auth, tenancy, rate limiting, schema validation and run lifecycle included.
Traced, with cost
Every model and tool call emits an OpenTelemetry span, with token usage and cost per run, per agent and per tenant.
Eval-gated
Each agent ships a versioned eval dataset and a baseline score that can only go up; CI re-runs evals on every change.
Use cases
Aro workflows, agent-assisted
Reconciliation break investigation
Shipped — V0 pilotOne click on "Investigate with AI": the agent returns probable cause, evidence links and a suggested resolution with accept, escalate or annotate buttons — minutes instead of an hour per break, every step traced.
Audit reconstruction on demand
ShippedAny past investigation can be reconstructed step by step: descriptor version (catalog SHA), model, ordered tool calls with inputs and outputs, tokens, cost, duration.
Solutions
Built for these teams
COO · Head of Operations · Fund Administrator
Operations
Absorb the investigation toil on breaks and exceptions. Analysts review structured findings — root cause, evidence, recommended action — instead of digging.
Fund Accountant · NAV Oversight
Fund Accounting
NAV exceptions will arrive pre-investigated with audit-grade evidence — a faster close without losing control, every finding ready for the depositary.
FAQ
Questions about agents in Aro
Is the reconciliation agent real today?
Yes. The aro.reconciliation-investigator is delivered end-to-end as the platform's V0 pilot, running on the governed runtime with catalog, tracing, audit log and eval CI. It currently runs against a contract-tested, fixture-backed connector; the live Aro database connector lands in Beta.
Does the agent resolve breaks automatically?
No. It investigates and recommends — auto-correct, manual review or escalate — and your analyst decides. Outputs are typed and schema-validated, with evidence links and a confidence score.
What happens when the agent is wrong?
A malformed output is a hard failure, never a silent success, and every finding carries its evidence so analysts verify before acting. The agent ships with a versioned eval dataset and a baseline score that can only go up.
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.