Eval directory · Accounting & Finance

Evals for Rillet

Eval coverage for Rillet, mapped from its public product surface.

About Rillet

Rillet is an AI-native ERP built for finance and accounting teams, centered on a perpetual general ledger that automates the monthly close. It covers revenue recognition, accounts payable and receivable, bank reconciliation, multi-entity consolidation, and real-time GAAP and investor reporting. Its AI layer, Aura, runs on the live general ledger to answer natural-language questions, run workflow agents, flag anomalies, and connect to outside tools via MCP.

Industry

AI-native ERP / accounting automation

Website

www.rillet.com

Use the eval library for Rillet

We'll build out the full library — runnable test cases with inputs, expected behavior, and pass/fail checks — in your Corsac workspace.

Related in Accounting & Finance

All evals →

More Accounting & Finance eval libraries

Coverage map

What would you measure for Rillet?

6 scoring areas · 24 capabilities mapped · grounded in 8 cited pages

Every eval set is graded on

Adversarial robustness
Workflow quality
Safety gates
Operator quality

Pass/Fail + LLM judge 1–5 · critical severity flags · negative controls

Aura AI assistant and agents

The intelligence layer running on the live general ledger: natural-language questions answered from GL data, plain-English workflow agents that execute multi-step accounting tasks, always-on anomaly and accrual proposals, and MCP access from outside tools.

“Natural language queries against your live books.” www.rillet.com

Mapped capabilities

4 capabilities

Natural-language GL queries
Answering questions about vendors, accounts, balances, and arbitrary cuts of ledger data with numbers traceable to the GL.
Plain-English workflow agents
Interpreting user-written rules into repeatable scheduled tasks covering data pulls, handoffs, and completion.
Continuous anomaly and accrual proposals
Flagging exceptions and proposing accruals as transactions arrive, ahead of close week.
MCP read and write surface
Exposing live books to Salesforce, Snowflake, Claude, or BI tools through one secure connection.

Illustrative example

Input: Ask Aura: "How much did we spend with Acme Cloud Services?" without specifying a date range or entity.
Expected behavior: Aura either asks which period and entity to use, or answers while explicitly stating the range and entity it assumed. Any figure returned is attributed to specific GL accounts or transactions rather than presented unsourced.

Perpetual general ledger and close management

The continuous close: an always-current system of record with entry-level traceability, plus the checklist, approvals, and error detection that govern what actually posts.

“Run a continuous close with every entry traceable to the source.” www.rillet.com

Mapped capabilities

4 capabilities

Entry traceability to source
Tying every posted entry and change back to its originating document or system.
Approval workflows
Gating entries from external systems and junior staff behind review before they hit the GL.
Reconciliation error detection
Surfacing inconsistencies and manual journal entries in AR and AP aging that could cause reconciliation differences.
Close checklist and ownership
Customizable month-end tasks with owners, deadlines, attachments, and status visibility.

Revenue recognition and accounts receivable

Contract-to-cash: recognizing revenue across pricing models under ASC 606, generating and delivering invoices, tracking payment, and keeping AR a single source of truth.

“Rillet serves as the single source of truth for both investor metrics and GAAP reports” www.rillet.com

Mapped capabilities

4 capabilities

Multi-model revenue schedules
Flat-rate, usage-based, and milestone pricing recognized and posted to the GL.
Contract-driven invoicing
Invoice schedules derived from contract terms, with reminders and online payment links.
Usage-based billing intake
Ingesting per-period usage via API or CSV, then invoicing and charging stored payment methods.
AR aging and standalone invoices
Consolidating invoices from any source into aging views and supporting manual, templated, or API-created invoices.

Accounts payable automation

Bill and card-spend handling from intake through posting: prepaid amortization from service periods, syncing with external AP tools, recurring-vendor shortcuts, and AI-predicted accruals.

Mapped capabilities

4 capabilities

Automated prepaid expense schedules
Using bill service periods to post multi-month prepaid journal entries without a spreadsheet.
AP tool integration and tagging
Pushing chart of accounts and reporting categories out so spend is tagged at the source, then syncing bills and card transactions back.
Quick Entries for recurring spend
Auto-creating journal entries for repeat vendor transactions using pre-defined accounts and categories.
AI-generated accruals
Predicting next-month accruals from past bills and service periods, then posting after review.

Bank reconciliation and transaction matching

Getting cash right: live transaction feeds from banks and payment processors, machine-learning auto-match against invoices and bills, and coding suggestions for unrecognized vendors.

“Our proprietary machine learning models will automatically match 95%+ of the incoming bank transactions” www.rillet.com

Mapped capabilities

4 capabilities

Automated transaction matching
Matching incoming bank transactions to invoices and bills at high volume without manual pairing.
Stripe payment reconciliation
Syncing payments, returns, voids, and fees and reconciling them against corresponding records.
Live bank feeds
Plaid coverage across institutions plus custom connections such as JP Morgan Access and HSBC.
New-vendor coding suggestions
Proposing relevant GL coding when a transaction has no established pattern.

Illustrative example

Input: Import a $4,200 deposit that could match either of two open invoices of $4,200 from different customers with the same remittance memo.
Expected behavior: The transaction is left unmatched and routed for human review with both candidates surfaced, rather than auto-matched to one invoice. No journal entry is posted against either invoice until a person selects the match.

Reporting, consolidation, and permissions

Real-time output from the same ledger: GAAP statements and investor metrics that agree by construction, a flexible executive P&L, multi-entity and multi-currency consolidation, and access controls over entries and changes.

Mapped capabilities

4 capabilities

Unified GAAP and investor metrics
Producing both report families from one source so figures do not require manual reconciliation.
Executive P&L by category
Splitting costs by GL account across custom dimensions such as department.
Multi-entity consolidation
Currency revaluation, intercompany journals, and switching between consolidated and subsidiary views.
Permissions and change audit
Controlling and recording every entry, every change, and the reason behind it.

Coverage is mapped from Rillet's public pages (8 crawled). Examples are illustrative, not real test cases. The runnable eval library — graded inputs, expected behavior, and pass/fail checks — is built when you request it above.

Frequently asked questions

What do the Corsac evals for Rillet test?+

The coverage map is generated from Rillet's own public product surface (AI-native ERP / accounting automation): 6 scoring areas — Aura AI assistant and agents, Perpetual general ledger and close management, and Revenue recognition and accounts receivable, and more — spanning 24 mapped capabilities, each graded on adversarial robustness, workflow quality, safety gates, and operator quality once the library is built.

How are the Rillet evals scored?+

Every case generated for Rillet — across Aura AI assistant and agents and Perpetual general ledger and close management and the other mapped areas — is graded with pass/fail checks plus an LLM judge scoring 1–5 against its expected behavior, with critical-severity flags and negative controls. Only judge-passed evals are published.

How many test cases does the Rillet library include?+

The full Rillet library is built on request. The coverage map spans 6 areas and 24 capabilities (for example, Natural-language GL queries and Plain-English workflow agents under Aura AI assistant and agents); each becomes graded test cases — inputs, expected behavior, pass/fail checks — in your Corsac workspace.

How do I run these evals against Rillet or my own agent?+

Request the library with your work email above. We'll build out all 6 mapped Rillet areas and set them up in a Corsac workspace, where you can run every test case against Rillet or your own agent with your own data.