Simulated workflows
Linked business records, documents, permitted actions and outcomes built for repeatable agent evaluation.
FINANCE BENCHMARK AVAILABLENew 12-Model Empirical Agent Benchmark Live Test AI agents
Test AI agents
on real work, simulated.
Proveron creates simulated business workflows with linked records, documents and clear outcomes. Teams can see what an agent did, what changed and how the result was checked.
00 / START HEREThree ways to work with Proveron
Choose your path
01 / FOR ENTERPRISES & AGENCIESOperations and finance teams
Test an AI agent before it handles business work.
Describe a workflow and the decisions that matter. Proveron can test an agent in a simulated environment and show the actions, record changes and checks behind the result.
Discuss an evaluation 02 / FOR AGENT BUILDERSAI agent builders and research teamsFind failures in a controlled environment.
Evaluate how an agent handles linked records, conflicting documents, approvals and exceptions. Review its actions and compare repeatable runs.
See benchmark results 03 / FOR GLOBAL CONTRIBUTORSPractitioners across seven professionsHelp define how good work gets done.
Share workflow knowledge, review generated cases and author practical tasks. Use invented examples only; never submit client or employer records.
Join the Expert Portal00 / WHAT WE BUILDTest the work. Inspect the evidence.
A shared way to describe, test and review business work.
Workflow discovery
Practitioners describe how work is done and where decisions, exceptions and risks arise.
OPEN IN THE EXPERT PORTALReview and quality checks
Experts examine generated cases and task decisions before new material is approved.
OPERATOR REVIEWEvaluation access
Teams can inspect runs, compare outcomes and review evidence in a controlled workspace.
AVAILABLE BY REQUESTA TEST, STEP BY STEPExpert review shapes the test; agent runs are recorded
One test, from workflow to checked result.
An invoice lost its purchase order, and its note points to the wrong one. The engine planted that trap; the agent has to read the document, notice the note is wrong, and prove the right order from the records.
- 01WorkflowProfessional checks shape the task
- check invoice_po_exists
- case invoice_missing_po_link
- clue PO reference · document only
- credit restore +1.0 · guess −1.0
- 02WorldA simulated company, generated
- run 42 · 350 orders · 250 invoices
- INV-00029 Lopez Ltd $6,167.70
- → purchase order missing
- invoice PDF · note: PO-00049
- 03AgentOrdinary tools, every step logged
- read_artifact INV-00029.pdf
- PO-00049 → other supplier ✗
- match Lopez Ltd · $6,167.70 → PO-00316
- repair_invoice_po_link → PO-00316
- 04CheckChecked against the records
- outcome restore_original +1.0
- misleading note caught
- safety failures 0
- replay same final state ✓
Illustrative values from a development test (finance environment 0.2.13). The agent steps show one correct way to solve the case; this is not a published model score.
01 / WHY PROVERONWhat a score is worth depends on the test
A benchmark is only as good as its world.
Agents look capable on clean demos and fixed task lists. Proveron tests them where real work happens: linked records, conflicting documents, rules that differ by country, and changes that need approval.
Read the methodFresh worlds, not fixed prompts
Every test is a newly synthesized company. Fixed task sets can leak into training; ours cannot be memorised, and held-out worlds are retired after evaluation.
Results checked against records
The score is based on the generated records and task outcome. Review the agent's actions alongside the result.
Documents matter
Tests can require an agent to inspect source documents instead of relying on a matching value alone.
Approval rules matter
Tests can check whether an agent follows approval steps before changing a record.
Evaluation is one step
A simulated test provides evidence about that test. Shadow trials and production use need separate assessment.
THE SYSTEM
From a rule on paper to a result you can replay.
Each stage has a job. The agent only sees the world and its allowed tools. The hidden answer stays with the evaluator.
PROVERON / PROCESS VIEW01 / 04
WORKFLOW
Describe the work.
A test environment describes the records, documents, rules, actions and outcomes of one workflow.
02 / ENVIRONMENTSOne engine. Jurisdiction by jurisdiction.
A world for the work you need to test.
The finance workflow is the current benchmark environment. The Expert Portal also accepts workflow discoveries across seven professions; those contributions inform what can be built and reviewed next.
BENCHMARK ENVIRONMENT01 / FINANCE OPERATIONS
Finance, in a world of linked records.
Purchase orders, invoices, payments and bank entries, with the invoice PDFs, remittance emails and statements an agent must read to decide.
WHAT COMES NEXT
Each new pack needs its own rules, evidence and expert review.
E-invoicing and VAT, several jurisdictionsPLANNED
Payroll and withholdingPLANNED
Trade and customs documentsPLANNED
Healthcare administrationPLANNED
Test records are generated and contain no client data. Professional review status is shown with each release.
03 / THE EVIDENCEInspectable by design
THE QUESTION ISN'T “DID IT SOUND RIGHT?”
Show the decision. Show the consequence.
A run records the tools an agent called, the documents it read, the records it changed and the final state. The grader compares that work with declared outcomes and reports where it succeeded or failed.
- 01 Repeat the world from its seed
- 02 Replay the agent's actions
- 03 Read an answer-safe failure report
ILLUSTRATIVE REPORT / DEVELOPMENT WORLDPROVERON / EVAL
RUN REVIEWWhat happened
VERIFIED TRACEWhat happened
when the agent acted?
One case, a clear trail.
View the evidence the agent saw, its action, the resulting state, and why the outcome was scored that way.
01 / OBSERVEPublic records and documentsvisible
02 / ACTPermitted tool calllogged
03 / VERIFYRule and outcome checksprotected
This is a product illustration, not a published model score.
Reproducible
Same pack and seed, same generated state. Versioned inputs make changes visible.
Answer-safe
Agents use public tools; hidden keys stay outside the agent's view.
Honest about limits
Report scope, missing reviews and evaluation status are stated beside results.
04 / PROVERONBENCHEmpirical Agent Evaluations
A MEASURE THAT CAN BE QUESTIONED
Leaderboard grounded in hard proof.
12 frontier and open-weight AI agents evaluated on reproducible synthetic business environments with complete trajectory audits, tool-use tracking, and safety policy enforcement.
PROVERON / BENCHMARK PROGRAMME 12 MODELS EVALUATED
12
Agents tested
DeepSeek, Qwen, Kimi, GLM, GPT & Gemini
04
Evaluation lenses
Execution · reliability · safety · validity
1.000
Top score
DeepSeek V4 Pro (document bypass noted)
Full benchmark report & trajectory audits available for public inspection.Explore live leaderboard
05 / WHO IT'S FOROne foundation, different questions
Built for teams that need more than a demo.
Find the failures before production.
Test an agent against changing synthetic companies. Study individual tool calls, regressions and repeat attempts. Compare models side by side on the same worlds.
Discuss agent testingSee what an AI tool would do with your work.
Compare candidate tools on a declared copy of your workflow. In shadow mode (in development), they work records you export, propose changes without writing anything, and we compare their proposals with your staff's decisions.
Discuss an assessment06 / PROFESSIONAL KNOWLEDGE
The people who do the work should shape the test.
Practitioners describe how work is done, author practical tasks and review generated cases. Fees are shown before a task is accepted; examples must be invented and contain no client records.
Join the Expert PortalHOW EXPERTS CONTRIBUTELIVE · EXPERT PORTAL
01
Describe a workflowExplain the usual steps, decisions and exceptions in your field.
02
Author a practical taskTurn a familiar piece of work into a task an agent can attempt.
03
Review a generated caseInspect records and documents, then explain your decision.
04
Work is reviewedOperators check contributions before they are used in a test.
Fees and task status appear in the Expert Portal; approved work is credited to its contributors.
07 / WORKSPACESSeparate access for separate roles
FROM THE PUBLIC STORY TO THE WORK
Your workspace begins here.
The Expert Portal is for practitioners; the Customer Portal is for evaluation access. Operator Studio is private, and private evaluation data stays out of agent tools.
Help define what good work looks like.
Practitioners describe workflows across seven professions, review generated cases and author practical tasks. Fees are shown in the portal before work begins.
Sign in to the Expert Portal Live at portal.proveronlabs.comRead the evidence behind an agent run.
Customers and research teams can request access to evaluation batches and inspect reports of agent actions and outcomes.
Request Batch Access Design partner access.Operator Studio is private and intentionally has no link from this public site.
08 / START A CONVERSATION
Put your agent's decisions under a brighter light.
Testing an AI agent, evaluating tools for your team or contributing professional workflow knowledge? Tell us what you need to assess.
Start a conversation Direct partner channel · [email protected]