The first AI admin for every enterprise system.

Every business change ends up as changes to your systems, and each system has its own specialists. Mando’s agents know every application, learn how yours is set up, and make the change across all of them.

Placeholder · demo video Play the demo

§1Thesis

Thesis

Enterprise admins translate business changes into system changes. Today that takes a specialist for every application, each with their own queue and their own way of working.

Thesis. Agents can do this across any enterprise system, with built-in knowledge of each application and the ability to learn each customer’s environment.

Today: one business change goes to three specialists, each changing their own system.

(a)Today. A specialist per system, each with a separate backlog.

With Mando: one business change goes to one set of agents that changes all three systems.

(b)With Mando. One set of agents across every system, sharing what it learns.

§2How it works

From a request to a tested change.

  1. 01 Learn

    Agents read calls, tickets and your admins’ past work, then inspect each system’s screens and backend to see how it is set up today.

  2. 02 Build

    Each agent works in its own browser with code execution. It plans the change and makes it on every screen it touches.

  3. 03 Prove

    Nothing ships until it passes in a sandbox. The agent records the state before and after, and keeps what it learned for the next change.

RunOpen an office in LisbonReady for production
TimeSystemStepStatus
Learn
0:00TicketRead the request: a new office opens in Lisbon
0:02MemoryRecall how the Madrid office was set up last year
0:06HCMInspect locations, pay groups and job profiles
0:09ERPInspect cost centres and approval rules
Build
0:14HCMCreate location Lisbon, Portugal
0:19HCMAdd pay group PT Monthly
0:27ERPCreate cost centre for Lisbon
0:31ERPRoute approvals to the Lisbon lead
0:38ITSMAdd laptop and access bundle for Portugal hires
0:44PayrollSet the Portuguese tax jurisdiction
Prove
0:52SandboxHire a test employee in Lisbon, end to end
1:05AllRecord state before and after every change
4 systems8 changesAll checks passed

§3Research

Open problems.

Screens are dense, state is hidden, and every action lands in a system people are working in. There is no reset button. These are the problems we work on.

P1

Reading screens

OCR and layout models read every frame down to the word, with a vision model for the rest. Dense tables and nested forms have to be read exactly.

Open questionRecovering table structure through merged cells and frozen panes.

P2

Masking sensitive data

Personal and proprietary values are masked on every frame at ingest, before anything is stored or seen by a model.

Open questionCatching company identity hidden in configuration names and free text.

P3

Tracking system state

Enterprise systems allow some changes only from some screens. We rebuild state from pixels and backend queries, and hold it across thousands of actions.

Open questionDetecting drift between what the agent saw and what actually committed.

P4

Remembering the setup

Each team’s history becomes a private record of how its systems were configured and why. The agent reads it before acting and adds to it after.

Open questionInferring why a rule exists when nobody wrote the reason down.

P5

Learning each environment

Agents learn by reinforcement inside each customer’s environment, from every change they build and test.

Open questionCarrying what works in one module to a module the agent has never seen.

P6

Working for hours

One remote browser session, hours long. The agent gathers context through clicks and code, builds the change, then retests it.

Open questionClean rollback after a half-applied change, hours into a task.

P7

Proving a change works

The unit of work is a verified change: the intent, the state before and after, and the runs that validate it.

Open questionScoring a change that is correct but was built a different way.

P8

Yours

If there is a problem in this space you would rather work on, tell us what it is.

talent@mando.work

§4Results

Measured on real systems.

Our agents lead on SCUBA and WorkArena. Public benchmarks cover only a handful of systems, so we build our own for every system our agents run.

BenchmarkFromWhat it measuresTasksSource
SCUBASalesforce AI ResearchAdmin, sales and service work in live Salesforce sandboxes, scored by milestone300arXiv 2509.26506
WorkArena++ServiceNow ResearchCompositional workflows on a live ServiceNow instance: planning, retrieval and memory682arXiv 2403.07718
CRMArena-ProSalesforce AI ResearchSales, service and quoting in a Salesforce org, single and multi turn19arXiv 2505.18878
EntWorldTsinghua UniversityERP, CRM and IT service tasks, each with a verifiable end state1,756arXiv 2601.17722
LedgerNeurIPS 202653 modules of a production ERP, scored on what reached the database424neurips.cc
EnterpriseBenchFujitsu ResearchEngineering, HR, finance and admin tasks behind access controls500arXiv 2510.27287
SOR-Bench OursMandoVerified questions across 31 modules, written with practitioners. The first of one benchmark per system we run.2,214sorbench.com

Help build the first AI admin.

talent@mando.work