The contributors come from the US occupations with the largest share of labor costs in digital work. Each one turns a piece of their own job into an environment. Computer Work Val v1.0 is 300 of those tasks, chosen to represent every domain, and the environments can be used for both testing and training.
Results #
How the environments are created #
We build each environment from high-fidelity clones of tools like Jira, Google Sheets and Salesforce, seeded with mock data from the person who actually does the task.
Simulators used
The environments run on 210 simulated apps, from Gmail and Excel to open-source ERPs, health records and helpdesks.
- Gmail
- Google Sheets
- Slack
- Word
- Excel
- PowerPoint
- Salesforce
- Google Drive
- Jira
- DocuSign
- HubSpot
- Shopify
- Google Docs
- Canva
- Zendesk
- Stripe
- Figma
- Power BI
- Trello
- Google Meet
- Freshdesk
- Amazon
- BambooHR
- Fathom
- Airtable
- 7shifts
- TurboTax
- DoorDash
- Airbnb
- Freshdesk (MCP)
- Xero
- Booking.com
- Jira Service Management
- Zoho Inventory
- Notion
- Splunk
- Zoho CRM
- PagerDuty
- UpKeep
- craigslist
Show all 210 #
- Grist
- BookStack
- Mattermost
- Docspell
- Nextcloud
- Metabase
- ERPNext
- WordPress
- Baserow
- WooCommerce
- OpenEMR
- Plausible
- Odoo eLearning
- ONLYOFFICE
- Znuny
- Bahmni
- Worklenz
- GLPI
- Superset
- Akaunting
- Bugsink
- osTicket
- Part-DB
- GitLab
- Plane
- listmonk
- Probo
- Odoo
- Odoo Manufacturing
- Grafana
- EspoCRM
- Forgejo
- Formbricks
- YAFFA
- Redmine
- CloudBeaver
- OpenCart
- eLabFTW
- Odoo Maintenance
- Datashare
- Odoo Invoicing
- Horilla
- Kanboard
- NetBox
- authentik
- CKAN
- Snipe-IT
- Dittofeed
- Frappe HR
- Dolibarr
- Docs
- Odoo Project
- CourtListener
- Memos
- GlitchTip
- Uptime Kuma
- ownCloud
- Odoo Purchase
- Healthchecks
- Odoo Inventory
- Documize
- Documenso
- Bigcapital
- Keycloak
- DocuSeal
- Easy!Appointments
- FreeCAD desktop
- Moodle
- Airflow
- Airbyte
- OrangeHRM
- GRR
- Answer
- Cascadia PLM
- FreeScout
- Bitcart
- Nextcloud Office
- doccano
- Redash
- Firefly III
- Mifos
- EveryDocs
- openIMIS
- Lychee
- Axelor
- Ghostfolio
- Odoo Expenses
- LocalStack
- bewCloud
- tillywork
- Invoice Ninja
- Odoo Employees
- CARE
- b1gMail
- Odoo Sales
- Econ market
- Bytebase
- docassemble
- GitHub
- BookLore
- Tumble
- n8n
- Kimai
- Chatwoot
- Wiki.js
- NocoBase
- InvoiceShelf
- VICIdial
- Odoo Email Marketing
- DAViCal
- Hyperswitch
- iDempiere
- Leantime
- Mindustry
- Twenty
- Zulip
- Bots EDI
- Hi.Events
- Kitchen 3D
- IceHrm
- alf.io
- InvoicePlane
- Penpot
- Docmost
- Flare
- Crater
- Bytepack
- Floor 87
- Lemmy
- Seafile
- Parallax
- Automatisch
- Vexa
- Odoo CRM
- Frappe Insights
- OpenAlgo
- KoboToolbox
- Gridfall
- Jeedom
- OLDP
- 2FAuth
- ArchivesSpace
- Cabot
- Halo
- DreamFactory
- Odoo Events
- Accent
- µTask
- TYPO3
- Blender
- Kestra
- AFFiNE
- 4ga Boards
- vutuv
- Odoo Time Off
- Ever Gauzy
- CodiMD
- Aleph
- Ballerine
- Databunker
- Peppermint
- SuiteCRM
- Cachet
- Etherpad
- Admidio
- ChiefOnboarding
- Aptabase
- Drupal
- Krayin
- Coder
How we select people #
Every task contributor goes through a rigorous pipeline to ensure they meet our quality bar.
- 1#### Initial screenWorks at a relevant organization · uses AI heavily in their job · 3+ years in their domain
- 2#### InterviewsA communication skills assessment and a domain expertise assessment
- 3#### Task creation trainingAn introduction to RL task creation and how to create high-quality tasks, then Plato platform training
- 4#### Work trialA three-hour work trial
- 5#### Final human reviewDomain experts and a Plato RL task specialist check accuracy, relevance, task quality and fit with RL objectives
Who the tasks come from
The 300 tasks span 10 domains and 36 roles.
How we ensure world quality #
We check every task on four dimensions to make sure it reflects real work, poses a meaningful challenge and is worth getting right.
Realistic
The environment and prompt match the author’s real job.
- The software they actually use
- Anonymized data, in the places it would normally sit
- A short prompt, the way work actually arrives
In the Juniper Bean example The accountant’s own stack: Excel, Gmail, Stripe and Mercury. Last quarter’s P&L sits in an email thread, where it normally would.
Fair
Everything needed to finish the task is in the environment.
- Information is where someone in the role would look
- Conflicting data has a clear source of truth
- Each grading criterion is objective and tied to what the task requires
In the Juniper Bean example Stripe shows every CAD payment next to its USD-converted amount, so the right revenue figure is findable.
Difficult
The work takes reasoning and several steps, not a lookup.
- Reconciling information across sources
- Working through conflicting records
- Difficulty from the work itself, never from missing information
In the Juniper Bean example Revenue spans USD and CAD. Summing the Stripe export at face value overstates it by $3,180.
Severe
A wrong answer has real consequences.
- Financial or legal impact, or harm outside the team
- Hard to reverse or likely to go unnoticed
- Actions taken without the required approval
In the Juniper Bean example The P&L goes straight to Elena, the CFO. An inflated revenue line reaches leadership unnoticed.
Reviews: two experts from the same occupation agree on each task’s realism and fairness, and the Plato team reviews it for environment defects, reward hacking and prompt/grading mismatches.
Want access? #
Partner with us to get access to any of these environments.
Work with us #
We’re hiring! Check out open roles or email hi@plato.so.