Practitioners author
Real asks, real documents, real judgement calls, from people who do the work every day.
RL environments for enterprise work
Real work.Ready to train on.
Lometa turns senior practitioners' judgement into realistic, verified environments, so AI agents learn the job, not the demo.
Writes the task from real work
“Close Q3 accruals. Split anything approved across quarters.”
- Real cases
- Runbooks
- Rubrics
- Edge cases
- Judgement calls
- Stakeholders
- ERP
- GRC
- Procurement
- Directory
- Teams
- Docs
Agent enters a live company
- Accrual split across Q3 and Q4
- Approver confirmed before posting
- C-12 evidence attached
- Memo explains the $48k variance
Why it matters
Agents ace benchmarks. Real work is messier.
Real work runs on incomplete records, unwritten rules and judgement calls. We put that into the environments models train on.
How it works01 02 03
From expert judgement to training signal.
We build the world
Systems that behave like ERP, GRC, procurement and the directory, messy data included.
Verified and calibrated
Reference solutions, hidden expert rubrics, and pass rates set at your model's edge.
Inside every environment
Built like a real company. Graded like a real expert.
Enterprise systems
ERP, GRC, procurement, directory and chat, with real API behaviour.
Real-world mess
Placeholders, look-alikes and records that disagree.
People with facts
Colleagues who answer only what they're asked.
Hidden rubrics
Expert criteria the agent never sees.
Verifiable rewards
Scored on outcomes and system state, not wording.
Your harness
Ships for Harbor, OpenEnv, Inspect and verifiers.
IndustriesWhere experience
Where experience
changes the outcome.
Starting with finance and enterprise operations, then every field where judgement decides the outcome.
01
Finance & enterprise operations
Close, controls, payables, revenue, treasury and audit.
View expertise
02Software engineering
Architecture, reliability and engineering judgement.
View expertise
03Legal & compliance
Contracts, regulation, privacy and compliance work.
View expertise
04Transportation & EV
Fleet operations, charging and deployment decisions.
View expertise
05Marketing & advertising
Campaign operations, measurement and brand.
View expertise
06Safety, emergency & risk
Incident response, risk and continuity.
View expertise
07Built environment & energy
Construction, facilities and energy systems.
View expertise
08Industrial design
Product design, prototyping and manufacturability.
View expertise
For expertsYour judgement,
Your judgement,
trained into AI.
Turn the calls you make every day into the environments that teach AI agents. Paid, flexible, project by project.
Software engineeringFinance & enterprise operationsLegal & complianceTransportation & EVMarketing & advertisingSafety, emergency & riskBuilt environment & energyIndustrial design
For AI teamsTraining agents
Training agents
for real work?
Tell us what your agents need to learn. We will scope environments at your model's edge.
Environment packs Calibration reports Custom builds with practitioners
Talk to Lometa
Tell us about your project and our team will follow up.