THOR Data overview

THOR Data is building a marketplace for commissioning the datasets models need, starting with expert robotics annotation. Buyers specify the work; qualified contributors produce it; THOR coordinates verification and commercial delivery.

THOR has signed contracts for robotics video annotation. Demand comes first: customer briefs define the work, the treasury reserves worker rewards and accepted deliveries generate customer receipts. Receipts allocated to buybacks fund purchases of SN112 alpha under the confirmed policy.

Demand-first token economics explains that model and why THOR starts with a private pilot.

Why work through THOR

Commissioning robotics data means finding capable suppliers, explaining the task and checking the result. THOR aims to coordinate those responsibilities through one accountable service, so a buyer can draw on specialist contributors without managing every supplier relationship.

Participant	What they need	THOR's approach
Buyers	What they needDependable output without managing every supplier	THOROne service coordinating the brief, suppliers and delivery
Contributors	What they needClear expectations before work starts	THORAgreed task scope, acceptance criteria and commercial terms
Network	What they needFoundations that can be reused across projects	THORSpecifications, tooling and evaluation knowledge that improve with use

Make dependable delivery easier to repeat.

For contributors, a defined brief makes the skills, effort and commercial terms clear before work begins. For the network, shared specifications and reusable verification methods can make each subsequent commission easier to deliver.

The private pilot

The THOR orderbook connects customer contracts with funded annotation work and explains the economics for buyers, suppliers and THOR.

The buyer and THOR agree the brief, budget, acceptance criteria and delivery terms. Qualified, whitelisted firms contribute across project batches, and accepted output forms the buyer's delivery. Annotation opens to permissionless participation after this private beta, within the Data stage.

1. Demand

   Buyer agrees a spec + budget

   Acceptance and delivery terms agreed before work

2. THOR coordination

   Turns the brief into defined annotation tasks

   Assign work · coordinate checks · organise delivery

3. Beta supply
   • Qualified professional firms
   • Whitelisted during beta
   • Permissionless annotation after beta
4. Verification

   Check output against the agreed spec

   Methods defined for each contract

Accepted output enters commercial delivery

Accepted output enters commercial delivery

From brief to delivery explains the workflow. Quality and verification explains how output is assessed. Participation explains the beta whitelist, the application route and permissionless annotation after beta.

From annotation to commissioned datasets

The broader ambition is to let developers and AI agents request complete training datasets. Commissions could include annotations on existing material, newly collected demonstrations, expert-created examples or synthetic data with human review. Each request would specify format, provenance, usage rights and acceptance criteria.

THOR's dataset market supports its wider open robotics mission. Robotics supplies the starting problems; commissioning and verification methods could also serve other model-training needs. Within THOR Group, data, research, simulation and physical systems remain connected through the Data → Models → Hardware direction.

In the longer term, an authorised agent could identify a model weakness, commission data within an approved budget and scope, pay contributors in crypto and evaluate the result in a training cycle. Human expertise and physical access remain essential inputs. Dataset delivery is one step in that loop; evaluation establishes whether the model improves.

Expansion into these services follows demonstrated demand and dependable verification.