Wave 1 segment
Real-world capture for teams training models that need grounded spatial understanding.
Roam helps world-model, multimodal video, and physical-AI teams procure real-world environments with clear formats, manifests, QA expectations, and review materials before a live program starts.
Likely buyer teams
What teams usually need to confirm
- Manifest structure, checksums, and delivery packet examples
- Open-format outputs tied to downstream training and evaluation workflows
- Clear line between public demo proof and NDA-only production review
First pains to address
Data diversity is the bottleneck, not model ambition.
Synthetic pipelines still need ground-truth calibration and evaluation data.
Procurement needs proof that the output contract is real before mobilization.
Example program shapes
Warehouse and industrial interiors for model post-training
Campus and retail environments for world-model evaluation
Indoor and outdoor geometry-rich corpora for multimodal grounding
Deliverable fit
Output families that fit this workflow.
Raw sensor data
Camera, LiDAR, calibration, pose, and trajectory data for teams that need first-order inputs rather than post-processed visuals alone.
- Authoritative for
- Training corpora, reprojection, calibration review, and downstream reconstruction pipelines.
- Preview use
- Subset validation, pose sanity checks, and pipeline bring-up.
Point clouds
Structured geometry for measurement, mapping, simulation prep, and change review.
- Authoritative for
- Measurement-adjacent review, geometry QA, GIS intake, and reconstruction baselines.
- Preview use
- Subset exploration, coverage checks, and visual inspection.
Gaussian splats
Neural-representation deliverables for fast scene review and model-facing visualization workflows.
- Authoritative for
- High-fidelity visual scene review and neural rendering research workflows.
- Preview use
- Web previews and human review.
OpenUSD scenes
Scene-graph packaging for simulation, twin ingestion, and composition-aware workflows.
- Authoritative for
- Simulation and scene composition where OpenUSD is the agreed delivery container.
- Preview use
- Scene inspection, layer review, and asset organization checks.
Materials you can review now
Sample manifest
Example manifest structure showing delivery versioning, formats, checksums, and CRS metadata.
Why review it: Roam can describe delivery structure in a machine-readable way before a live program starts.
Boundary: This is an illustrative public example, not a customer-specific delivery commitment.
Checksum example
Example checksum file showing artifact validation format.
Why review it: Roam can expose integrity checks as part of the delivery packet.
Boundary: Checksums shown here are example values for public proof only.
Sample dataset catalog
Live demo datasets, manifests, and gated artifact downloads.
Why review it: The public demo routes are real and runnable today.
Boundary: Demo subsets are not equivalent to a production program or customer packet.
Sandbox and readiness APIs
Short-lived demo keys, sample dataset listing, and readiness scoring.
Why review it: Technical evaluators can verify endpoint behavior and sample shapes now.
Boundary: Not a production API commitment.
Likely objections
We do not need another scanning vendor.
Roam is positioned around delivery packets, manifests, QA expectations, and handoff structure, not a generic capture gallery.
We need training fuel, not a one-off demo.
Roam is built for repeatable dataset programs and environment refreshes, not one-off demo scans. Production scale is scoped directly with the team.
Ways to engage
Technical evaluation
Start with docs, sample data, format references, and the sandbox before opening a live program.
Audience: ML engineers, simulation leads, GIS analysts, platform teams
Available now: Public and available now through developer docs, sandbox APIs, and sample-data examples.
Validate technical fitLive program scoping
Scope environments, outputs, review constraints, and timeline through an intent-first intake path.
Audience: Founders, VPs, robotics PMs, autonomy leads, operations buyers
Available now: Public intake is available now; commercial and operational specifics are finalized in the engagement.
Start program scopingSecurity, privacy, and legal diligence
Bring data handling, deletion, privacy-sensitive capture, and diligence questions in early.
Audience: Security reviewers, privacy counsel, legal, procurement
Available now: Public diligence materials are available now; customer-specific review expands under NDA when needed.
Start diligence reviewFrom evaluation to scoping
Review sample delivery proof
Inspect sample artifacts, manifests, and gated downloads before starting a live conversation.
Raise NDA, security, privacy, and delivery constraints in the first conversation instead of after procurement restarts the process.
Need immediate scheduling? Use the programs inbox.