Sorting and Categorization
- Categorization
- Fine manipulation
- Repeated actions
EGXO
Human Behavior,
Captured for Machines That Act.
Custom human demonstration data from egocentric, exocentric, or synchronized viewpoints—designed around the tasks, environments, and signals your model needs.
Start With What Your Model Needs to Learn.
Prove It in a Pilot.
Then Scale.
Tell us which behavior, environment, or failure case your model needs to understand. We’ll design a focused pilot to test the right capture approach before expanding collection.
Custom data programs
Start With Real Samples. Build Around Your Model.
Review six rights-approved egocentric task videos to evaluate viewpoint, task complexity, and capture quality.
Need different tasks, environments, contributor profiles, exocentric views, synchronized capture, metadata, or annotations? We’ll design a focused pilot around your model, delivery format, quality standards, and licensing needs.
How it works
Review Samples. Run a Focused Pilot. Scale With Confidence.
Start with real task footage, then test a custom collection approach against your technical and commercial requirements.
- 01 / Review
See Real Task Footage
Review six first-person previews showing hand-object interaction across varied everyday tasks.
View Sample Data ↗ - 02 / Define
Define the Data You Need
Choose the tasks, environments, viewpoints, contributor criteria, data format, rights, and success measures.
- 03 / Scale
Expand What Works
Use the pilot results to refine the collection approach before committing to a larger program.
Tell us what your model needs to learn and where the current data falls short.
Discuss a PilotContributor access
Reach Real Contributors Across Two Priority Markets.
EGXO Data is the dedicated B2B data brand for the GIG Rewards contributor network. The network publishes contributor and telco-subscriber reach across the Philippines and South Africa.
- Published contributors
- 800K+
- Telco-subscriber reach
- 120M+
- Priority markets
- 2
- Philippines + South Africa
These are GIG Rewards network figures—not a promise that every subscriber will participate, qualify, consent, or be available for a specific project. Pilot capacity, targeting, throughput, and turnaround are confirmed against the actual brief.
Choose the Viewpoint That Captures the Right Signal
Use egocentric data for actor-aligned detail, exocentric data for scene context, and synchronized views when the model needs both.
Turn Your Use Case Into a Clear Data Brief
Before collection starts, agree on what to capture, how it will be delivered, and how you will judge whether it works.
- Model objective
What the model needs to observe, predict, or execute—and the result you want to improve.
- Coverage
Tasks, environments, objects, people, viewpoints, variations, and hard cases.
- Data requirements
Viewpoints, modalities, metadata, labels, file structure, privacy, and licensing.
- Success criteria
The measurable quality, coverage, and ingest checks the pilot must pass.
A Pilot Should Answer One Question: Is This Data Worth Scaling?
Test the capture setup, task design, data quality, rights, and delivery format against the buyer’s actual use case.
- 01
Scope
Agree on the task, environment, viewpoint, contributor criteria, delivery format, and success measures.
- 02
Capture
Collect a representative sample that includes the conditions most likely to challenge the model.
- 03
Validate
Check task visibility, file quality, metadata, labels, privacy, rights, and buyer ingest.
- 04
Decide
Review the results, refine the approach, and decide whether to expand collection.
Sample library
View Real Egocentric Task Footage.
Explore six first-person previews covering assembly, cleaning, deformable objects, packing, sorting, and tool use.
Object Cleaning and Dishwashing
Continuous handling, washing, rotation, and inspection of household objects in a natural environment.
- Object rotation
- Contact-rich actions
- Continuous handling
Quality and governance
Good Data Comes With Proof.
Every delivery should make it easy to verify how the data was captured, reviewed, licensed, versioned, and prepared for your pipeline.
See Our Quality Approach ↗Start With the Technical Decision
All Resources ↗Egocentric vs. Exocentric Data: Choosing the Right Viewpoint
A buyer’s guide to first-person, third-person, and synchronized capture for robotics and physical AI.
Read Guide ↗Egocentric Video for Robot Learning
How first-person demonstrations can support manipulation, VLA systems, navigation, and long-horizon task learning.
Read Guide ↗Public vs. Custom Egocentric Datasets
A procurement framework for deciding when public benchmarks are enough and when custom collection is justified.
Read Guide ↗Start a conversation
Tell Us What Your Model Needs to Learn.
Share the task, environment, viewpoint, format, timeline, and success criteria for your data program.
Start a Project Brief