Training datasets
Structured operational histories that can support the development of AI systems capable of understanding real-world technical work.
Ground
RUNRGEN
The real world is the training ground.
Decades of work. A new kind of asset. Every service call, installation, repair, migration, and technical decision leaves a record. RUNRGEN turns that history into evidence-grounded data for AI—not another system for managing the work itself.
The problem
Decades of work, scattered through the record.
Across industries, decades of operational knowledge are scattered through work orders, technician notes, diagrams, equipment inventories, emails, and project documentation.
Most of that history was created to get work done—not to teach machines how work gets done.
RUNRGEN changes that.
We transform fragmented operational records into structured, traceable datasets and evaluation benchmarks designed for AI development.
Fragmented records
- Work order
- Technician note
- Diagram
- Equipment inventory
- Project file
Resolved into
A structured, traceable dataset—with the gaps still marked.
What RUNRGEN produces
Built for the next generation of AI.
Evaluation benchmarks
Evidence-grounded scenarios for testing how AI agents interpret documentation, reconcile conflicting records, identify missing information, and reason about operational outcomes.
Domain-specific intelligence
Specialized datasets across technical environments, including telecommunications, networking, healthcare infrastructure, hospitality, and commercial facilities.
Evidence architecture
From operational history to machine intelligence.
Real-world work is rarely clean. Plans change. Equipment fails. Technicians improvise. Documentation conflicts. Jobs are completed without every detail being recorded.
That complexity is exactly what makes operational history valuable. RUNRGEN reconstructs the relationships between what was planned, what was reported, what was verified, and what remains unknown.
Not synthetic scenarios pretending to be reality. Actual operational records, reconstructed with their uncertainty intact.
01
Planned
What the work order, design, or schedule said would happen.
02
Reported
What a technician, installer, or operator said was done.
03
Verified
What an independent check, test, or customer confirmation established.
04
Unresolved
What the record still does not support. Missing evidence stays visible.
Intelligence without inventing certainty.
A work order says a task was scheduled. A technician reports that equipment was configured. A customer confirms that the system is operating. Those are three different kinds of evidence.
RUNRGEN preserves those distinctions.
It doesn't turn a scheduled task into completed work, a reported result into an independently verified outcome, or missing documentation into an invented fact.
For AI systems operating in the physical world, knowing what isn't established can be as important as knowing what is.
Every derived assertion maintains a relationship to its supporting source. Reported activity is distinguished from verified outcomes. Missing evidence remains visible.
Commercial opportunity
A new revenue stream for operational businesses.
Organizations accumulate valuable technical histories while running their businesses. RUNRGEN is designed to help transform eligible records into licensable data assets—subject to ownership, contractual rights, confidentiality, privacy, and quality requirements.
The original business retains its operational focus. Its accumulated knowledge gains a potential second life supporting AI development. Datasets can be organized and licensed by domain, platform, application, or defined collection.
Source
Operational businesses
Historical records created to get the work done.
158 West
RUNRGEN
Structured, traceable datasets and benchmarks.
Use
AI developers
Training and evaluation grounded in real operations.
Built from the field.
RUNRGEN's development is informed by real technical operations: structured cabling, network migrations, nurse-call systems, communications infrastructure, and the documentation challenges that accompany them.
The objective isn't simply to digitize paperwork. It's to preserve the operational knowledge inside the record—without losing the difference between what happened, what someone said happened, and what can actually be established.
Next step
Turn operational history into AI infrastructure.
For operational businesses: explore the potential value of your historical records.
For AI developers: explore domain-specific datasets and evaluation benchmarks grounded in real-world operations.
RUNRGEN is a developing 158 West product. Dataset availability and licensing are subject to qualification and agreement.