What the network is
The Saolabs expert network is a pool of working professionals who have been sourced, interviewed and tested before they are placed on an AI project. It spans medicine, law, finance, engineering, software, science, mathematics and languages.
We built it because a model learns its judgement from the people who label its data. When the question is clinical, legal or numerical, a generalist annotator can mark a fluent answer as correct when an expert would see the error in the second line.
The network exists to close that gap. Every expert is chosen for the field they will judge, and assessed on work that looks like the work they will do.
Who is in it
We recruit people who do the job, not people who describe it. Each industry draws on the professions whose judgement a model needs to learn.
Each industry has its own page explaining why the field matters for AI safety, what its experts do and how they are vetted.
- Medicine: doctors, nurses and pharmacists
- Law: lawyers, paralegals and compliance specialists
- Finance: analysts, accountants and risk specialists
- Engineering: mechanical, electrical and civil engineers
- Software and science: software engineers, architects, security specialists, physicists, chemists and biologists
- Mathematics and languages: mathematicians, statisticians, native speakers, translators and linguists
What experts do on AI projects
Experts produce the data a model learns from and the data it is judged against. That covers labelling images, video, audio and sensor data; writing expert answers, rankings and reasoning for training; evaluating model outputs; red-teaming; and producing alignment data.
The work spans language, code, vision, voice, video, agents, simulation, robotics and the physical world. A pharmacist might rank answers about drug interactions. A security specialist might try to make a coding agent write an exploitable function.
In each case the output is expert-labelled data, delivered to the team training or evaluating the model.
How an expert reaches a project
Every expert passes through the same three stages: sourced, vetted, matched. Sourcing runs through our agentic Hiring OS, outreach agents, a recruiter network and a jobs marketplace.
Vetting combines interviews by AI voice agents, real-world work tests in the expert's field, skills assessments and any other assessment the project needs. Matching places the expert on the right AI project, fast.
The full method is on the vetting page. The short version: we test the work before we trust the work.
Who works with the network
AI model companies and labs come to Saolabs for expert data, evaluations and expert talent. Companies that want to run the same hiring process themselves can license the technology behind it.
Experts join the network through Workor at https://www.workor.ai. From there they are vetted and matched to AI projects in their field.