Skip to content

A world model,
for people.

Every model predicts something. Ours predict what happens when a person, a team and a company meet.

Language modelReads wordsPredicts the next word
World modelReads the state of the worldPredicts what happens next
People world modelReads a person, a team and a companyPredicts how it goes when they meet

Halloway builds people world models. Orbit is the first.

The difference isn't how clever it is.
It's what it has seen.

Will Hannah do well on our Platform team?
A language model

Hannah's CV matches most of the job description. She looks like a strong candidate.

It has read her CV. It has never seen your team.

A people world model

Yes, for about 18 months. Platform has nobody who scales systems, and she has done it twice. People with her path at companies your size tend to outgrow the role after that.

Her career pathYour team mapSimilar cases in the market

Not enough evidence on how she leads people. Ask in interview.

It learns the world, then your company, then predicts.

1

Learns how work works

Career paths, team changes and company growth across the market. Who does well where, how long they stay and what changes when someone joins or leaves.

2

Learns your company

Your teams, leaders, goals and culture, read from the tools you already run. Private to you, and never used for anyone else.

3

Predicts, and shows why

It finds real cases like yours and what happened in them. Then it gives the answer, the evidence and what it could not see.

Every hire you make teaches it.Outcomes and interview decisions feed back into your private model.

People, teams and companies, in one language.

People

Path, not keywords

What they have built, how fast they grew and the kind of builder they are.

Teams

How work gets done

Who starts, scales, runs, fixes and goes deep, and what is missing.

Companies

Stage and culture

Size, pace, how fast it is changing and how people really work there.

Roles

What the work needs

The problems the hire will own, and what good looks like at 6 and 18 months.

Signals,
never scores.

Orbit reads a career role by role, then adds what it finds into eight plain-English signals. Five describe the person. Three depend on your team, and change as it does.

Separately, it measures how visible someone is to the market and how much evidence there is of their quality. Brand only ever counts towards visibility. Where the two disagree, you have found someone the market mispriced.

TrajectoryHow fast scope and level have grownAbout the person
Pace of learningNew domains picked up, and how quickly they stuckAbout the person
ResilienceWhat they shipped through when it got hardAbout the person
OwnershipDecisions they owned, and the outcomes that followedAbout the person
How they workStarts, scales, runs, fixes or goes deepAbout the person
Trust signalsWho vouches for them, including your own peopleAbout your team
Team pairingWhether they fill a gap your team hasAbout your team
TimingWhether they are likely to be open to a move, and whenAbout your team

Built to be trusted with people.

Private to you

What it learns from you never trains anyone else's model.

Shows its working

Every answer opens to its evidence, and says what it could not see.

Team totals only

Your own people are read as teams. Nobody gets a score.

A person decides

It never rejects, moves or contacts anyone on its own.

How we keep your data safe →

Orbit is the first. More are coming.

Our first people world model, built for the teams that hire. See it read your company in a 30-minute briefing.

Book a briefing

Common questions about people world models

What is a people world model?
A people world model predicts how people, teams and companies interact: who succeeds in a role, what a team is missing, and how hiring changes delivery. It is trained on careers, org context and your company's own outcomes, not text alone.
How does Orbit learn our team?
Orbit connects to your ATS, HRIS and tools your people already use. It maps teams and reporting lines, reads interview notes where you allow it, and learns from hires, rejections and who thrives after they join.
Does Orbit get better over time?
Yes. Every action teaches your private model: hires, interviews, stakeholder briefs and searches. That learning stays in your workspace. It is never pooled into a shared database other companies can use.