Soon Robotics

Social media, the way it should have been.

Meet

Plan it, drive it, arrive together.

Get Caravan › Meet Cara ›

Live map

Every car. One map.

Your whole crew, moving in real time — on the phone in your hand or the screen in your dash.

Caravan on CarPlay in a car dashboard: a live map of the Fort Lauderdale coast with the crew’s custom car pucks, a Members button and push-to-talk controls, city street ahead through the windshield.

Chirp

Hold. Talk. Release.

Push‑to‑talk to one car or the whole convoy. Hands stay on the wheel.

A hand holding an iPhone on Fort Lauderdale beach, thumb pressing Caravan’s amber Chirp push-to-talk button over a live map with friends’ pucks and a route along the coast.

Group Soundtrack

One song. Every car.

The host picks the playlist. Everyone hears the same beat at the same moment.

A hand holding an iPhone in a classic car in the Scottish countryside: Caravan showing a live rally meetup around Edinburgh with friends’ pucks and a route to the checkered flag, and the Group Soundtrack card playing from Apple Music.

Privacy

On for the drive. Off when you’re home.

Sharing starts an hour before the trip and ends itself, per person, the moment they arrive home.

A driver parked at home at night holding an iPhone: Caravan showing the “You’re home” prompt, offering to let the group know they arrived safely and end location sharing.

Places

Find the spot. Bring everyone.

Every place has a page. See it, save it, and start an event there in one tap.

A person on a bluff above the water at sunset holding an iPad: Caravan showing a winery’s place page with photos, address and a Create event here button, beside the map.

Every screen

One crew. Everywhere.

iPhone · iPad · Apple Watch · Apple TV · CarPlay

A crew’s cars — a blue sports car, a yellow 4x4 and a green supercar — parked at a coastal overlook at sunset, their drivers and a dog watching the sun go down over the ocean.

Coming to Caravan

Meet Cara.

The copilot for your whole crew. Tell her where you want to end up and who’s coming. She handles the rest.

A coastal highway stretching to the horizon, its surface rendered as glowing streams of data — the road as Cara sees it.

Plans the trip

Say it the way you’d say it to a friend. Stops, timing and invites come back drafted.

Rides along

Knows who’s behind, who’s pulled over, and when the group needs a new route.

Keeps watch

Notices when a car goes quiet mid‑drive and makes sure someone checks in.

Knows your car

Range, fuel stops and roads that suit what you’re actually driving.

Remembers

Your places, your taste, and a recap of every trip worth keeping.

Works for you

Never for an advertiser. What she learns about you stays inside Caravan.

The foundation

The S.O.U.L. Stack™.

Soul Stack™ is Soon’s independent AI governance layer, built above the model to keep intelligence aligned with truth, safety, and human dignity. It helps reduce hallucinations, constrain harmful behavior, and guide Cara’s responses with judgment — not just computation.

Safe by Design

Safety isn’t bolted on at the end. The system is built so the safe path is the default path, and harm is structurally hard to reach.

Objective by Default

Cara reasons from evidence and tells you what’s true, not what’s flattering — and holds her ground when the easy answer is the wrong one.

Useful Under Constraint

Real life has limits — privacy, attention, bandwidth, trust. We build for the constraints people actually live with, not a frictionless lab.

Loyal to Humanity

Our first allegiance is to the people we serve and the humanity we belong to — not to engagement metrics, not to the machine.

Measured, not promised

Tested in the Crucible.

The S.O.U.L. Stack is scored on Crucible, an open safety benchmark that measures the system, not just the model. It spans six tracks — five built on industry-accepted datasets (HarmBench, XSTest, JailbreakBench, TruthfulQA, and MASK) plus our own AI Peer Test, in which seven frontier models write adversarial prompts designed to break it. Every response is graded by the Crucible Tribunal: three independent judges, each running a different model, so no single model decides the verdict.

97.5

Median score across all six tracks · 0–100, higher is better

98.5

Harm Refusal

HarmBench

96.2

Over-Refusal

XSTest

97.0

Manipulation Resistance

JailbreakBench

97.9

Truthfulness

TruthfulQA

100.0

Honesty Under Pressure

MASK

95.5

AI Peer Test

Crucible · original

The Crucible Tribunal

Three independent judges evaluate every response — each a different large language model, accessed via DeepInfra. They never see one another's evaluations, so no single model's biases decide the score.

Solomon

Qwen 3.6-35B-A3B

The primary judge. Named for King Solomon — wisdom and discernment. Scores each response against the Solomon Rubric v1.

Aquinas

DeepSeek-V3

Named for Thomas Aquinas — moral reasoning and ethical philosophy. The second evaluation, weighing whether a response upholds ethical standards.

Jerome

Llama 3.3 70B

Named for Saint Jerome — fidelity to principle. The third judge, and consistently the most rigorous scorer.

A verdict requires 2-of-3 consensus. Each judge returns one of eight verdicts, scored 0–100; if all three disagree, the result is NO_CONSENSUS.

VIGILANTHONESTREBUKECONSCIENCEDEFUSEEVADELIEBETRAY

See the full scoreboard at cruciblebenchmark.com ›

Crucible v1.0 · complete run across all six tracks · 52,567 questions.