CRYPTIK

[ DEMO · LIVE SIMULATION ]

Position and atomic time, onboard.

Accurate when GPS and ground contact are not.

Two satellites share one orbit. At T+10 min GNSS goes dark. One keeps time with a chip-scale atomic clock, the other with CRYPTIK. Each bubble is how sure it is of where it is.

The numbers Fly it

[ PROBLEM ]

GNSS is a single point of failure.

Jamming and spoofing are routine near conflict zones, and they reach orbit. Ground contact comes in passes of a few minutes. In between, a satellite navigates on whatever its clock remembers.

[ CLOCK ]

Every nanosecond is 30 cm.

Ranging to beacons turns time into distance, so clock drift becomes position error. After 4000 s without GNSS a chip-scale atomic clock is off by about 40 ns: 12 m of range. CRYPTIK’s optical reference is off by 0.68 ns: 20 cm.

MVP time error @ 4000 s0.68 ns0.68 ns ✓
Range uncertainty @ 4000 s≈ 20 cm20.4 cm ✓
CSAC drift @ 4000 s≈ 40 ns40.0 ns ✓
Less drift than a CSAC≈ 60×58.8× ✓

Left: cryptik.space. Right: recomputed by this sim from the stability spec.

[ CONJUNCTION ]

Then something is in the way.

A conjunction is cleared by miss distance: the hard body plus a multiple of your own uncertainty. A bigger bubble means a bigger burn. Same debris, same rule. The CSAC satellite burns about 22 mm/s; CRYPTIK about 3.

Illustrative: 20 m hard body + 5σ at closest approach, 3 h warning, along-track burn (Clohessy–Wiltshire).

[ ORBITAI ]

OrbitAI flies it. So can your LLM.

OrbitAI plans the burn on the satellite’s own estimate. Take over: burn by hand, or copy the briefing into any chatbot and paste its answer back. Same physics, same log, and the propellant counter doesn’t lie.

[ LIMITS ]

What a clock doesn’t fix.

With no measurements at all, the orbit itself drifts: density forecasts are 10–30 % off, and a drag error grows along-track whatever the clock. A clock that holds 20 cm keeps every range clean. Turning those ranges into a tight orbit is OrbitAI’s job.

[ TRY IT ]

The error model behind this page, headless. Change any number, rerun, sweep, download the data.

Open the full terminal ↗

Accurate when GPS is not.

anurag@cryptik.space
[ GNSS LOST ] T+00:00:00
CRYPTIK MVP ±0.0 mm CSAC ±0.0 mm

Live simulation by Veenie. Clock figures from cryptik.space; the mission, debris and screening rule are illustrative.

Talk to Veenie ↗

About the CRYPTIK simulation

Satellite navigation without GNSS, simulated

CRYPTIK MVP stability
1.7 × 10⁻¹³ at τ = 4000 s
Time error after 4000 s
0.68 ns ≈ 20 cm of range
CSAC drift over 4000 s
≈ 40 ns ≈ 12 m of range
Avoidance Δv (5σ rule)
CRYPTIK 2.9 mm/s · CSAC 22.4 mm/s (7.7×)
Orbit
550 km sun-synchronous (illustrative)

A live demo of what an onboard optical atomic clock is worth: two satellites on one orbit lose GNSS; one ranges on a chip-scale atomic clock, the other on CRYPTIK’s optical reference. A debris conjunction follows, and the avoidance burn is flown by OrbitAI — or by any LLM you paste the briefing into.

Questions

Are these CRYPTIK’s real numbers?+

The clock stabilities (1.7 × 10⁻¹³ at 4000 s, roadmap ~1 × 10⁻¹⁴) and the CSAC comparison are from cryptik.space and are reproduced exactly. The orbit, debris, screening rule and propagation model are illustrative.

Does a better clock give a better position by itself?+

No. It keeps timing error out of every range measurement, so the clock bias stops costing a measurement. Position still comes from measurements and an estimator.

Can an LLM really fly the burn?+

Yes, within an allowlist: copy the briefing into any chatbot, paste its JSON reply back, and the command executes as the AI pilot. Anything outside the five known commands is rejected.