🛰️ Use case 25 HorizonRetrofit now · Greenfield next
Space robots: who keeps the memory of every operation when Earth is far away?
Persistent operating memory for space robots (station arms such as the Canadarm family and Dextre, free-flyers, lunar arms and rovers): every operation, anomaly, decision and handover kept on board when the ground is minutes or days away.
Researched Sep 27, 20264 solutions reviewedAI in the cloud: 0 of 4
Prismatic Agents makes Prismatic. Its row is first, highlighted and labelled, so you can compare every solution against it; the others follow alphabetically and are ranked by nobody. Every value carries an evidence label; "Where Prismatic leads" is our view as its maker.
The situation today
The Canadarm family built and maintains the ISS; Canadarm3 is being designed to be highly autonomous and, with NASA's Gateway paused in March 2026, redirected toward the Moon's surface. GITAI demonstrates autonomous arms; Perseverance drives Mars mostly by itself on a radiation-hardened processor. None publishes an on-board memory of operations that survives hardware changes. Full facts: file 15.
Machines and chips in use: Canadarm2 and Dextre on the ISS; Canadarm3 in development (CA$999.8M contract); GITAI S2 arms; Perseverance with two RAD750 CPUs and an FPGA vision co-processor.
Greenfield for Prismatic. No product here keeps a private, lasting local mind inside the machine: Prismatic on the Jetson AGX Orin 64 GB is designed for this gap.
The chart
MeasuredPublishedReportedEstimateNot disclosedNot disclosed is never shown as “no”.
Solution
Type
Machine and chip
Where its AI thinks
Lasting memory
Offline
Price
PrismaticWe make it
Lasting local mind (never controls motion or safety)
A 64 GB Mac mini (M5 Pro) at a ground test bed or analogue station first (retrofit); a Jetson AGX Orin 64 GB in new machines next (greenfield); flight hardware much later
Local only Measured
Lasting memory with sources, receipts and handover Measured
Yes Measured
US$99 per arm a month, unlimited turns; US$500 activation + Jetson AGX Orin 64 GB module US$2,999 once per arm (proposed sector licence) Estimate
Canadarm2 + Dextre · MDA / Canadian Space Agency
Station arm and dexterous handyman
ISS robotics
Operated from the station and the ground Published
Mission records on the ground Reported
Operator-controlled Published
Government programme Published
Canadarm3 · MDA Space / CSA
Autonomous next-generation arm
AI, machine vision; in development
Highly autonomous, on board Published
Not described Not disclosed
Designed to work without ground oversight Published
CA$999.8M contract (reported) Reported
GITAI S2 / Inchworm · GITAI
Autonomous space arms
Dual arm on the ISS; lunar studies
Autonomous demos Reported
Not described Not disclosed
Not disclosed Not disclosed
Not disclosed Not disclosed
Perseverance AutoNav · NASA JPL
Mars rover autonomy
Two RAD750 CPUs + FPGA co-processor
On board Reported
Not described Not disclosed
Yes (~88–90% of driving autonomous) Reported
Mission Published
What Prismatic would cost you
Sector price: US$99 per arm a month, unlimited. Proposed list prices in US dollars; change the numbers and everything updates.
First year—
Each year after—
Five years—
Prismatic licence only, five years—
What this sector pays today, five years—
Cloud AI instead, five years—
What this sector pays today: Government programmes; no per-unit software price published (Not disclosed). That is often a whole product; Prismatic adds the lasting local memory beside it. Hardware is bought by you from Apple or NVIDIA. Electricity at 14.2¢/kWh around the clock. Prices are proposals until the first customer agrees them. How Prismatic is licensed
Run your numbers
How far is the ground? Signal delay and the case for on-board memory
The farther the robot, the longer every question to Earth takes.
One-way delay (seconds)—
Round trip (seconds)—
Waiting a day (minutes)—
How it works: delay = distance ÷ speed of light (299,792 km/s). Physics, not an estimate: only the distance changes.
Reviews and gaps
One card per solution. Strengths first; gaps are about this job, not about the product being bad.
Canadarm2 + DextreMDA / Canadian Space Agency · Station arm and dexterous handyman
Our review. Canada's flagship: 25 years of building and maintaining the ISS.
Gaps for this job. Ground-operated; no on-board memory of lessons.
Every operation and anomaly kept on board with receipts, when the ground is far.
Recurring anomalies surfaced before the next operation.
Same memory across a decade of hardware and model changes.
Built for tight power budgets.
How Prismatic solves it
Host: A 64 GB Mac mini (M5 Pro) at a ground test bed or analogue station first (retrofit); a Jetson AGX Orin 64 GB in new machines next (greenfield); flight hardware much later.
Start on the ground: a Mac mini beside a space-robotics test bed (no motion, read-only).
A Jetson AGX Orin 64 GB in analogue habitats and rovers on Earth; flight needs radiation-tolerant hardware.
The robot's journal, anomalies and handovers kept with receipts (F33, F12, F34); recurring anomalies surfaced (F04).
Work orders offered by operators and accepted or refused (F36); every write confirmed (F11).
Mission procedures searchable with sources, missing values said missing (F25, F07, F05).
Its identity and memory move to new hardware or a new model (F44). Prismatic never controls motion or safety.
Hosts. Prismatic runs on a 64 GB Mac mini (M5 Pro) beside machines that already work, and is carried next into new machines on the Jetson AGX Orin 64 GB, the host for greenfield sites.
Retrofit: Mac mini with M5 Pro, 64 GB, 64 GB unified, 307 GB/s; about US$2,699 (US$1,699 base plus US$1,000 for 64 GB (Apple configuration prices, reported)).
Licence here (proposed sector price, US$): US$99 per arm a month, unlimited turns. What this sector pays today: Government programmes; no per-unit software price published Not disclosed. Design-partner pilot: US$5,000 for 90 days, credited in full to the first year's licence. How Prismatic is licensed
Status: Operations journal, receipts and resume run today (PPAI); space-grade hardware is a long road. Features used: PPAI gateway, Machine resume, Receipt ledger / audit, Recurrences, Work orders, Confirm-before-write, Local-only, Portable genome.
A day with it: Before an arm operation, the operator asks what happened the last three times the arm grappled this handrail. The on-board mind answers with the dated journal entries: one joint-temperature warning in shade, cleared by waiting 20 minutes, and proposes the same wait, for a yes.
Genesis Blueprint · Project blueprint (PRD)
The team Genesis would build for this
Every operation, anomaly and handover of a station robot arm kept with receipts, proven first on a ground test bed. A funded ground study on a robotics test bed is a project with a defined end.
Flight-readiness adviserWhat changes for flight hardwareCautious, forward-looking
Research liaisonLab and fundersCurious, diplomatic
First 30 days
Agree the read-only telemetry link
Journal ten test operations
Publish the first pre-operation brief
Write the grant study plan
How we'll know it works
Every operation journaled with a receipt
9 in 10 history questions answered correctly
Memory survives a restart and a new computer
In a Prismatic studio this blueprint waits in the Marketplace, ready to be born: Genesis gives each role an original name, look, voice and character, and nothing starts until the founder presses Generate. Founders can also write their own blueprint step by step in the Blueprint Builder.
Investment outlook
Score47/90
Money in the market2/5
Ease of funding now3/5
Local-mind advantage5/5
Approach: Ground test bed on a Mac mini first; on-board much later on flight hardware.
Market: Small in units, large per contract: a few robots, government budgets.
Funding: Moderate: space-tech funds and public contracts; long cycles.
Business model (hypothesis): Engineering study contracts, then a licence per robot or per mission with a space-robotics maker (hypothesis).
Why fund it, and the risks: A credibility halo for every other market, a Canadian flagship story (Canadarm), public money that reduces risk. Risks: very long cycles, small volumes, qualification cost.
First validation step: Offer a no-motion 'operations memory' study to a space-robotics maker or a university space-robotics lab; run it on a ground test bed. Goal: a funded study (for example a CSA technology-development contract).
Scores are first-pass hypotheses from general industry knowledge (UseCase file 05), not researched market sizes.
For patient investors
The same mind, market after market.
Retrofit sites pay first on a 64 GB Mac mini (M5 Pro) beside machines that already work. Greenfield machines carry the same mind inside, on the Jetson AGX Orin 64 GB, licensed per machine. The horizon (blackouts, wilderness, space) is where a mind that needs no cloud is not a feature but the point.