Patrick · Sovereign AI Appliance

Your AI workforce, inside your walls.

Patrick is Tymeline's sovereign AI appliance — a sealed, self-contained hardware unit that brings your AI Employees fully on-premise. Everything happens on the box. Your data, your models, and every decision stay in your environment and never touch the cloud. Built to run completely air-gapped: disconnect it from the network and it keeps working. Purpose-built for sensitive, regulated, and mission-critical programs where information simply cannot leave the building.

100%
Sovereign · nothing leaves the building
Air-gapped by design
Zero
Cloud dependency · zero data exposure
Single, sealed unit
Turnkey
Ship, install, switch on · private AI live
Deploy where the work is
SOVEREIGN AI APPLIANCE — TECHNICAL SCHEMATIC CONFIDENTIAL · TYMELINE × ELEMENT 31 · NOT TO SCALE SHOEBOX FORM FACTOR 1U · SEALED A · OLED DISPLAY sovereign state · audit D · HEAT FINS compute cooling E · TAMPER SEAL integrity guarantee STATUS LED ARRAY · B pwr · secure · audit VENTILATION · C passive thermal TYMELINE DWG NO. PK-01 PATRICK SOVEREIGN AI APPLIANCE SCALE NTS BUILT WITH ELEMENT 31 REV A
The Sovereignty Gap

Some programs cannot use cloud AI. At all. Ever.

Defense programs. Sovereign wealth. National labs. Silicon design at 3nm. Regulated financial infrastructure. Classified research. There's a class of work where the data cannot cross a network boundary — where the answer to "can we use OpenAI or Anthropic here?" is a hard no, backed by regulation, national security policy, or hundred-million-dollar IP. Until Patrick, that class of work was excluded from the AI Employee revolution entirely.

$100M+
semiconductor design IP where a single leak resets years of R&D
Silicon programs at 3nm and below carry IP so valuable that traditional cloud deployment is a non-starter. Design houses run air-gapped by policy — and were locked out of AI Employees until now.
Industry · 3nm and below
Classified
defense programs where cloud AI use is prohibited by mandate
Government contractors and defense primes work under classification levels that prohibit any external inference. Patrick brings AI Employees inside the SCIF, inside the secure facility, inside the classified enclave.
National security · regulated environments
Sovereign
data residency regulations for financial and health infrastructure
GDPR, HIPAA, national data residency mandates, and sovereign wealth policy require data to stay physically within specific jurisdictions or facilities. Patrick guarantees this at the hardware layer.
Regulatory · global
Patrick closes the gap. Every AI Employee, the full orchestration layer, the intelligence surface, the audit chain — all of it running inside a single sealed appliance behind your firewall, or entirely disconnected from any network. Sovereign AI, not sovereign compromise.
What Runs On Patrick

The whole Tymeline platform. None of the exposure.

Patrick isn't a stripped-down version of Tymeline. It's the full platform — AI Employees, orchestration, and organizational intelligence — running locally on sealed, sovereign infrastructure. Whatever your team does with Tymeline in SaaS or VPC, they do inside Patrick. The only difference is where the compute lives, and who never touches the data.

3 layers
Platform · one appliance
Local
Inference · no telemetry
Sealed
Tamper-resistant · Element 31 infrastructure
Turnkey
Ship, install, switch on
Layer 01 · Execute
Verified AI Employees
Every AI Employee you deploy in Tymeline SaaS or VPC runs identically on Patrick — RTL design, verification, DFT, code review, project coordination, vendor liaison, and more. Same behavior, same interfaces, same audit. Different physical location.
Deployment: Local inference · no cloud call
Layer 02 · Orchestrate
Plan · Execute · Govern
Full initiative orchestration across teams and programs. Live status, scope drift detection, milestone tracking, decision governance — the entire operational surface, running on Patrick's local compute.
Availability: Works fully air-gapped
Layer 03 · Intelligence
Organizational Intelligence
Workforce, project, resource, and risk intelligence — with Via, Tymeline's conversational AI command interface. Leaders ask questions, get answers, run scenarios. Every query stays on the box.
Data flow: Never leaves Patrick
Trust · On-Device
Blockchain-Anchored Audit
Every consequential AI Employee decision is cryptographically anchored to a tamper-evident record on-device. No cloud dependency. No external service. The chain of custody starts and ends inside Patrick.
Storage: Local · immutable · on-appliance
Identity · Local IdP
Verified Tymeline ID
Same identity model as SaaS — AI Employees and humans are verified named identities. Patrick supports local identity providers (LDAP, Active Directory on-premise, local Okta bridges) so the identity chain never leaves your walls either.
IdP: Local · sovereign
Hardware · Element 31
Sovereign Infrastructure
Patrick's compute is integrated by Element 31 — sovereign AI infrastructure purpose-built for on-premise, tamper-resistant AI operation. No shared cloud components, no supply-chain exposure, no telemetry channels.
Compute: Sovereign, integrated, on-premise
Where Patrick Deploys

Wherever the work is. Wherever the network can't reach.

Patrick is designed to be deployed exactly where your most sensitive work already happens. Sealed. Powered on. Live. The unit needs power, cooling, and — optionally — network access to peer instances. That's it.

S
Secure Facility
Semiconductor design floors, R&D labs, SCIFs, secure data rooms, sovereign wealth trading environments. Deploy Patrick inside — AI Employees live where the sensitive work happens, not on the other side of a firewall.
F
Forward Site
Field operations, disconnected environments, mobile command centers, remote research stations. Patrick works fully air-gapped — deploy where connectivity is unreliable, restricted, or forbidden.
R
Regulated Office
Regulated financial institutions, healthcare providers, government agencies, and jurisdictions with strict data residency mandates. Patrick guarantees data locality at the hardware layer.
D
Distributed Sovereignty
Multi-site organizations can deploy Patrick units per facility, per program, per classification level. Each unit sovereign to its own scope. Peer to peer coordination optional.
Sealed. Tamper-resistant. Independently attested.
SOC 2 Type IIISO 27001FIPS 140-2 targetCommon Criteria roadmapAir-gapped by designElement 31 infrastructure
Built With Element 31

Sovereign infrastructure meets sovereign AI. One appliance.

Patrick is built in partnership with Element 31 — a sovereign AI appliance semiconductor startup building the hardware layer for on-premise, air-gapped AI compute. Tymeline provides the AI Employees, the orchestration, and the intelligence surface. Element 31 provides the purpose-built sovereign AI infrastructure that runs it all sealed, sovereign, and tamper-resistant.

Partnership · Tymeline × Element 31
Sovereign AI stack·Infrastructure to platform
Element 31 · The Infrastructure
Purpose-built sovereign compute.
Element 31 designs and integrates the sovereign AI infrastructure inside Patrick — a compute platform built from the ground up for on-premise, tamper-resistant AI inference. No shared cloud components, no supply-chain exposure, no telemetry channels.
HWSovereign AI infrastructure
compute integrated for on-premise inference
HWTamper-resistant enclosure
sealed unit · physical integrity guarantees
Tymeline · The Platform
AI Employees, orchestration, intelligence.
The full Tymeline platform runs on Element 31 infrastructure. Every AI Employee, every governance surface, every intelligence query — all executed locally on the box. Same platform experience as SaaS or VPC. Different sovereignty guarantee.
SWAI Employees
full roster deployed · local inference
SWPlan · Execute · Govern
full orchestration on-device
SWVia + intelligence surface
conversational AI · local execution
Together · The Appliance
Turnkey sovereign AI.
Patrick ships as a single, sealed unit. Infrastructure, platform, models, identity, audit — all pre-integrated, pre-hardened, ready to deploy. Ship it to your facility, connect power, switch it on. Your AI workforce goes live inside your walls the same day.
UNITSealed appliance
infrastructure + platform pre-integrated
UNITShip-and-switch-on
operational within hours
UNITOn-device audit chain
blockchain-anchored · local
Two teams. One appliance. Tymeline's AI workforce and orchestration platform, running on Element 31's sovereign AI infrastructure. Purpose-built for the class of programs where cloud AI is not an option.
Who Patrick Is For

The programs where cloud AI ends. The programs where Patrick begins.

Patrick is not for every Tymeline customer. It's for the class of programs where the sovereignty requirement is absolute — where cloud AI is not a design preference but a policy line that cannot be crossed. If any of these describe your work, we should talk.

01Semiconductor design at 3nm or below?
A single leak of RTL, verification data, or process node IP resets years of R&D and hundreds of millions in engineering investment. Design floors that have historically banned cloud AI can now deploy AI Employees inside the design environment itself. Patrick keeps the silicon IP where the silicon IP is designed.
02Defense or classified programs?
Government contractors and defense primes work under classification levels and export control regimes that categorically prohibit cloud-based inference. Patrick brings AI Employees inside the SCIF, inside the secure facility, inside the classified enclave — with local audit and no external touch points.
03Sovereign wealth or regulated finance?
Investment strategy data, trading logic, portfolio composition, and client positions cannot leave the fund's control. Patrick provides AI Employee capability for research, allocation, risk analysis, and back-office coordination — all inside the fund's sovereign infrastructure.
04Data residency mandates?
GDPR, HIPAA, national data residency laws, and sovereign wealth policies increasingly require data to remain within specific jurisdictions or facilities. Cloud AI providers cannot guarantee this at the hardware layer. Patrick can — it is the hardware layer.
05Forward or disconnected operations?
Field research, remote sites, mobile command posts, distributed operations centers. Cloud AI depends on connectivity that may be unreliable, restricted, or forbidden. Patrick works fully air-gapped — disconnect it from the network entirely and every AI Employee keeps operating.
06Foundational research or trade secrets?
National labs, pharmaceutical R&D, industrial process IP, and other environments where the research value is measured in decades and billions. Patrick removes cloud AI from the threat model entirely — the model that generated the answer never left the building.
If your programs fit any of these, Patrick was designed for you specifically.
Time To Sovereign AI

Ship. Install. Switch on.

Patrick is a turnkey appliance. It ships pre-integrated, pre-hardened, and pre-configured to your program scope. Installation is a physical process, not a software project. Your team plugs it in, powers it up, and your sovereign AI workforce is live the same day.

Order
4-8 weeks
Configuration · hardening · shipment
Install
1 day
Physical install · power on · validation
Live
Same day
AI Employees operational · on-device

Ready to see it in your walls?

Patrick briefings are hands-on. We arrive with the appliance specifications, security architecture documentation, deployment scenarios matched to your facility, and a partnership deck with Element 31. Bring your CISO, your Head of R&D, your Chief of Staff, and whoever owns the sovereignty decision. Let's talk about what "AI Employees inside your walls" would look like for the specific programs you can't run in the cloud.

Sealed · Tamper-resistantElement 31 infrastructureAir-gapped by designOn-device audit