Concept in development · Tenet by TrueMadeAI
The district edge cloud

The district's private AI can live in the charging cart.

Tenet Grid ships the cart with a built-in AI appliance the district owns outright — a reliable engine for governed overnight work. Capable Chromebooks in the same cart join in as an optional subagent fleet that compounds every refresh, and every device is returned charged, secure, and ready for students by morning.

An appliance the district owns Subagent fleet as upside Charge first, compute second
Overnight
idle hours reclaimed
1 model / device
independent parallel jobs
Device identity
never student identity
By morning
charged · cool · ready

Figures and examples throughout this page are illustrative and conceptual — no customer results, benchmarks, or savings are implied.

The back office is next

AI automation is coming for the district back office.

Translation, document processing, ticket triage, compliance review, curriculum alignment — the administrative work that used to take people is becoming agentic. And the pattern frontier AI assistants use to do it is a fleet of parallel subagents: each one takes a single bounded task and runs it independently.

Fleets of subagents

Anthropic's Fable and OpenAI's Sol Ultra fan out fleets of small parallel subagents — one job each — instead of one monolithic call. Tenet Grid runs that same shape, on district hardware.

Sovereignty, not just savings

Your data never leaves the district. You control the cost — it's electricity, not a metered per-token bill — and you run the exact model you validated, with no silent quantization, rerouting, or deprecation by an outside vendor.

An appliance you own

The cart ships with a built-in unified-memory AI appliance as the reliable engine. Capable Chromebooks join as an optional subagent fleet that compounds every refresh — and harder work escalates to the box or an approved cloud.

A private appliance, with a subagent fleet as upside.

The cart's built-in unified-memory appliance — a Mac Studio, an NVIDIA DGX Spark, or an AMD Strix Halo mini-PC class box — is the primary engine and orchestrator: reliable capacity the district owns from day one. Every capable Chromebook in the cart can then join as a worker running a complete small quantized model, so the fleet compounds as devices refresh. The harder work escalates to the box or an approved cloud model. It's the same pattern behind Anthropic's Fable and OpenAI's Sol Ultra, brought inside the district walls.

Appliance first, fleet as upside. The built-in box is the dependable engine; the Chromebooks are optional parallel capacity that grows every refresh cycle.
Owned, not rented. The compute is already in the building; the marginal cost is the electricity, not a per-token bill.
Governed by default. Tenet is the orchestration and policy layer over both the appliance and the fleet.
The asset nobody is using

The unused infrastructure already sitting in schools

Every night, K–5 districts power down dozens or hundreds of centrally managed Chromebook Plus devices into charging carts. Plugged in. On the network. Owned outright. And completely idle until morning.

Fleets you already own

Hundreds of commonly configured devices per district — hardware that's already been paid for.

Centrally managed

One admin console, consistent policy, and remote control over every node in the fleet.

Predictable idle hours

Students sign out; the carts fill; the network quiets. The same window, every school night.

Physical ownership

Every node is district property behind district walls — not a stranger's device on a public marketplace.

A private district edge cloud — not a public marketplace.

The district owns the hardware, controls the workloads, decides what data may be processed, and keeps all resulting value. Tenet Grid simply assembles a private “district edge cloud” from assets the school already has — no third party, no outside tenants, no student devices leaving district control.

Charge first, compute second

How the overnight compute cycle works

The cart's built-in appliance runs the coordinator and handles the primary work, then distributes independent jobs to capable Chromebooks in parallel — not one giant model split across the network. Each device holds a complete, small quantized model and does its own bounded work, so only small jobs and results ever cross the local network, never model weights.

Sign-out

Students return and sign out of their Chromebooks for the day.

Into the cart

Devices are placed in an AI-ready charging cart and begin charging.

Verify readiness

The coordinator checks charge, temperature, hardware capability, schedule, and policy.

Divide the work

Approved jobs are split across the devices that qualify right now.

Local inference

Quantized models run private, on-device inference — no active student session involved.

Return results

Outputs flow back to the district's control plane over encrypted channels.

Wipe

Temporary job data and short-lived credentials are removed from every device.

Ready by morning

Compute stops early — every device cool, charged, updated, and ready for students.

Parallel, not collective. One Chromebook classifies documents, another translates communications, another builds embeddings, another evaluates AI-generated materials. Larger or harder tasks escalate to a district GPU server or an approved cloud model.
Batch institutional work

What a governed fleet can do overnight

Batch-oriented institutional tasks — not unsupervised agents making consequential decisions about students.

A new hardware category

The AI-ready charging cart

The classroom charging cart becomes an edge micro–data center overnight — and reverts to a plain charging cart by morning. Its first job is never compute. Its first job is student-device readiness.

For charging-cart manufacturers

Own the category before it exists.

The charging cart has been a commodity metal box for a decade — it moves power and nothing more. Tenet Grid turns it into a premium, software-attached edge appliance. The first manufacturer to ship a certified AI-ready cart sets the reference design and the standard everyone else follows.

Why a cart manufacturer should move first

A premium category

Create the "AI-ready cart" tier — a new product line above the commodity metal box.

A reason to refresh

Gives districts a fresh, fundable justification to replace aging charging carts.

Features that pay off

Active ventilation, smart power, and telemetry become value drivers — not cost.

Recurring revenue

Potential per-cart or per-device software licensing attached to the hardware.

Escape the commodity

Real differentiation from interchangeable metal-cart competitors.

Thought leadership first

Lead the narrative now — build strong positioning before committing product investment.

A seat at the AI table

Positions the company inside the school-AI conversation, not adjacent to it.

Pull-through hardware

Encourages districts toward higher-spec Chromebook Plus fleets the carts serve.

What's inside an AI-ready cart

Active, monitored ventilation Thermal sensors + auto shutdown Per-device power telemetry Smart USB-C power management Charge-first scheduling Embedded network backhaul Integrated orchestration controller Model-compatibility certification Physical emergency stop
Readiness beats compute, always. The cart automatically reduces or halts workloads when devices run warm, battery health shifts, the network degrades, charging falls behind, or the school day approaches.

District dashboard — illustrative

Energy

Consumption per cart, per night.

Readiness

Devices charged, cool, and updated.

Throughput

Jobs completed and success rate.

Avoided cost

Estimated cloud inference not purchased.

Safety

Thermal events, auto-pauses, emergency stops.

Dashboard metrics shown are conceptual placeholders, not measured results.

Security & district governance

Designed around district control and data minimization

We don't claim automatic FERPA compliance or perfect security. Tenet Grid is an architecture built around explicit governance, narrow permissions, and keeping sensitive work inside the district.

District-owned hardware, district-approved workloads only.
No execution inside an active student session.
Device identity, not student identity — no student passwords or general Workspace credentials.
Narrow, short-lived permissions issued per job.
Encrypted job delivery and results, end to end.
Physical, single-tenant isolation — each job runs alone on a device holding nothing else, then wiped. No shared multi-tenant GPU batching.
Ephemeral local storage with automatic deletion of temporary data.
District-defined data-classification limits on what may be processed locally.
Full workload audit trail — without retaining unnecessary prompts or sensitive content.
Automatic exclusion of lost, outdated, unhealthy, or noncompliant devices.
Optional escalation to a district server or approved cloud for tasks unfit for the edge.
Model sovereignty — you run the exact weights you certified, and they don't silently change underneath a district that validated their safety behavior.

A Chromebook can still be lost, stolen, or physically inspected. The institutional trust boundary and data-minimization posture reduce exposure; they do not eliminate it. Governance is explicit and district-defined.

Why now

Quantized models change the economics

Four trends are converging — and each Chromebook refresh cycle and model generation makes the edge more capable per parameter.

Fleets already exist

Schools already own large, centrally managed Chromebook fleets — the capital cost is sunk.

Capable Chromebook Plus

Newer devices carry stronger CPUs/GPUs and enough memory for small local models.

Quantization

Models are getting dramatically smaller and more efficient — small INT4 models now fit modest devices.

Private capacity demand

Districts want affordable AI without sending every document to an external cloud.

The trajectory favors the edge

As on-device model families (for example, Google's Gemma line) improve, useful narrow agents — classification, translation, transformation, evaluation — become increasingly practical on hardware districts already own. Today's sweet spot is bounded batch work; the capability envelope widens with every cycle.

Fleet wins on massive parallelism — thousands of independent document, translation, and classification jobs.
A single GPU wins when the district just wants one strong, always-on chatbot. Tenet Grid routes to whichever fits.
Model names and capabilities referenced are public, illustrative context — not benchmarks or endorsements.
Business & partner vision

A differentiated hardware category + recurring software

Delivered through charging-cart manufacturers, Chromebook OEMs, education resellers, and managed-service providers.

Retrofit kit

Upgrade existing compatible charging carts into AI-ready carts.

Certified “AI-ready” cart

A premium cart built for governed overnight compute from day one.

Orchestration software

Tenet Grid licensed per cart or per active device.

Bundled packages

Chromebook Plus + cart + deployment + management + support.

Approved agent catalog

A vetted library of institutional workloads districts can enable.

Hybrid routing

Local, district-server, and cloud routing under one control plane.

Chromebook OEM tier

For device makers — Lenovo, Dell, Acer, HP — a fresh reason to sell higher-spec Chromebook Plus fleets that the fleet-as-upside tier directly rewards.

Silicon & appliance tier

For unified-memory silicon — Apple, NVIDIA, AMD — the built-in cart appliance is a new K–12 channel for the exact chips these vendors want inside schools.

For partners

A differentiated premium hardware category and recurring software revenue on assets already in the channel.

For districts

Private AI capacity generated from existing, underutilized hardware — new institutional capacity, not new capital.

A disciplined first step

The one-cart pilot

Small, measured, and honest. One school, one cart, low-risk work, and hard limits — compared head-to-head with a cloud API and a single GPU workstation.

Scope

One school · one cart · ~30 Chromebook Plus devices
One small quantized model (≈270M–1B, INT4)
Limited low-risk, non-PII batch workloads
Strict power and temperature ceilings
Automatic shutdown well before morning
Direct comparison vs. a cloud API and one GPU workstation

What we measure

Useful jobs completed Accuracy / success rate Docs & tokens processed Energy consumed Device temperature Morning charge levels Battery-health impact Network utilization Est. cloud cost avoided IT admin time
The primary success criterion

“How much useful district work can one cart complete per dollar — without affecting device health, security, or morning readiness?”

Closing vision

The next school data center may already be sitting in a charging cart.

An appliance the district owns, a fleet that compounds, and new institutional capacity. Tenet doesn't just govern the district's AI usage — it turns the cart into governed, private AI capacity the district controls end to end.

Charge first, compute second Governed local inference Ready for students every morning