Element 31 × 10xGrowth

Your AI, kept close.
Your bills, kept small.

Element 31 builds Czar, a sealed AI appliance and the software to run it. Your models and data stay inside your own walls, and your AI costs stop climbing with every prompt.

Through our partnership, 10xGrowth helps banks, insurers and large enterprises in India explore whether it's right for them.

Start a conversation Or say hello on WhatsApp

Sealed & tamper-evident
·
Air-gap capable
·
Fixed, predictable cost

Czar by Element 31: a sealed, secure AI appliance with DC in, 5GbE, four 25GbE SFP28 ports and a service panel

Czar · Rev A · Assembled in the USA

−82.4%

monthly AI spend at a listed financial services firm

Read the story

The honest version

Almost every leader we talk to has the same two worries about AI.

“Where does our data actually go?”

Every prompt sent to a cloud AI service is processed on someone else's servers. For a bank or an insurer, that's a hard conversation with risk, compliance and audit, and often the reason AI pilots never leave the pilot stage.

“Why does the bill keep growing?”

Usage-based pricing means the more your teams enjoy using AI, the more you pay. There's rarely a ceiling, and every new use case quietly adds to the meter.

Czar quietly answers both. AI runs on hardware you own, inside your security perimeter, at a cost you can actually plan for.

Case study · Public financial services

A listed financial services firm cut its AI bill by 82.4%, and brought every prompt back in-house.

82.4%

Lower monthly AI spend

$444,960

Projected saving each year

100%

Data kept on their own premises

Monthly AI spend

Before · metered cloud AI$45,000

After · their own Element 31 appliance$7,920

Where they started

Cloud AI had become popular across teams, and the bill had grown to $40,000–50,000 a month with no ceiling in sight. Meanwhile, every prompt was being processed on third-party servers, which worried risk, compliance and audit.

What changed

Element 31 placed a sealed, air-gap capable appliance inside their own environment. Inference now runs on hardware they own, and teams kept working exactly as before, with no new habits to learn.

What it means

A fixed, predictable cost instead of a meter. A full, self-hosted log of AI usage. And integrity evidence they can hand to auditors and examiners whenever they're asked.

“The appliance made AI cheaper. Substrate made it defensible.”

From the Element 31 case study

Client name withheld under a mutual confidentiality agreement. Savings are modelled on the client's reported $40,000–50,000 monthly baseline (midpoint $45,000) and the 82.4% reduction achieved after deployment.

Your numbers

Curious what it could mean for you?

Pop in roughly what you spend on cloud AI each month. It's only a rough sketch, but it's often a useful place to start a conversation with your CFO.

Nothing you type here is saved or sent anywhere unless you choose to share it with Karan.

A gentle back-of-envelope

What would 82.4% look like for you?

Your current monthly spend on cloud AI (APIs, tokens, GPU hours)

Today, per month

—

With an appliance*

—

Potential saving in a year

—

*Applies the 82.4% reduction from the case study above. It's a starting point for a conversation, not a quote. Your real number depends on your workloads and models.

Made for BFSI

For banks, insurers and lenders, and the people who audit them.

Financial institutions sit on some of the most sensitive data there is. Czar lets your teams use AI on it every day, without that data ever leaving your building.

Credit & underwriting

Summarise financials and draft credit notes from your own documents.

KYC & AML review

Read and cross-check customer documents without them leaving your network.

Policies & circulars

Ask questions of internal policies and regulatory circulars, with citations.

Claims & operations

Triage claims, pull out key details and draft first responses.

Service copilots

Help agents answer faster, grounded in your product and process docs.

Internal knowledge

A private assistant for every team, across every department.

The questions your risk team will ask, answered kindly.

Keeping AI and data on your own premises makes conversations about RBI and SEBI data-localisation expectations, IRDAI guidelines and the DPDP Act far more straightforward. Your compliance team will still have the final word, of course.

Where does the data live?: On your premises, on hardware you own. It can run with no path to the internet at all.
Can we prove nothing was altered?: Sealed, tamper-evident hardware and continuous integrity checks produce evidence you can hand to auditors.
Who used AI, and for what?: Every request is logged locally. It's your record, not a vendor's retention policy.
What if a vendor changes its terms?: Your AI runs on hardware you own, so access doesn't depend on someone else's pricing, uptime or policy.

Not in financial services?

The same appliance serves any large enterprise with sensitive data or a growing AI bill.

Healthcare
Legal
Manufacturing
Government
Critical infrastructure
Conglomerates

What's inside

One quiet box. Everything it takes to run AI well.

Underneath Czar is Substrate, the software layer that turns a sealed appliance into a working, governable AI system. It arrives ready, so the system you accept is the system you run.

Assistants that remember context across sessions.

Answers grounded in your documents, with citations.

Your chosen models, served locally behind one API.

Fine-tune open models on your data, on-premise.

Tracing and quality checks for every AI operation.

Integrity verification and tamper-evident logs.

Connects only to the systems you approve.

Way 01

Branded appliance

Your own application, sealed onto Czar and delivered under your name.

Way 02

Team appliance

A private AI workspace where your teams run the internal apps they need.

Way 03

Platform direct

Just the platform and its APIs, for engineering teams who build everything themselves.

How we'd begin

No pressure. No jargon.

And if it isn't a fit, we'll tell you so.

1
A coffee-length chat
Thirty relaxed minutes about how you use AI today and what's on your mind.
2
Your numbers, together
We look at your current AI spend and the use cases you care about.
3
An honest view
A clear estimate, and a straight answer on whether an appliance fits.
4
A pilot, on your premises
If it makes sense, you try it where your data already lives.

Say hello

Hi, I'm Karan.

Karan Grover · 10xGrowth

I'm your point of contact for Element 31 at 10xGrowth. Tell me a little about what you're working on, and I'll get back to you personally.

Not ready to fill in a form? That's completely fine. Just drop me a message on WhatsApp.

WhatsApp+91 79884 83308

Good questions

Things people usually ask.

Do our teams need to change how they use AI?

Usually not. In the financial services case study, the appliance slotted into existing workflows, so people kept working the way they already did. What changed was where the work ran and what it cost.

Can it run completely offline?

Yes. Czar can run fully air-gapped with no path to the internet, or connect only to the internal systems you already secure. You can choose differently for each site.

Which AI models can we run on it?

Open-weight models of your choice, served locally behind a stable API. You can also fine-tune them on your own data, entirely on your premises, and keep the result.

How is it priced?

Instead of paying per token or per request, you pay for infrastructure you own, so the cost stays predictable as usage grows. Karan can take you through the purchase and subscription options.

Is this only for banks and insurers?

No. BFSI is where the need is sharpest, but the same appliance suits any large enterprise with sensitive data or a growing AI bill: healthcare, legal, manufacturing, government and critical infrastructure.

How do we get started?

With a relaxed conversation. We'll look at how you use AI today and what it costs, and give you an honest view on whether an appliance makes sense. If it doesn't, we'll say so.

Your AI, kept close.Your bills, kept small.