Private AI inside your firewall. Starts at $500/month.
Sovereign AI for CMMC, HIPAA, FERPA, and other regulated environments. Your data and its search index never leave the building.
Data and index stay inside.
No third-party APIs. No off-site copy of your files or your search index.
Managed end to end.
We patch, monitor, and maintain the software stack, and the hardware too when you lease. Your team does not run an AI stack.
Audit-ready.
Built to the standard of CMMC L2, HIPAA, FERPA, and NIST 800-171.
Predictable cost.
No per-token billing and no surprise overages. Pay per seat on your own hardware, or one flat fee with a leased appliance.
Keeping the file local isn't enough. The index is the leak.
Keeping your files local is not enough if your AI ships the search index to the cloud. Most "on-prem" AI creates a second, queryable copy of your sensitive data off-site. DPLYD runs on physical hardware inside your perimeter. Your data and your index stay behind your firewall.
Read the full explainer →Two ways to run it.
Starts at $500/month
Includes 1 seat, then $25 per additional seat per month. Software, models, and management on hardware you buy and own.
$5,000/month, fully leased
As many seats as the hardware supports. Hardware, models, maintenance, patching, and support. No CapEx.
No per-token metering, ever. No surprise overages.
Example: a 10-person team on your own hardware is $725/month. Leased is $5,000/month, all-in, with as many seats as the hardware supports.
| CLOUD AI | DPLYD | |
|---|---|---|
| billing unit | per seat × per token × egress | per seat on your hardware, or one flat fee leased. Never per token |
| budget shape | grows with usage, unpredictable | predictable — seats and price are known up front, no usage-based surprises |
| capex | none, but perpetual metering | none if leased; you buy the hardware on the $500 tier |
| who can use it | licensed seats only | leased: as many seats as the hardware supports. Own hardware: $25/seat |
A workspace your team already knows how to use.
Skills, tools, and knowledge-base search — running entirely on your own hardware.
Compliance is a property of the platform — not a feature we bolt on.
Every deployment inherits the same hardened base — identity, isolation, segmentation, audit — via Lighthouse, our compliance layer. Sentinel records every check in an immutable audit log, so compliance is evidence on demand, not "trust us."
Identity & access
Zero-trust by default.
Data isolation
Strict separation of workloads and secrets.
Network segmentation
Narrow, enforced communication paths.
Supply-chain provenance
Signed artifacts, verified integrity.
Content policy enforcement
Continuous oversight of every AI interaction.
Managed lifecycle
Auto-reconciled updates, always consistent.
CMMC 2.0 L2 — deployed inside contractor environments preparing for Level 2 assessment; DPLYD keeps CUI and every derived index within the assessed boundary. See the case study →
HIPAA — PHI is processed on hardware inside your network and never transmitted to a third-party model API, so no model-vendor BAA is required. Access control, audit logging, and isolation map to the Security Rule safeguards.
FERPA — student records never leave district infrastructure; the district remains the sole custodian of education records and their derived data.
NIST 800-171 — the platform's identity, segmentation, and audit architecture is built against 800-171 control families; Sentinel exports the evidence.
Proof from the field.
DEFENSE & AEROSPACEAI Infrastructure for a Defense Contractor Preparing for CMMC 2.0
Deploy generative AI on 20 years of BD content without violating CMMC Level 2 posture or exposing controlled data to frontier models.
Read the case study → View all case studies →Built exclusively for regulated verticals.
Straight answers.
Q ▸ Does any data leave my network?
No. The model, the retrieval index, the embeddings, and the logs all live on-premises, inside your perimeter. Nothing derived from your documents is transmitted to a third-party API.
Q ▸ What is sovereign AI?
Generative AI that runs locally on hardware inside your premises, managed by us. You get the productivity of modern AI; your data never crosses your network boundary.
Q ▸ How is this different from a private cloud or GovCloud AI service?
GovCloud and "private cloud" AI is still someone else's computer — your content and its search index are processed and stored outside your walls, under someone else's controls. DPLYD moves the computer to you instead.
Q ▸ What does it cost?
Two options. On hardware you own: $500/month including 1 seat, then $25 per additional seat per month. Fully leased: $5,000/month flat with as many seats as the hardware supports, including hardware, maintenance, patching, and support. There is no per-token metering on either.
Questions your security team will ask
Q ▸ Who patches and audits DPLYD?
DPLYD manages patching, monitoring, and updates of the software under the monthly fee, and of the hardware on the leased tier. Audit logs are available to your team and your assessor.
Q ▸ Does it work in an air-gapped or restricted network?
Yes, on request. DPLYD can be deployed fully disconnected. Managed updates are delivered on a schedule and by a mechanism that fits your environment, including offline transfer for air-gapped enclaves.
Stop exposing your most sensitive data to the cloud.
Bring AI productivity to where your value lives.
See it on your own data.
A 30-minute walkthrough with the founder — DPLYD, Lighthouse, and Sentinel, on your own data.
Open scheduling calendar →