Skip to main content
AtomicIQEnterprise · Air-Gapped · API & SDK

Auditable AI built on your own knowledge, inside your perimeter.

Deploy auditable AI infrastructure across your organization: distribute signed Expansion Packs (data and Playbooks packaged together), generate examination-ready Run Records, and eliminate cloud token bills on servers you control.

AtomicIQ · deployment paths
  • Team Rollouts

    Package policies, manuals, and standardized Playbooks once into private Expansion Packs and distribute updates across every team without maintaining per-team RAG pipelines.

  • Self-Hosted & Air-Gapped Servers

    Run on standard Linux servers or disconnected networks with zero cloud calls, no GPU clusters, and $0 per-query API spend.

  • API & SDK Embedding

    Embed the same AI infrastructure layer inside your own software: scoped Expansion Packs, repeatable Playbooks, abstention guarantees, and sealed Run Records via API and SDK.

Built for regulated, high-stakes work.

Every deployment packages data and actions into portable Expansion Packs, keeps access scoped, eliminates per-token cloud meters, and backs every output with an auditable trail.

Source-grounded abstention & sealed Run Records

AtomicIQ releases statements only when backed by verified source passages and explicitly abstains when evidence is missing (0 / 396 unsupported statements released across internal banking-compliance and rulebook suites). Every run seals its inputs, citations, and outputs into an examination-ready Run Record.

Repeatable Playbooks in Studio

Pick from ready-to-run Playbook templates in Studio for executive summaries, compliance audits, contract reviews, and checklists. Choose the exact Items, Collections, or Expansion Packs from your Library for each run so out-of-scope files are never read.

Portable Expansion Packs for data & actions

Package pre-compiled Knowledge, pre-built Playbooks, or both into one cryptographically signed Expansion Pack. Equip desktops and air-gapped servers in seconds from the Marketplace or internal releases with zero manual ingestion or per-seat re-indexing.

Air-gapped, model agnostic & $0 per query

Eliminate metered cloud API bills, cloud GPU clusters, and frontier-model lock-in. Runs offline in 1.5 to 3.1 GB of memory on existing laptops or standard Linux servers, ships with a compact 4B open-source model, and swaps models cleanly without rebuilding Expansion Packs or Playbooks.

No GPU clusters. No per-token API bills.

Run the native desktop app on macOS (Apple Silicon) and Windows (64-bit), or deploy self-hosted Linux servers with Docker Compose. Measured at 1.5 to 3.1 GB of memory in use on an Apple M4 Pro laptop, eliminating cloud inference spend while staying protected by U.S. Patent 12,153,884.

macOS
A Mac with Apple Silicon. Intel Macs are not supported.
Windows
A 64-bit (x64) PC.
Linux
Docker with the Compose plugin. Runs as a self-hosted server.
Memory
1.5 to 3.1 GB in use, measured on an Apple M4 Pro laptop.
Network
None needed to use it. AtomicIQ runs fully offline.

Rolling it out to a team?

Individuals can just download. Organizations usually want help with a rollout, a server, or a pilot on their own documents.

  1. Tell us which you need. The form takes a minute.
  2. A person at NOLA AI reads it and replies by email.
  3. You can download and try the app in the meantime.
Prefer to talk first? Contact us
AtomicIQ · teams and pilots
What do you need?

Tick “I’m not a robot” above to turn on the button.

Ready to test it on your hardest documents?

Start with a focused pilot on your own rulebooks, contracts, or policies, running on hardware you control.