Your AI Team, Ready to Work.

Private AI is an AI system that runs on hardware you control, so your prompts and documents never go to a public AI provider. Inceptus Reeve is a managed version: one sealed box on your own network, running open AI models, set up and looked after by Inceptus. It answers your team’s questions from your documents and helps run your IT and security tools.

They’re ready-made AI tools for specific jobs, like writing marketing copy, drafting proposals, triaging an inbox or answering HR questions. Each one comes set up, then adapts to your organization by learning from your approved documents, templates and brand voice. Your team picks the task instead of writing prompts, and a person approves anything that goes out.

On free and personal plans, it can. OpenAI’s help center says that for its services for individuals, “we may use your content to train our models” unless you turn that setting off. OpenAI says its business, enterprise and API plans don’t train on your data by default. For most companies, the bigger risk is employees using personal accounts for work.

On Claude’s Free, Pro and Max plans, it can. Since Anthropic’s August 2025 terms update, it trains new models on those chats when the user’s training setting is on, and keeps that data for up to five years. With the setting off, chats are kept for 30 days. Anthropic says this doesn’t apply to its commercial offerings, including Claude for Work and the API.

Not on its standard plans. OpenAI lists only specific offerings as HIPAA-eligible under a Business Associate Agreement, such as ChatGPT for Healthcare, ChatGPT Enterprise with a regulated workspace, and its API with modified retention. Free, Plus and Pro accounts aren’t on that list, so patient information shouldn’t go into them. Inceptus Reeve processes it on a box inside your practice instead, and a Business Associate Agreement is available on request.

Only if the AI system sits inside your assessed CUI boundary, or is a cloud service that meets DFARS 252.204-7012 requirements (FedRAMP Moderate or equivalent). Consumer AI tools don’t meet that bar. A Sealed Inceptus Reeve box sits on your network with no route out, so it’s assessed as part of your own environment.

Yes. Open AI models can run fully offline on a local server. In Inceptus Reeve’s Sealed mode, the box has no route out of your network at all, and updates arrive as a verified media bundle. Connected mode adds a pull-only update path, approvals from your phone and approved connections to your cloud apps, and only calls a cloud AI model if you turn that option on.

Yes, on a Connected box. Reeve connects to the SaaS tools you approve through their APIs or MCP servers, so it can look up a record, update a ticket or file a document where your team already works. The AI model still runs on your box, only the data each task needs goes to that app, and every call is recorded.

MCP, the Model Context Protocol, is an open standard for connecting AI systems to tools and data. An MCP server offers an app’s actions, like “search contacts” or “create invoice,” in a form an AI can use. Inceptus Reeve can use MCP servers to reach your apps, with each connection approved, limited to the actions you allow, and logged.

Inceptus Reeve starts at $995 a month with $0 down for one team of up to five users, on a 36-month term that covers setup, the hardware and Uniview backup. The Growth plan, for three to four teams and up to 50 users, is $4,495 a month. If you’d rather pay setup upfront, the monthly fee drops to $500 for Starter and $2,500 for Growth. Final pricing is set after a readiness assessment.

It depends on how much GPU memory the box has and how many people use it at the same moment. As a planning estimate, a 24 GB office box suits one team of about 5 to 10 people, a 48–64 GB workstation up to about 50, and a 96–128 GB server up to about 150. Larger organizations use two or more servers. Every size runs on AMD GPUs. Final sizing is set during the readiness assessment.

Enterprise AI plans protect your data by contract, but it’s still processed on the vendor’s cloud, under its retention rules and sub-processors. Inceptus Reeve protects your data by location: the models and your documents sit on one box on your network, under your control and managed by Inceptus. Many businesses use both, a cloud tool for general work and Reeve for sensitive data.

Yes, if you choose to. Cloud AI models are off by default. You can turn one on for specific tasks, such as general research that doesn’t involve sensitive data, and decide which teams may use it. Everything else keeps running on your box, and every request sent to a cloud model is recorded. [confirm how the cloud-model option is controlled]

Not at everything. The largest cloud models are still stronger at open-ended reasoning and broad general knowledge. The open models on a Reeve box are picked and tested for specific jobs, like searching your documents, summarizing, drafting and working your tools. They do those well because they work from your own data and cite their sources.

Blocking alone rarely works, because people switch to personal phones and laptops. Most businesses need three things: a written AI use policy, monitoring for traffic to public AI tools, and an approved alternative that’s just as easy to use. Inceptus Reeve is that alternative, and the readiness assessment starts with how your team uses AI today.