RentServers.comPrivate AI for law firms, on a server no one else uses
An open-weight model on a dedicated server in a US data center, run for your firm by our engineers. Prompts and outputs are never stored, nothing trains a model, and every access is logged. One fixed monthly price for the whole firm, not per seat.
What firms use it for
Work on your own files. Verify any case citation against a citator or the primary source, as your duty of competence requires.
- Document Q&A. Ask questions across contracts, discovery sets, policies and your own precedents, with answers that point back to the passages they came from.
- Summaries. Depositions, hearing transcripts, long agreements and correspondence chains, condensed for the file.
- First drafts. Letters, memos, clause variations and checklists, drafted from your templates and reviewed by the lawyer who signs them.
- Internal knowledge. Search firm guidance, intake notes and know-how without sending them to a public AI service.
Your vendor diligence file, ready
Ethics guidance asks you to understand how a vendor protects client confidences. We hand you the documents in one packet. The full mapping is on trust and security.
Request the security packet- A mapping of ABA Formal Opinion 512 vendor checks to how the service works
- Our SOC 2 status and a completed SIG Lite or CAIQ questionnaire
- Data processing terms and the US subprocessor list
- Contract terms on confidentiality, US-only data, 48-hour breach notice and deletion at the end of service
- Audit log and usage exports for your supervision records
One price for the firm
Capacity sets the plan, not headcount. Everyone at the firm can use the server within its load. Sold and billed by Wells Software Consulting LLC, with our margin shown on every plan.
Private LLM Core
$649/month
1 to 3 users at once
Private LLM GPU
$1,399/month
About 20 to 40 concurrent streams on an 8B model
Private LLM 70B
from $4,995/month
70B-class quality for long documents, on a 36-month term
Accounting and tax firms
The FTC Safeguards Rule asks firms to choose service providers that maintain appropriate safeguards and to require them by contract. Our Private LLM Hosting terms and data processing terms cover confidentiality, security incidents and deletion. If your client consents under Internal Revenue Code section 7216 must name the provider, we give you our legal name and address for the form.
Questions firms ask
Does a private server remove the need for client consent?
Not by itself, and we don't claim it does. ABA Formal Opinion 512 calls for informed client consent before client information goes into a self-learning tool. The model on your server is not self-learning: nothing you send trains it, and its weights change only when you approve a model change. Whether your engagement letters or a client's outside counsel guidelines still call for consent is your firm's judgment; the security packet gives you the facts to make it.
Who at RentServers.com can read our documents?
No one, in the ordinary course. Prompts and outputs are processed on your server and never written to disk or logs, and our staff do not access their content. When an engineer opens your organization to help you, every action is recorded in your audit log under their name.
Where is our data?
On a dedicated server in a US data center that only your firm uses. We will not move it outside the United States without your written consent, and every subprocessor we use is based in the United States.
Can it connect to our document management system?
Through its OpenAI-compatible API. Any tool that accepts a custom OpenAI endpoint can use it, and so can an integration your IT provider builds against iManage, NetDocuments, Clio or a file share. Tell us what you run and we scope the connection with you.
Does it work with our managed IT provider?
Yes. Many firms have their IT provider handle the integration and user setup. We work with them directly, and the console gives their staff their own roles, so they never need your owner account.
Which plan does a firm need?
Capacity, not seats, sets the plan. Private LLM Core suits a handful of people at once on a 7B to 14B model. Private LLM GPU serves about 20 to 40 concurrent streams on an 8b model with streaming. For 70B-class quality on long documents, Private LLM 70B runs on a dedicated 128 GB accelerator slice.
What happens to our data when we stop?
We delete the model, your data and logs from the server within 7 days after service ends. Ask before then for an export of your configuration or fine-tuned weights.