Skip to content

πŸ€– On-Premise Agents & Assistants

You have the hardware. You have the inference engine. The next question is inevitable:

This section answers that question by distinguishing two complementary types of tools that most on-premise architectures end up combining.


πŸ—ΊοΈ The two tracks

πŸ§‘β€πŸ’Ό Track A β€” Personal Assistants

AI that knows you.

A local personal assistant is software that:

  • remembers your documents, notes, and work context;
  • lets you ask questions about your own data;
  • ideally never sends your information outside.

πŸ‘‰ Compare personal assistants β†’


πŸ€– Track B β€” Custodian Agents

AI that acts for you.

A custodian agent is an autonomous agent that:

  • runs on a trigger (scheduled or event-driven);
  • maintains a vault, audits code, proposes sourced fixes in branches/PRs;
  • waits for human validation before any irreversible change.

πŸ‘‰ Discover custodian agents β†’


❓ The question that comes first

Before evaluating any tool in this section, one question must be asked:

The interface can be local, the marketing page can say β€œprivacy-first”, and yet the model may run on the vendor’s servers. This trap is common β€” and often not documented on the first page.

The sovereignty evaluation grid provides a 6-criterion protocol to audit any tool in under 15 minutes.


Give my opinion

Everything is optional: one answer is enough. This site sends nothing: you send it, the way you choose.

Your noteNo note
Your opinion is about
About what? (several choices possible)
What will be sent
Send by e-mail

Private, no account: opens your mail program with the message ready, for ia-on-prem@damien.becherini.fr.

Open a GitHub issue β†—

Public, with a GitHub account. You read the issue again before you publish it.

To paste it wherever you want.

Give at least a note, a subject or a message to be able to send.