AI Assistant Overview

Boltise ships with an AI assistant that understands FiveM and RedM development. Open it from the chat panel on the right of the IDE.

Two ways to use it

• Boltise AI (paid plans). Plus, Pro and Max include hosted Boltise AI on your billing cycle, no API key needed. Answers are grounded in a FiveM/RedM knowledge base. See Boltise AI.
• Bring your own key (all plans). Use your own provider — usage is never metered by Boltise. See Bring Your Own Key.

One assistant, not modes

There is no mode selector. Every message runs the same agent: it can read your project, search the knowledge base, run tools, and propose edits — and it asks before anything that writes a file, touches your database, or runs a command.

What changes between a quick question and a big refactor is not a setting, it is how much the agent needs to do. Ask “what does this event do?” and it reads and answers. Ask “add a job to qb-core” and it reads, edits and shows you diffs to approve.

See Tools and approvals for what it can reach and what it has to ask about first.

Knowledge base

Hosted Boltise AI requests are grounded in a FiveM/RedM knowledge base: official docs, native references, framework source (QBCore, ESX, QBox, VORP, RSG, ox_lib), and curated community solutions. Retrieval happens on our servers, before the request reaches the model — that is what makes the answers framework-aware instead of generic.

Note

This is the one thing your own key does not get. With BYOK the request goes straight from Boltise to your provider, so there is no server-side retrieval step and no knowledge base in the prompt. Everything else is identical: the same tools, the same project context, the same approvals.

Native lookup is the exception, and it works on every tier — the natives index ships inside the app, so the assistant can always check a signature rather than guess one.