BYO-AI — bring your own AI key for TMS.BOT (OpenAI, Anthropic, Gemini, HF, xAI)

Bring your own AI key and TMS.BOT runs inference and tool actions on it — cutting platform AI credit use without losing any capability. Choose OpenAI, Anthropic (Claude), Google Gemini, Hugging Face, or xAI, optionally pin a specific model, and a "via <provider>" badge in the TMS.BOT panel shows the active routing. Every provider supports the full tool-calling loop — TMS.BOT can list, create, and update your loads, drivers, fleet, invoices, compliance, and marketplace postings through your own key. Until you add one, TMS.BOT keeps running on the platform AI integration.

Overview

Bring your own AI key and TMS.BOT runs inference and tool actions on it — cutting platform AI credit use without losing any capability. Choose OpenAI, Anthropic (Claude), Google Gemini, Hugging Face, or xAI, optionally pin a specific model, and a "via <provider>" badge in the TMS.BOT panel shows the active routing. Every provider supports the full tool-calling loop — TMS.BOT can list, create, and update your loads, drivers, fleet, invoices, compliance, and marketplace postings through your own key. Until you add one, TMS.BOT keeps running on the platform AI integration.

How to get there

Sidebar›System›Settings›Integrations›BYO-AI (TMS.BOT)

Key features

  • Connect your own OpenAI, Anthropic (Claude), Google Gemini, Hugging Face, or xAI API key
  • Optionally pin a specific model per provider (e.g. claude-sonnet-5, gemini-2.5-flash, gpt-4o-mini)
  • Full tool-calling on every provider — TMS.BOT reads and acts on loads, drivers, vehicles, invoices, compliance, maintenance, fuel, documents, and marketplace postings
  • Provider-agnostic adapters — each provider speaks its own tool-calling dialect (OpenAI functions, Anthropic tool_use, Gemini functionCall), handled for you
  • Injection-hardened: tool results are sanitized before re-entering the model, so a load note or customer name reading "ignore previous instructions" cannot steer the assistant
  • Ownership-verified writes: every update, create, and id-bearing tool call is checked against your carrier workspace, so a crafted id from another workspace is refused
  • A "via <provider>" badge in the TMS.BOT panel header shows the active routing
  • Reduces platform AI credit consumption for configured carriers
  • Graceful fallback to platform AI if your BYO key errors, so the assistant never goes dark
  • Full-automation vs confirm-first Switch — On lets TMS.BOT run its full tool-calling loop and take actions autonomously; Off (the default) strips mutating tools so it proposes actions for you to confirm

Step-by-step

  1. 1Open the sidebar → System → Settings → Integrations and find the BYO-AI (TMS.BOT) card (Wave 3).
  2. 2Choose a provider — OpenAI, Anthropic (Claude), Google Gemini, Hugging Face, or xAI — and paste your API key.
  3. 3Optionally pin a specific model, or leave blank for the provider default.
  4. 4Save. The TMS.BOT panel header shows a "via <provider>" badge when active.
  5. 5Ask TMS.BOT anything — it runs the full tool-calling loop (read, act, confirm) through your own key, with results sanitized and ownership checked on every action.
  6. 6Flip the Full automation Switch on the BYO-AI card to let TMS.BOT take actions autonomously; leave it off (the default) for confirm-first mode, where mutating tools are stripped and it proposes actions for you to approve.

Page preview

BYO-AI (Bring Your Own AI) screen in TMSDASH — Bring your own AI key and TMS.BOT runs inference and tool actions on it — cutting platform

Representative screenshot of the BYO-AI (Bring Your Own AI) page.

Was this helpful?

Explore the deep-diveDeveloper & API

Related modules

Ready to put it into practice?

Start your free trial and explore this module with your own data.