BOS

Business Operating System

Business Growth, Made Simple

Menu

BOS AI Settings Assistant

Use this assistant to choose AI providers, understand model defaults, keep keys safe, and control how AI works across BOS.

Smart Tools ▾

Smart Tools

Open tools assigned to this page or choose from your saved tools.

On this page

No Smart Tools are selected for this page yet.

Choose from saved toolsOpen ▾

No saved Smart Tools are available yet.

Manage Smart ToolsOpen ▾

Add new tools, edit saved tools, or manage page assignments in the full Smart Tools settings area.

Open Smart Tools Settings
Provider-Agnostic AI

AI Settings

Configure BOS Assistant behavior, provider selection, model defaults, and safe AI rules. Cost approvals are handled separately under AI Approvals & Cost Controls.

BOS AI Settings Assistant

Ask BOS which provider to use, how models should be assigned, or how Assistant behavior should work.

AI keys must stay server-side. Production user keys should not be stored until encrypted secure storage is added.

V1 uses server environment variables. User-level provider keys are planned next.

Safe setup mode

Ask a question

Tell BOS what chart or workflow you need.

Provider + Model Quick Switch

Choose the default AI provider and model that powers BOS Assistant, email drafting help, page guidance, and normal AI support. External video, scraping, and design tools are selected on their own workflow pages and imported back into BOS.

Use custom when your provider releases a newer model that is not listed yet.

Advanced model routing

Later, BOS can allow different models for specific tasks like email rewriting, research prompts, funnel strategy, and campaign planning. For V1, one default keeps the normal BOS Assistant simple.

API Key Storage

Current mode: server environment keys only. Customer API key fields are intentionally disabled until encrypted secure storage is added.

Current provider

Google Gemini

Active through server environment settings.

Fallback model

gemini-2.5-flash

Controlled by server env until workspace model selection is added.

Key storage rule

Secure workspace keys next

Add encrypted provider settings before storing real customer keys.

Assistant Behavior

Universal BOS Assistant

Available across the CRM with page-aware guidance, workflow help, and next-step suggestions.

Confirm before action

Sending emails, texts, bulk research, enrichment, or future paid actions should require clear user confirmation.

Page-specific behavior

The same Assistant engine can appear as BOS Dashboard Assistant, Contact Assistant, Task Assistant, Funnel Assistant, and Video Creator Assistant.

Provider flexibility

Workspaces should eventually choose Gemini, OpenAI, Claude, BOS managed AI, or disabled mode.

Model Defaults

BOS should let each workspace choose provider, model, budget style, and daily research caps. Gemini and OpenAI should be primary choices, while Claude and other providers remain optional for users who accept higher costs.

BOS Assistant

Recommended: Gemini server default

Best first target for page-aware CRM guidance, email drafting, funnel ideas, chart help, and general assistant questions.

Email writing

Recommended: Fast / cost-controlled model

Most email improvements should be quick, affordable, and reviewed before sending.

Lead research / enrichment

Recommended: Daily capped model

Research and enrichment can create surprise costs, so BOS should use caps, approvals, and clear usage warnings.

Funnels / video scripts

Recommended: Stronger creative model

Campaign strategy, landing page copy, video scripts, hooks, and CTAs benefit from higher creativity and reasoning.

BOS Managed AI
Uses active server provider

Use BOS managed AI with built-in usage tracking and cost controls.

For V1, BOS Managed AI routes through the configured server-side Gemini provider.

Mode: managed

API key required: No

OpenAI
Provider shell only

Use your own OpenAI API key.

This provider is listed for future user-selected API key support, but it is not wired yet.

Mode: bring your own key

API key required: Yes

Anthropic
Provider shell only

Use your own Anthropic API key.

This provider is listed for future user-selected API key support, but it is not wired yet.

Mode: bring your own key

API key required: Yes

Google Gemini
Connected by server env

Use your own Google Gemini API key. Recommended first because it connects well with Google Workspace, Gemini, Flow/Veo, and future BOS video/funnel workflows.

Gemini can be used server-side without exposing the API key in the browser.

Mode: bring your own key

API key required: Yes

Mistral
Provider shell only

Use your own Mistral API key.

This provider is listed for future user-selected API key support, but it is not wired yet.

Mode: bring your own key

API key required: Yes

Groq
Provider shell only

Use your own Groq API key for fast model inference.

This provider is listed for future user-selected API key support, but it is not wired yet.

Mode: bring your own key

API key required: Yes

Local Model
Provider shell only

Use a self-hosted or local model later.

This provider is listed for future user-selected API key support, but it is not wired yet.

Mode: local

API key required: No

AI Disabled
Available

Turn off AI generation and use manual workflows only.

Users can disable AI generation and use manual workflows only.

Mode: disabled

API key required: No