Balance answer quality behind the brand
Use owned content to answer visitor questions with less friction.
7-day free trial · No charge during trial
What this feature covers
Why it matters
The practical reason to use it.
Multi-model support is useful when it improves a real visitor experience, not when it turns the product into a provider comparison page.
How it works
A step-by-step look at the workflow.
Step 1
Choose a default model for the assistant based on the task and traffic profile.
Step 2
Switch to a different model when a conversation needs more speed, depth, or multimodal support.
Step 3
Keep the same knowledge and tools available so the assistant stays consistent across model changes.
Step 4
Review cost, latency, source use, and answer quality signals to refine which model should handle each question type.
Step 5
Turn those routing decisions into a repeatable rule so the team can improve quality without changing the website launch experience.
Core job
The main job this feature handles.
OpenAI models
Use OpenAI options for stronger reasoning and flexible routing when a visitor question needs more depth.
Anthropic models
Use Anthropic options for nuanced writing, long-context source review, and customer-facing responses where tone and reliability matter.
Google models
Use Google options when documents, images, and multimodal reasoning support the same grounded assistant.
Open and alternative models
Use additional providers when teams need cost flexibility, portability, or a different reasoning profile for a specific assistant route.
Daily use
How teams use it after launch.
Switch mid-conversation
Change models without losing chat history, retrieved context, or enabled handoff paths when the question needs a different response profile.
Cost control
Use lower-cost models for repetitive answer paths and reserve stronger models for questions where poor guidance would create follow-up work.
Assistant defaults
Set default models by assistant so support, product discovery, and content guidance each start from the right quality target.
BYOK support
Bring your own API keys when procurement, billing, or provider governance requires the model relationship to stay directly under your own vendor.
Control points
What to keep controlled.
Start with one bounded workflow
Use Models on the narrowest workflow where the team can measure whether the feature reduces friction, improves clarity, and creates better cost.
Keep edge cases visible
Review the conversations, prompts, and system actions tied to models so operators can see where the rollout still depends on manual judgment.
Connect the surrounding systems
Models is stronger when the feature sits beside the knowledge, integrations, and routing rules that already determine what happens after the first.
Expand after the first proof
Once the first deployment is stable, teams can extend models into more surfaces and assistants without rebuilding the same control model from.
What you get
The changes teams should notice first.
- Better cost control with model flexibility
- Higher quality for complex conversations
- Faster responses with optimized model selection
- More provider flexibility without changing the visitor experience
The facts do the selling
Plan facts, platform capabilities, and worked examples — every claim here is checkable, not a pitch.
White-label included — never a paid add-on. Copyright removal from $98/mo. Full white-label — custom domain, branded portal, your-domain emails — from $198/mo.
The white-label wedge
Platform fact
Training runs on your sitemap, PDFs, docs, and YouTube transcripts. Answers cite the source pages they came from.
Trained on your content
Platform fact
Five clients at $300/mo on a $198/mo Agency plan is $1,300+ of monthly margin before usage.
A 5-client agency on one flat plan
Worked example
Your questions, answered.
Tap any question about the product, pricing, security, or setup to see a straight answer.
InsertChat
Answers about InsertChat
Hi! Tap any question below and I'll answer it for you.
Models questions
Can I switch models without rebuilding the assistant?
Yes. The assistant configuration, approved sources, brand settings, and enabled handoff paths stay in place while the serving model changes. That lets teams compare providers without rebuilding the website experience. The operational question is whether models makes the workflow clearer once real conversations, real ownership, and real edge cases show up. That is the bar teams should use before they expand the rollout across more assistants, more channels, or more teams.
Why use multiple models instead of one?
Different visitor questions need different trade-offs. Multi-model support lets you use lighter models for common answers and stronger models for complex, long-context, or multimodal questions while keeping source grounding and brand behavior consistent. The operational question is whether models makes the workflow clearer once real conversations, real ownership, and real edge cases show up. That is the bar teams should use before they expand the rollout across more assistants, more channels, or more teams.
Does multi-model support help with cost control?
Yes. Teams can route traffic to the least expensive model that still meets the quality target, then escalate only the conversations that justify deeper reasoning or richer multimodal capability. That keeps model cost aligned with the business value of the request instead of treating every chat like the most expensive possible workload. The operational question is whether models makes the workflow clearer once real conversations, real ownership, and real edge cases show up. That is the bar teams should use before they expand the rollout across more assistants, more channels, or more teams.
Ready to get started?
Start your 7-day free trial. No charge during trial.
7-day free trial · No charge during trial