AI models
Pick the model.Switch it later.
The model is configuration, not architecture. You choose which one answers, per bot, and changing it does not mean rebuilding a flow or retraining from scratch.
- 4
- model familieschosen per bot, switchable at any time
- 30
- languagesgenerated from the API constant
- 24
- published processesthe model answers inside these
The options
What each one is good at.
Every one of these runs against your own content. The difference is how it reasons over it, not what it can see.
- 01ChatGPTRun GPT-4 for context-aware customer support answers, trained on your own content.GPT-4 Powered
- 02ClaudeUse Anthropic Claude for safe, nuanced conversations with your customers.Constitutional AI
- 03GeminiUse Google Gemini for multimodal chats that handle text and images.Multimodal AI
- 04AI AgentsDeploy autonomous AI agents that resolve complex customer issues end-to-end.Autonomous resolution
Bring your own key, and the traffic is yours.
Add an OpenAI, Anthropic or Google key and requests run on your account under your terms with that provider. Same for Twilio on the phone line.
EzyConn holds no SOC 2, ISO 27001 or HIPAA certification — the security page describes the mechanisms instead of showing badges.
Try it on your own content.
Point it at your site, ask it something only your documentation answers, and see which model you prefer.