Features Overview

AIsh.li is a privacy-first platform for building, embedding, and integrating purpose-built AI agents with real-world tools, structured forms, and zero model training on your data.

Jump to: AI Agents • Chat Widgets • Conversational Forms • Privacy • Models • Thinking • i18n • Tools • Developer API

AI Agents #

Purpose-built assistants tailored for specific domains, tasks, or customer workflows. Customize persona instructions, models, and capabilities visually or via declarative YAML code.

👉 Try it in practice: Explore Demo Agents • Bistro Concierge • Create Custom Agent

📖 Learn more: AI Agents Guide

Embeddable Chat Widgets #

Embed custom AI assistants on any website with two lines of standard HTML. Includes origin domain security to protect your quota, responsive mobile styling, and dark mode out of the box.

👉 Try it in practice: Test the embed in our Interactive Widget Playground

📖 Learn more: Chat Widgets Guide

Conversational Forms & Progressive Disclosure #

Replace tedious static web forms with natural dialogue. Form agents extract and validate required JSON fields conversationally while a reactive side panel progressively discloses recognized values in real time.

👉 Try it in practice: Launch the Transport Planner and say "I want to travel from Paris to Rome tomorrow with Luigi" to watch fields update live!

📖 Learn more: Conversational Forms Guide • Forms API

Privacy-Respecting AI #

Unlike consumer AI chatbots that may train on your prompts or sample transcripts for human review, AIsh.li is built on commercial, enterprise-grade APIs with an ironclad zero-training commitment:

• Zero Model Training: Your conversations and customer messages are never used to train foundation models.
• Enterprise Data Protection: Commercial API terms prohibit third-party human review of private chats.
• Data Sovereignty: Bring Your Own Key (BYOK) for direct provider billing and cookieless telemetry.

👉 Try it in practice: Connect your credentials under Your Secrets

📖 Learn more: Privacy Overview • Privacy Policy

Next-Gen AI Models #

Powered by Google's latest multimodal Gemini models, optimized for complex reasoning, high-speed conversation, and cost-efficient throughput. When model versions evolve, existing agents and API endpoints are automatically redirected with zero downtime.

👉 Try it in practice: Switch models in Chat or select a model for a new agent

📖 Learn more: Supported AI Models

Transparent Thinking & Reasoning #

Watch complex reasoning unfold in real time. AIsh.li streams the model's inner chain of thought into an expandable accordion, providing complete auditability while keeping user answers concise.

👉 Try it in practice: Ask the Gemini Agent a logic puzzle or math riddle in Chat and click Thought process to inspect its reasoning

📖 Learn more: Thinking & Reasoning Guide

Agent Internationalization (i18n) #

Deploy agents globally with native understanding of 100+ languages. Localize instructions, starter prompts, and form schema descriptions across English, German, French, Italian, and beyond.

👉 Try it in practice: Switch languages in the top header, or chat with the Bistro Concierge in French ("Je voudrais réserver une table")

📖 Learn more: Agent Internationalization Guide

Connected Real-World Tools #

Equip AI agents with eyes and hands to interact with the live web and physical world:

👉 Try it in practice: Ask Google Gemini in Chat "What's new in the world today?" to see live search citations in action

📖 Learn more: Connected Tools Overview

Developer REST API #

Everything in AIsh.li can be automated via an open REST API with Server-Sent Events (SSE) streaming for real-time tokens. Integrate assistants into Python, Node.js, Go, or custom frontends.

👉 Try it in practice: Generate an API Key or test endpoints in the Interactive API Reference

📖 Learn more: API Reference • API Key Management