AI features for your app

1. 1

   Decide what AI feature you want to add to your app, such as a chatbot, text summarizer, or image generator.

   Lovable can add many different AI features, so knowing what you want helps you ask for the right thing.

2. 2

   If you want to change how Lovable asks for permission to add AI features, go to 'Connectors', select 'AI', and adjust settings under 'Manage my agent's permissions'.

   This lets you choose if Lovable should always add AI features, ask you first, or never add them.

3. 3

   Tell Lovable what AI feature you want to build in your app, describing the function and, if you have a preference, which model to use (e.g., 'Build a chatbot using Gemini 3.6 Flash').

   Lovable will set up the necessary AI infrastructure and code for that feature directly in your app.

Add a chatbot to my app that can answer questions about my product documentation. Use the Gemini 3.6 Flash model.

API keys

Secret codes that identify you to an online service, like a password, allowing your app to use that service.

Backend

The part of an app that runs on a server, handling data and logic that users don't see directly.

Streaming

Receiving information in small pieces continuously, rather than waiting for the entire response to be ready.

Server-sent events (SSE)

A way for a server to automatically send updates to a web page, allowing for real-time information flow.

Semantic search

Searching for information based on the meaning of words, not just exact keyword matches.

Retrieval-augmented generation (RAG)

An AI technique where a model first finds relevant information from a knowledge base and then uses that information to generate a more accurate answer.

Embedding models

AI models that convert text or other data into numerical representations, making it easier for computers to understand and compare their meanings.

Latency

The delay between when you make a request and when you receive a response.

• Do not try to set up API keys or manage billing for AI providers yourself, as Lovable handles this automatically.
• Do not expect AI features to run outside your app; they are distinct from the Lovable agent that helps you build.
• Be aware that some models (like GPT-5.5 Pro and GPT-5.4 Pro) always 'reason before answering' and are not suitable for simple chat replies.
• Remember that showing a model's 'thinking' or using 'priority processing' for faster responses will increase the cost and potentially the time taken for an answer.
• Understand that video generation is charged per second, can be expensive, and has limits on simultaneous clips, even if your app doesn't show the video.