From Keyword Search to AI Search: 3 Ways to Add AI Capabilities to Your AddSearch

Your content is already indexed and your keyword search is live, so adding AI is not a rebuild. It is a question of which method you use to switch it on.

In this recorded session, Roberto Roca, Implementation Manager at AddSearch, walks through three ways to add AI Answers and AI Conversations to an existing AddSearch setup, hands-on in both code and the dashboard. Every answer you will see is generated from your own indexed content and shown with its sources, no external sources used. Using AddSearch's own content as the live example, Roberto takes each method end to end: the request, the response, the settings that matter, and the result running on a real page.

What Roberto covers

Method 1, the API: a single conversations endpoint that starts a conversation, handles follow-up questions with context, refines a conversational question into a clean keyword query, returns a single AI Answer on its own, and streams responses as they generate. Because you work directly with the API, you can build and style your own front end and combine AI Answers, AI Conversations, and Keyword Search in any layout, whether that is the full AddSearch Experience or any one capability on its own.

Method 2, the AddSearch libraries: adding AI Answers on top of an existing search using the Search UI and JS client libraries. Roberto shows the two changes it takes, then an answer that streams in above your current results with its sources and a thumbs up or down signal, while all your existing filters, tabs, sorting, and pagination keep working exactly as before.

Method 3, AI Conversations widget: AI Conversations widget to your site for multi-turn, context-aware conversational search. Roberto shows both routes, building and styling the widget in the dashboard with the widget builder and then embedding it, and the code-first route that takes a single script line. He covers sample questions, themes and light or dark mode, launching the widget from your search bar or an "Ask AI" button, and the automatic hand-off from a conversation into keyword results.

Everything Roberto demonstrates is captured in analytics too: the questions visitors ask, the answers given, sentiment, and timestamps, so every question becomes insight into what your visitors actually need.