Blog

Find answers in podcasts with GoodBarber Chatbot

A listener remembers an expert explaining how the city plans to reduce summer heat, but cannot remember the guest, programme or date. GoodBarber Chatbot can now use podcast descriptions and optional audio transcription to help listeners find answers in your archive and return to the relevant episode.

Backend & DevOps Engineer

  4 min read

The interview they remember, the episode they cannot find

Take a local radio station with live programmes, a schedule and interviews available on demand in its app. Every recording has a title and a description. Those help listeners browse, but an interview often contains useful details that never make it into the summary.

In our example, the episode is called “The city we share”. Its description introduces a conversation about everyday life in the city. During the recording, an urban planner explains the city's approach to summer heat: planting trees along school routes, adding shade to playgrounds and replacing some paved surfaces with planted areas.

None of those measures appears in the title or description. A listener remembers the subject and asks:

Which episode explained how the city plans to reduce summer heat?

We have added audio podcasts as knowledge sources for GoodBarber Chatbot so questions like this can lead back to your programmes. There are two levels: the written information you publish about each episode, and the spoken content you choose to have transcribed.

Start with the episode descriptions

In the Chatbot Sections tab, the Podcasts group brings together podcast sources created in the GoodBarber CMS and connected through external podcast feeds. Choose the programmes and categories you want the assistant to draw on.

The written descriptions of those episodes can become part of its knowledge base, without transcribing the audio. If a description names the guest or explains the subject of an interview, that information can already help Chatbot answer a listener's question and identify the episode.

For our station, a question about which episode features an urban planner could be answered from the description. The specific measures discussed during the interview require the recording itself.

Your archive stays organized around the programmes you publish. The assistant adds another way to explore it, starting with a question rather than a list of episode titles.

Bring the spoken details into the knowledge base

The optional Automatically transcribe podcast audio setting lets Chatbot draw on what was actually said in your recordings, not just their descriptions. It uses OpenAI Whisper or Mistral Voxtral Mini Transcribe 2 to turn spoken content into text and add it to the assistant's knowledge base.

This is where the summer-heat interview becomes useful in a different way. Its short description can stay focused on introducing the programme. Transcription makes the expert's explanation available for retrieval, including details the editor did not put in the summary.

The station does not have to anticipate every question a listener might ask when writing that summary. Once the recordings have been transcribed and added to the knowledge base, Chatbot can use the relevant passage to answer a question about it.

Automatic transcription uses your existing Chatbot credits. You can start with one interview series, then expand the sources your assistant uses as your archive grows.

From the answer back to listening

Once the interview's transcript has been indexed, the listener can ask about the summer-heat measures without knowing the programme title or the guest's name. Chatbot can explain the measures using the spoken passage and identify “The city we share” as the episode behind the answer.

With interactive source cards enabled, the answer can include an episode card with a time range to help locate the relevant passage. The card opens the episode in the app, giving the listener a route from the answer back to the recording.

That connection matters for a radio station. A short answer helps someone find the information they remembered. The episode lets them hear the guest's reasoning and the conversation around it. Discovery leads back to the station's own audio archive.

The listener starts with an idea. The station's programme names, categories and publication dates remain useful for browsing, but remembering them is no longer the starting requirement for this search.

Keep the archive's access rules

For podcast content created in GoodBarber and protected through Memberships, Chatbot respects the listener's subscription status. Subscribers can receive answers from the full indexed content available to them, including the audio transcription. For non-subscribers, the transcription of a premium episode remains protected. Answers use only the paragraphs of its description that you mark as free in the back office. Those paragraphs make up the episode's free preview text.

If our example interview were part of a subscriber-only series, the assistant would work alongside that access model. Your archive can become easier to explore while keeping your subscription offer in place.

Try a question your descriptions cannot answer

Start with one interview programme in your app and add its podcast source to GoodBarber Chatbot. Enable automatic transcription, then, once the recordings are available to the assistant, ask a question about something mentioned in the audio and follow the source card back to the episode. The Chatbot documentation covers the configuration steps.

The value is in that connection: a listener remembers an idea, asks a question and finds the programme that helps them hear more.

Keep reading

  4 min read

Your Twitch live streams and videos, now in your app

The short version. Twitch is an extension of GoodBarber, the no-code app builder, that adds a Twitch channel as the source of a Video section. When the channel is live, the stream sits at the top of the section; the rest of the time, the section lists the channel's videos. You connect it once, with a Client ID and a Client Secret from Twitch's developer console. It is free for Content apps: each section counts as one of the extensions included in your plan.The audience is on Twitch. In September 2026, viewers watched 1.51 billion hours of live streams there, with 2.1 million people watching at any given moment on average, according to TwitchTracker. Here is how the extension plays out for a creator who goes live every Tuesday at 8 pm.

  4 min read

Turn documents into answers with GoodBarber Chatbot

In any app, publishing a useful document is only part of the job. People also need to find the right information when they are about to act. A clearly named file helps someone browsing a library, but a reader may arrive with a more immediate question.Employee communication is only one possible application. The same approach can support learning resources, member guides, technical references or any other useful documents shared through an app. For our example, consider a colleague trying to book a client trip who needs to know:Which travel expenses need prior approval?We have added document sources to GoodBarber Chatbot so that the files you already maintain can support these conversations. You select the documents. Users ask in their own words. The assistant looks for relevant passages and uses them to build a response.The question becomes a way into the policy. Employees can start with the decision in front of them, even when they do not know the title of the document that explains it.

  5 min read

Siri Actions: ask your GoodBarber iOS app for its latest article or video

Siri Actions now answers two new questions: the latest article and the latest video in a section of a GoodBarber iOS app. Siri names the item and offers to open it by voice. They reach every app with its next store update. Siri Actions is the set of questions a GoodBarber iOS app answers out loud, for newspapers, radio stations, schools, organizations and any app that publishes content. Its first question, "What's new", gives an overview of your home screen. When we launched it in September 2026, we said the next one would be the latest article in a section. Here it is, and the latest video comes with it. Each goes straight to one item and offers to open it. Your audience goes from a question to the article in a single conversation, without searching for it.