Dexa AI

What is Dexa AI?

Dexa AI was a free answer engine that searched podcast transcripts and replied with short summaries linked to the exact clips behind them. As of October 2026, dexa.ai shows only a brief notice saying the service has shut down.

The system transcribed episodes, labelled each speaker and split long conversations into topic chapters. A case study by AssemblyAI, its speech-to-text supplier, reports that Dexa processed more than 3 million hours of audio this way. When you asked a question, Dexa searched those transcripts and wrote an answer with timestamped citations.

Riley Tomasek, who earlier co-founded the note-taking app Flight (bought by Figma in 2021), started Dexa in 2023. It raised a $6 million seed round in February 2024 from investors including Abstract Ventures, The General Partnership, Maple VC and Vercel founder Guillermo Rauch. The founder later moved on to Charlie Labs, and we found no published shutdown date or announcement.

Key Features

Dexa AI’s core features were cited answers and timestamped clip links, which let you check every claim against what a host or guest actually said.

• Cited question answering: Dexa answered natural-language questions with a short written summary drawn from podcast transcripts. Each point linked to its source, so you could tell which expert had made which claim.
• Timestamped source clips: Citations opened the original episode at the relevant moment rather than at the start. That spared listeners from scrubbing through two- or three-hour interviews to find a single passage.
• Speaker identification: Transcripts were labelled by speaker using diarisation from AssemblyAI. Answers could therefore separate a host’s opinion from a guest’s, which mattered on long interview shows where guests and hosts disagreed.
• Show filters with @mentions: Typing an @mention narrowed a search to one podcast instead of the whole index. TechCrunch reported more than 120 shows indexed in early 2024, so filtering kept answers focused.
• Follow-up questions: A 2024 redesign added follow-up prompts beneath each answer. You could refine a broad question, such as sleep advice, into a narrower one about timing or dosage without starting again.
• Shareable answer pages: Each answer had its own public link, built to be indexed by search engines. Signed-in users could also save people and topics to a Radar list to follow later.

Pros

Dexa AI’s strongest point was verifiability: every answer linked straight to the spoken source, so you never had to trust a summary blindly.

• Free to use: Dexa never charged listeners, and TechCrunch noted the company earned no revenue while it weighed subscriptions or ads. Students and casual listeners could ask as many questions as they wanted.
• Checkable answers: Timestamped citations made it easy to hear the context behind a claim. That was more dependable than general chatbots, which often summarise podcasts without showing where a statement came from.
• Growing expert archive: Coverage grew from about 200 Huberman Lab episodes to more than 120 shows by early 2024. Health, fitness and business listeners could compare what several experts said on one topic.
• Time saved: Instead of replaying a long episode, you could jump straight to the few minutes that answered your question. Busy listeners used it as an index to shows they had already heard.
• Simple interface: The site worked like a search box, with no settings to learn. First-time users could type a question and read a cited answer within seconds.

Cons

Dexa AI’s biggest drawback is that it no longer works: the homepage and show pages now display only a short closure notice.

• Service shut down: The site now shows a brief thank-you message instead of a search box. Any answers, saved topics or shared links you relied on are gone, so research notes must be rebuilt elsewhere.
• No export or warning: We found no public announcement, export tool or migration path. Users who had bookmarked answer pages lost them without an obvious way to download their history.
• Narrow library: Coverage centred on partner and expert shows, mostly in health, science and business. Questions about news, comedy or niche hobby podcasts often fell outside its index.
• No check on claims: Dexa reported faithfully what speakers said but did not judge whether the science was sound. Health answers still needed checking with a qualified professional before you acted on them.
• No mobile app: A mobile app with short clips was planned in 2024, but we found no official iOS or Android release. Phone users relied on the mobile website instead.

Who is Using Dexa AI?

Dexa AI was used mainly by podcast listeners, health enthusiasts and writers who wanted quick, sourced answers from long-form expert interviews.

• Huberman Lab listeners: Fans used the search embedded on the Huberman Lab site, which began life as Huberman AI. They looked up protocols on sleep, light exposure and supplements.
• Health and fitness enthusiasts: Readers compared advice from several doctors and researchers on one topic. Citations helped them hear caveats that a short summary might drop.
• Writers and researchers: Journalists and students traced who said what, and when. Timestamped links made quoting a podcast accurately much faster.
• Podcasters: Hosts such as Andrew Huberman partnered with Dexa to give audiences a searchable archive. It turned large back catalogues into a free reference tool for listeners.

Pricing

Dexa AI was free throughout its life and never launched a paid plan before it shut down (checked October 2026).

Early coverage said the company had considered subscriptions or advertising, but neither appeared before the closure.

These details are now historical, so check each alternative’s own pricing page before you sign up.

What Makes Dexa AI Unique?

Dexa AI answered only from what podcast hosts and guests had actually said, rather than from web pages. Every answer linked to the source moment, so it behaved like a listening companion rather than a general chatbot.

Its partnership model was also unusual. Dexa worked directly with hosts such as Andrew Huberman and embedded a custom-trained search on their own websites, which gave creators a reason to support it rather than see it as a threat.

Compatibilities and Integrations

Dexa AI ran only in a web browser and offered no public API, apps or export options before it closed.

• Platforms: Desktop and mobile browsers; no official mobile app was ever released.
• Content: Transcripts from more than 120 partner and expert podcasts, processed with AssemblyAI.
• Embeds: A custom search tool ran on the Huberman Lab website.
• Sharing: Public links for individual answers, now offline.
• API: No public API was ever advertised.

Dexa AI Tutorials

Dexa AI tutorials no longer apply because the service has shut down, so the steps below show how it worked and how to replace it.

1. Users opened dexa.ai in a browser and typed a specific question.
2. They read the summary and opened the timestamped clips to hear the context.
3. Follow-up prompts or @mentions narrowed an answer to a single show.
4. Today, try Perplexity for general cited answers drawn from the web.
5. Use Snipd or Podwise for AI summaries and notes on episodes you listen to.
6. For a single show, search its published transcripts or show notes where the host provides them.

How We Rated It

We rate Dexa AI 3.4 out of 5 for the period it operated, because cited answers and free access were offset by a narrow feature set and few integrations.

Summary

Dexa AI suited podcast listeners who wanted quick, cited answers without replaying whole episodes, and while it ran it did that job well. Its main limitation is now final: the site displays a shutdown notice, and saved answers are gone. We no longer recommend it, as there is nothing left to sign up for. Perplexity, Snipd and Podwise cover most of what it did between them. Whichever replacement you choose, keep your own notes with timestamps, so a future closure does not erase your research.