Should you take heed to loads of podcasts, you most likely keep in mind humorous anecdotes and surprise… “Wait, who talked about consuming french fries with Sriracha once more?” Or extra severe questions. To seek for solutions, you first want to seek out the podcast after which search its transcripts. Dexa is attempting to make trying to find podcasts simpler by leveraging synthetic intelligence.
The instrument means that you can ask questions on a single podcast, like Andrew Humberman’s Huberman Lab podcast within the screenshot under, or question all podcasts within the Dexa database – there are at present over 120 with extra being added. The search outcomes provides you with an AI-generated reply abstract in addition to tips that could the podcast the place the participant mentioned the subject.
For instance, you possibly can ask questions like, “What’s the easiest way to get extra sleep?” And discover solutions to that from Dexa’s podcast library with time-stamped hyperlinks to these conversations. You can even tag a selected podcast to slender down your search outcomes.
![](https://techcrunch.com/wp-content/uploads/2024/02/dexa-launch-5.jpeg?w=680)
Picture credit: dexa
Once I requested Dexa “What does Mark Zuckerberg take into consideration Threads?” She introduced again the outcomes of his dialog on the Lex Friedman present about The Social Community. Nevertheless it additionally confirmed me outcomes the place the podcast hosts had been speaking about Mark Zuckerberg’s remark about Threads on one other platform.
You can even share these searches and outcomes with others by hyperlinks. For that reason, Dexa may also index some content material for engines like google like Google.
Firm
Dexa was created by Riley Tomasek in 2023, who beforehand co-founded an app referred to as Flight for customers to seize notes, photographs, and concepts on completely different boards and collaborate throughout them. Figma acquired the corporate in 2021.
Tomasik stated he is gotten actually wholesome. As a part of the data constructing part, take heed to the Huberman Labs podcast. He wished to take the complement Huberman instructed however forgot the dose. So he began researching once more when the neuroscientist talked about it and the method took a very long time.
“After that have, I made a decision that I’d construct one thing to make it simpler as a result of I did not need to undergo that course of once more. So I transcribed all of the loops and used instruments like OpenAI and Whisper embeds to allow the search and revealed it to X,” Tomasik stated.
After Tomasik posted about his instruments, Hoberman retweeted them, sparking inside curiosity from podcasters and listeners in growing the instruments. He stated loads of podcasters talked about folks not having the ability to discover particular content material.
The startup makes use of completely different indexing methods mixed with a data graph of individuals, episodes, and reveals to boost its search. Tomasik stated the instrument was launched practically 9 months in the past and has answered greater than 1,000,000 inquiries with 50,000 folks visiting the location every month.
Dexa has raised $6 million in seed funding led by Summary Ventures and The Normal Partnership with participation from Maple VC and Guillermo Rauch. The corporate at present has three folks on the staff and is now planning to rent extra engineering workers.
The corporate is launching a redesigned web site and asserting a partnership with Huberman Labs the place Dexa analysis might be embedded into the podcast website. For that reason, the startup skilled a particular mannequin — consider it as GPT in ChatGPT.
The brand new design permits customers to ask follow-up questions and lets them watch the linked dialog with out having to go away the window.
Future roadmap
Dexa’s progress has been principally natural as podcasters refer folks in the direction of the instrument. He stated the strategy nonetheless has momentum as the corporate works with common podcasters. So she’s not seeking to spend money on networking proper now.
“We briefly tried promoting on podcasts, and we found that our makes an attempt resulted in a relationship the place we listed their content material, added it to Dexa, and so they began speaking about it organically,” he stated.
The corporate is not at present making any cash, however has thought-about subscriptions and adverts as a supply of earnings because it expands.
Dexa additionally plans to create a cellular app, which might entice extra consumption to podcasts with brief clips. Moreover, the corporate plans to supply an account creation function so folks can save their searches.
Alternatives and challenges
There are many instruments together with Riverside, Podcastle, and MusixMatch that assist podcast creators with transcription. When it comes to search, Very Disco’s newly pivoted service is attempting to grow to be a type of IMDB podcast with data like books, folks, and TV reveals talked about in podcasts.
Marco Paglia, chief product officer at MusixMatch, sees promise in what he calls the “assistant-driven person expertise” in Dexa. Moreover, he feels there may be potential to permit folks to retrieve the opinions of a specific individual by podcasting. However the cause why such analysis is just not included in MusixMatch is that it feels that textual content is just not the easiest way to work together with grasp’s diploma holders.
“I really feel like typing in chat is just not one of the best person expertise for assistants. As a substitute, for this use case, I strongly consider voice is the long run.
Ben Cmigla, associate at The Normal Partnership, believes Dexa’s method to utilizing data from innovators for analysis is exclusive.
“The success or failure of the platform will rely upon the belief it builds with customers and content material companions,” he stated. “Dexa’s method places belief first — the person’s belief that the solutions to their questions are coming from human consultants, and the associate’s belief that their content material might be considered.” “And never simply use it for coaching knowledge or anonymized summaries.”