Sign in Subscribe

AI Tools I've Been Digging

AI Tools I've Been Digging

AI Twitter and WhatsApp groups mostly cover two things: frontier labs battling for AGI supremacy and startups building AI infrastructure. Essentially, there’s a large buffet of AI capabilities and tools to serve them up, but the world needs founders to build real apps that give people tangible value.

I really admire programs like AI Grant that incentivized this early on. I'm always on the hunt for people who are finding new ways to apply generative AI and build something useful. Here are 3 AI-powered tools that I’ve been using and really enjoying.

Consensus: Scientific Literature Made Accessible

One tool that I’m enjoying is Consensus. It's essentially a LLM that's been tuned across a bunch of research papers. You can ask it any sort of question, like "Is creatine good for recovery?" and it will look across all the scientific literature and summarize the papers.

It has a brilliant way of pulling out the relevant metadata about each paper, e.g. the type of study, the number of participants, whether it was double-blind, and how much it's cited. It uses that information to give you the strength of each paper and creates a nice summary. If you ask a yes/no question, it even goes as far as to give you a direct answer they call the "Consensus Meter." It tries to take a stand in a way that a lot of AI products do not.

They have 200m+ papers in almost every field so there’s lots of ways to use it. For example, I’ve been tweaking my diet and exercise routine heading into 2025. Here are some of the suggested questions on the homepage:

  • “Is topical retinol skin cream safe to take while pregnant?” (3 out of 5 papers studying this say “yes”)
  • “Does increasing the minimum wage lead to higher unemployment?” (this seems to only hold up among younger workers).

Their mission is “to democratize access to the world's best knowledge” and the product definitely makes academic research accessible to non-academics like me.

You can sign up here: https://consensus.app/. There’s a free plan with unlimited searches (but limited AI) or $108/yr for premium.

GigaBrain: Unlocking Reddit and YouTube

GigaBrain is a search product across Reddit and YouTube. The product is built on a huge vector database of Reddit threads and video transcripts and uses RAG with all this content to source its answers. 

This is great for product research that I used to do on Wirecutter. For example, researching the best ultra-wide desktop monitor would normally require scrolling through dozens of Reddit threads but GigaBrain used 89 relevant, recent comments to come up with its recommendation:

I also really like it for home improvement questions. For example, we recently had a heat pump installed. We were considering several options and I wanted to know how one particular heat pump performed in warmer climates. It was awesome to find very specific Reddit threads discussing that, where previously I would do a Google search and have to scroll through threads to synthesize the information on my own.

I do have a few critiques with this site:

  • The "Pro" mode apparently uses a more advanced model but I didn't see any improvements when testing it out.
  • They’re experimenting with the UI so there are a few different ways to do things. They have a standard search view and they also let you start threads like ChatGPT. There’s a separate “Shopping” section and an “Interest Feed” that I think is based on your Reddit profile. I’ve just been sticking with the standard search but it would be nice if they picked a single UI.
  • It’s not clear how to turn on Youtube. I think this might only be on the paid version.

Overall, I still highly recommend GigaBrain and I’m sure they’ll make it more polished. Sign up here: https://thegigabrain.com/. This one is $80/yr.

Snipd: Actually Remember Podcasts

I've been a huge fan of a podcast app called Overcast for years, but Snipd was the first thing that got me to switch. It uses the latest AI models to process all podcast audio and make it better.

The first way is through “snips”. You can set a shortcut (I just double-tap on my AirPods) and it marks whatever concept was most recently discussed and saves a little snippet so you can review it later. What’s cool is these snippets have an AI summary of what’s being discussed vs. just marking the timestamp in the audio.

They've taken this a step further. If you're listening passively, you hear a little ding, and it will automatically snip the most interesting parts as you listen. I used to scramble to write these down in the Notes app or with a voice memo. Now, if you’re listening to podcasts while commuting or walking the dog, it builds up a library of interesting snippets from everything you’re listening to. 

I love using this combined with Readwise, a tool that centralizes notes from all of the content you consume. It pulls in highlights from Kindle and from read-it-later apps like Matter, then sends you a daily reminder of a few of these. Snipd integrates nicely so you get a morning reminder of the best parts of the podcasts you’re enjoying, moving beyond passively listening and helping you absorb the material. 

I just started doing this recently but I'm excited to build up a library of these snippets. I think there are a lot of interesting threads to pull, tying things together between different interviews, especially given how many podcasts I consume. It's definitely replacing a certain amount of book reading for me so I’m pretty excited to get more leverage out of this content.

You can get 1 month free with my link here: https://get.snipd.com/pAbF/ytfeo3kj. It’s $84/yr after that.