Perplexity.ai

,

I’ve been using Perplexity more in recent months. I don’t have the Pro subscription yet, but considering it. Perplexity seems seems really good at summarizing search results. Not sure exactly why I like it, but it seems to provide me the information I need in a form that is very usable

Apparently, I’m not the only one:

Founded less than two years ago, Perplexity has become a many-times-a-day-use product for me, replacing many of my Google searches—and I’m not alone. With fewer than 50 employees, the company has a user base that’s grown to tens of millions. They’re also generating over $20 million ARR and taking on both Google and OpenAI in the battle for the future of search. Their recent fundraise of $63m values the company at more than $1 billion, and their investors include Nvidia, Jeff Bezos, Andrej Karpathy, Garry Tan, Dylan Field, Elad Gil, Nat Friedman, Daniel Gross, and Naval Ravikant (but sadly not me :sob:). Nvidia CEO Jensen Huang said he uses the product “almost every day.”

Is anyone else using Perplexity? If so, what do you think of it?

I am avid user for many months now.

I am continuing to use Perplexity more. Just for fun, I asked if it knew anything about a developer named Cliff Brake:

Hmm …

1 Like

One of the neat things about Perplexity is that it shows us how it thinks and does things. It often generates Python programs. See these examples:

https://www.perplexity.ai/search/recommend-a-momentum-trading-s-SaiYAR5LSKeuUp8ZqlbJKw?0=d

https://www.perplexity.ai/search/im-planning-a-business-trip-to-bBcXju4wRRCi_30W2zN4mw?0=a

Notice that in the Assets tab, there are Python, Markdown, HTML, CSS, and other files. Why don’t we as humans work more like this? When we have a problem to solve, start with a Markdown doc, put some data in CSV files, write a little Python to generate tables, graphs, diagrams, etc. Perhaps because humans are too lazy …

There are even tools like Juypter notebook, that do some of this in a single notebook format, but there are disadvantages with Juypter notebooks as they are stored in XML and are more difficult to edit collaboratively and review changes. It also mixes code and documentation, which is a little messy in the final presentation. Juypter also generally limits what you can do to Python and the things it supports. The freeform directory of files in a Git repo has no limitation.

What if AI could collaboratively work with us in a Git repo or workspace? What if we could record and version the AI prompts along with everything else? I’m sure this is already possible to some extent with tools like Cursor, Aider, and Copilot. It seems the future belongs to tools and humans who can work on same files in a local workspace. This allows for an iterative approach.