My first impressions of Evernote AI Search
I tried the new Evernote AI Search feature and I must say that it is well-executed. However, it requires some understanding of both Evernote and how LLMs work to utilize it most effectively.
On a side note, I think the term AI is still overused. AI Search, like Chat-GPT and others are using Large Language Models are extracting and manipulating the given text to try and give you the correct answer. Although, it is technically impressive, there is no real “intelligence” here.
How does Evernote AI Search work?
Evernote AI Search operates by initially generating a standard search that gathers the notes potentially containing the answer to your question. These notes are then forwarded to a third party (currently OpenAI), where the answer is returned along with a list of relevant notes.
You also have the option to request all notes from a specific date range, with particular tags, within certain notebooks, and/or that match given keywords. Your question is sent to OpenAI, and the appropriate Evernote advanced search syntax is returned. You will then see all the notes that match your query. This adds a natural language layer to Evernote’s advanced search, which is quite beneficial, especially since it does not send your notes to a third party in this scenario.
Is Evernote AI Search trustworthy?
Evernote does a good job of telling you exactly what you’re getting into. Not all of your notes are sent to OpenAI, only those relevant to the search, and they are automatically deleted from their servers after 30 days. If you’re not comfortable with this, don’t use it!
You can still use Evernote with regular search without the AI.
I wish there were a toggle to hide it from the UI altogether.
I think Evernote needs to add more options to allow users to tag notebooks, tags, or specific notes that will never be sent to the AI.
Evernote AI Search can help you find answers, maybe.
If we store notes, PDFs, images, etc. for reference, this AI search could be extremely useful for quickly finding what we need.
However, when I tested these AI queries, I did not get the results I expected. I thought that the question “When did I last change my oil in the Honda?” would find the receipts I have for my oil changes and return an answer with the date of my last oil change. Instead, all I received was a response stating that the information was not available, followed by a list of notes that matched some of the words I used in my query, most of which were far off. A regular search yielded better results.
I then ran the exact same query again and it found the answer (based on my latest receipt in Evernote).
I ran it yet again, and it failed to find the answer. Apparently, like any other LLM, its not consistent.
I created a new note that contains the written answer. However, even after 30 minutes, this note has yet to appear in my regular search results. Frustrating!
From my observations, the more descriptive your notes are, the easier it is for AI to find the answer. This feature still relies on Evernote search, which is generally excellent, but not perfect.
I found the same results with some of my other tests. Some tests worked just fine.
With that said, I feel like if you spend the time to make your notes better, then they’re even easier to find with regular search. Is the AI Search worth it?
Garbage In, Garbage Out
Another concern of mine is how some people may want to use this AI Search feature. Evernote is known as a place where you can capture anything, and it accumulates over the years. While the AI search can help extract specific content from this collection, it’s not addressing the main issue.
Another popular service, Readwise, is an excellent tool for capturing highlights from books, articles, podcasts, videos, tweets, etc. Since Readwise can automatically export to Evernote, this makes Evernote’s AI Search even more powerful.
The problem I have with this is that it skips the most important step, your thinking! If we blindly capture everything we find interesting, knowing that we’re building a giant repository of all this information, and Evernote will make it easy to find it again, are we getting real value out of this or are we simply creating our own personal search engine?
There is certainly some value there, but if your goal is to become a better thinker, then I’m guessing the act of thinking may need to be involved at some point during the process.
Will I use Evernote AI search (or Evernote in general?)
While Evernote has certainly been through tough times recently, I believe the service is better than it has ever been. The apps are quick and efficient. Sync also seems to be reliable after recent updates.
I’ve had an active Evernote account for many years. However, I haven’t been actively using Evernote for a while. I’m likely to let my account expire this year, particularly after the significant price increases.
If you’re looking for an all-in-one tool that can store anything and you’re okay with all of your data being stored in a single service, then Evernote is still a good option. If you want a tool to help you learn and think, or just want to control your data locally, then I would recommend looking elsewhere.
I don’t mean this to be a dig on Evernote. I actually think its a great service and hopefully has a bright future. I simply don’t think its for me, at least not today.