SerpApi.Org
I long avoided using AI in my workflows when it first came out. As a writer, I was irritated that it was taking jobs like mine. It also didn't work well when it first came out, and its answers were incorrect; students were using it to plagiarize term papers, and some tech websites were using it to hastily create blog posts that were half true.
Over time, however, I started to discover the power of AI and how to integrate it into my workflow. For instance, using Perplexity for my workflows. Another feature of Perplexity and Claude is the ability to switch models and use the deep research functionality. It is meant to be used as an assistant that provides more data than simple search queries by creating overviews, specific information, and data points.
Using Deep Research in Perplexity
Simply change the mode
When I needed to do deep dives into technical information, I started using Perplexity's Deep Research feature. All I have to do is change the mode in the search box before submitting my query. Once my query is submitted, it does, well, deep research and puts everything into headers and data points. It is more than doing a simple search query.
I can also combine it with Perplexity Computer for extra power; for instance, when I want to organize my Outlook inbox for my workflow. While I have a Perplexity Pro account, the deep research feature takes longer than a simple search, but it provides more detailed information about the subject I'm learning about. Also, it's important to note that free users get only a few Deep Research queries per month, so save them for the research you really need detailed information on. Still, even a Pro account has a limited number of queries — around 20/month.
Once the cap is reached, a user must wait for the Deep Research queries to regenerate — about one every 36 hours. So, unlike using the Perplexity Computer feature, I can't purchase new tokens; I simply need to wait it out. However, I can still use the regular search function, which lets me choose from nine different models. A Pro account also allows up to 200 search queries per day, with the limit replenished every 24 hours.
Perplexity Deep Research thinking
It's not a quick query
When using Deep Research, I found that it doesn't provide fast answers to queries like the basic search function does, with the "fastest" model enabled. It actually thinks and does additional research for complex or simple queries you ask it. So, expect the results to take longer. I used it like this: I enter my query, leave it alone, and once it's done, read the results.
I can do follow-up queries, but it's important to remember that even with a Pro account, or especially a free account, the number of queries you can research is limited. Still, when I want something more than a basic search-based answer to my research questions, the Deep Research function does a much better job. Also, when combined with Perplexity Computer — for which you can buy additional tokens — I get more detailed data to support my workflow.
Claude starts to creep into my workflow
I found out Claude can summon Perplexity
Now, I know most people reading this site will likely turn to Claude first. However, I started mastering Perplexity while I was a reporter for a newspaper gig, and I dig it. However, since everyone around here is a Claude-first user, I decided to give it a shot. What I found interesting is that Claude's Model Context Protocol (MCP) tools can "call" Perplexity, but not the other way around. It can use Perplexity's ask, search, reason, and deep research functions.
What I like about Claude is that I can just start typing queries in the chat box, and its deep research functionality is built in. I don't need to select a special feature to access it. I dabbled with it in free mode but went ahead and purchased a month of the Pro plan to see if I could convert my AI systems or simply use both.
Well, the more I used Claude, the more I came to like the research capabilities it offers. For deep research and reasoning, especially in extra and max modes, Claude provided better data for me. In fact, it provided a lot more compared to Perplexity Deep Research with the "Extra" and "Max effort levels. However, it is important to note that using the Max effort levels in Claude can cost you more tokens over time.
I enjoy the Cowork feature, which I use as a workspace to organize files, check documents and spreadsheets, and use across browsers. It works better with my local files and is a good assistant for managing certain tasks in my workflow. However, in my opinion, Claude can research well, but Perplexity still shines in workflows on Windows and Mac.
I love Perplexity, but Claude creeps in constantly, helping my workflow
Overall, I still turn to Perplexity Deep Research 70 percent of the time when I start my workflow. However, Claude has been creeping in more and more. When using the "extra" or "max" effort levels, Claude took longer than Perplexity's Deep Research functionality, but it gives me better information I can use in my workflow. I will keep experimenting, but the more I use Claude, the more it becomes part of my daily workflow. It's a tool I just can't seem to keep away while I am working. Which AI assistant tool I use depends on my workflow at the time.
