Perplexity

The Research Workflow That Replaced My Old Browser Tabs Habit

May 2026

By Mark Reeves — author of all 47 AI Field Guides. Written and tested in a working business.

AI-generated portrait. Pen name. Real work.

Twenty open tabs was never a research system. Here's what replaced it.

By Mark Reeves — AI-generated portrait. Pen name. Real work.

The tab problem isn't about tabs

Twenty-two open tabs sounds like a productivity problem. It isn't. It's a thinking problem in disguise.

Each tab represents something you found but haven't processed. A piece of information you're holding in reserve because you don't know yet what to do with it. The browser becomes an external hard drive for half-finished thoughts.

The actual cost isn't the clutter. It's what you're doing instead of thinking: managing. Scrolling back to re-find the tab you need. Trying to remember why you opened the one about the 2023 market report. Checking whether the article you saved contradicts the one from last week. That's not research. That's overhead.

Real research is: get oriented, drill into specifics, push on what you found, capture what matters, move on. Browser tabs make that sequence almost impossible because they're structured around saving links, not thinking through problems.

The moment the habit started to break

You're a consultant. Your client asked you to evaluate three CRMs before Thursday. You've got twelve tabs open. It's been 45 minutes. You have impressions, not understanding.

That's the version of this I lived through. I used to research software purchasing decisions the same way. Open a tab for the category overview. Open three for comparison articles. Open two for Reddit threads with real user opinions. Open the pricing pages for the two or three tools I was actually considering.

Then I'd read all of it, badly, because reading six long pages of mixed quality simultaneously is not reading. It's scanning. And scanning produces impressions, not understanding.

The first time I tried running this in Perplexity instead, the experience was jarring in a specific way: I had an answer in four minutes. Not a reading list — an actual synthesized answer with the sources attached so I could verify it.

The question was something like: What are the main criticisms of [software category] tools among small business users, and what do the alternatives get right that these tools miss?

Perplexity pulled from review sites, Reddit, tech publications, and a couple of specialist blogs. It gave me a coherent summary with numbered citations. I clicked three of them to verify the specific claims that mattered. Done.

I had not looked at a single tab. I had not managed anything.

The workflow in practice

The approach I use now has three moves.

Start with an orientation query. Keep it broad. You're not trying to get a final answer yet. You're trying to understand what the question actually involves.

Something like: Give me an overview of [topic] — key concepts, main players, and what the current debate is about. This collapses an hour of backgrounding into five minutes. You find out what you don't know, which is more useful than finding out what you already suspected.

Drill with specific questions. Once you're oriented, you know what to push on. This is where Perplexity earns its keep. Ask the specific question you actually have, not the general version of it.

Instead of: What are the best CRMs?
Try: What CRM options work well for service businesses with fewer than ten people, where the main use case is tracking client relationships rather than a sales pipeline? Focus on pricing, ease of setup, and what users say doesn't work.

The specificity changes the answer quality dramatically. Perplexity is better at synthesis than at navigation, and the more precisely you tell it what you want synthesized, the better the result.

Use follow-up queries to test the answer. Don't just accept what the first answer gives you. Ask a follow-up that challenges it.

What are the strongest arguments against [conclusion from prior answer]? or What do critics of this approach say? This is the step most people skip, and it's where you find the caveats and edge cases that would have bitten you later. A research session that only confirms your existing inclination isn't research.

Moving research into a working document

Perplexity is where you find and synthesize. It is not where you write or organize.

When I finish a Perplexity session, I do one pass to capture the things that actually matter. Not everything. The temptation is to paste the whole answer into a doc. Resist it. You'll just recreate the tab problem in document form.

What to capture:

• The key conclusions, in your own words. Forcing yourself to restate it means you actually understood it.
• The specific citations that back up any claim you'll act on or repeat. Not the whole source — just the title, URL, and the one fact it's vouching for.
• The open questions the research raised. If you ran a good follow-up query, there will be some. These are often more valuable than the answers.

What to leave behind: the full Perplexity output, any source you clicked and found didn't add anything, and any tangent that was interesting but not relevant to the actual decision.

The working document should be shorter than what Perplexity gave you. If it isn't, you're compiling, not thinking.

Handing off to Claude or ChatGPT

Perplexity is a research tool. It is not a writing or analysis tool.

Once I have a working document with research notes, I move to Claude or ChatGPT for the next step — whether that's drafting something, analyzing what I found, building an argument, or structuring a recommendation.

The handoff is simple. Paste your research notes at the top of the conversation and give the task. Something like: Here's what I found researching [topic]. Based on this, I need you to [write / analyze / structure / draft]. The notes give the model something accurate to work from rather than relying on its training data, which may be outdated on specifics.

This is the division of labor that actually works: Perplexity for what's true now, Claude or ChatGPT for what to do with it. Trying to do both in one tool produces mediocre results from both.

When the tab habit is still right

I've been making the case for replacing tabs with Perplexity. But there are tasks where browsing manually is still better.

When you need a specific document you can already name. If you're looking for a company's terms of service, a specific regulatory filing, a press release from a known date — Google and direct navigation are faster. Perplexity is for synthesis, not retrieval.

When you're doing primary source research. Academic papers, government datasets, original reports. You need to read those yourself. Perplexity can find and summarize them, but for anything you'll act on professionally, you need to verify the primary source, not a synthesis of it.

When you're doing deep competitive analysis on a live business. Perplexity is good at synthesizing existing public information. It's less good at catching the nuance in a competitor's website copy or the specific language in their pricing page. For that kind of close reading, you need to be in the tab looking at it.

When the topic is moving fast. Perplexity has a real-time search capability, but for news events in the last 24-48 hours, Google's indexing is more current. If you're tracking something that's actively developing, start with Google and move to Perplexity once the initial wave of coverage settles.

Perplexity replaces tabs for synthesis tasks. It does not replace tabs entirely. Outside those cases, every time you open twelve tabs for a synthesis task, you're doing the research pipeline manually — time Perplexity handles in under five minutes.

Run the Three-Move Workflow Once Before You Move On

Take the next research question you have — a real one, not a test. Run the three-move sequence: orientation query first, specific drill-down second, a follow-up that challenges what the first answer gave you.

Paste the query below into Perplexity to start:

Give me an overview of [your topic] — key concepts, main players, and what the current debate or disagreement is about.

Then follow up: What are the strongest arguments against the most common view on this?

That two-query loop is the whole method. If you've got twenty tabs open right now, close them. Run those two queries instead.

This article covers the three-move research workflow. It doesn't cover two things that determine how well it works in practice:

How to use Perplexity's focus modes to control what sources it retrieves. The three-move workflow defaults to Web mode. For medical, academic, or specialized research, the mode choice matters more than the query structure.

When to trust the synthesis vs. when to verify the source. Not every citation in a Perplexity result needs the click. The guide covers the triage logic for deciding which claims are load-bearing enough to verify.

Related reading

• Perplexity vs Google for business research
• Perplexity focus modes explained

Mark Reeves is a pen name. AI Field Guide publishes role-specific, practical guides for using AI tools in real work.