Tools and Workflows That Actually Work

Let me start with a confession: I have, at various points in my life, been a devoted user of Google Reader, Instapaper, Pocket, Evernote, OneNote, Notion, Obsidian, Roam Research, Logseq, Bear, Apple Notes, and at least four apps I’ve forgotten the names of. I’ve built workflows in IFTTT, Zapier, Make, n8n, and hand-rolled Python scripts. I’ve tried Getting Things Done, Building a Second Brain, the Zettelkasten method, and something called “PARA” that I still can’t fully explain.

Each of these tools and systems worked. For a while. Then my needs changed, or the tool changed, or I changed, and I moved on.

The lesson isn’t that tools don’t matter. They do. A good tool reduces friction; a bad tool creates it. The lesson is that tools are the most ephemeral layer of your information architecture. The specific app you use to read RSS today will be different from the one you use in five years. The patterns — why you use RSS, how you triage, what you save — those persist.

So this chapter will talk about both: specific tools that work well today, and the enduring patterns behind them. When the tools change, the patterns will still be useful. When the patterns feel abstract, the tools will make them concrete.

RSS Readers: The Cockroach of Information Tools

RSS has been declared dead roughly once a year since Google killed Google Reader in 2013. It is still, thirteen years later, the best mechanism for following serial content from sources you’ve chosen.

Why RSS endures:

You control the source list. No algorithm decides what you see. You subscribed to a feed; you get that feed. This is the killer feature. In a world of algorithmic curation, RSS is the last bastion of “I asked for this and only this.”

It aggregates without editorializing. Your RSS reader shows you what was published. It doesn’t reorder by engagement, insert sponsored content, or bury posts from accounts you “don’t interact with enough.”

It’s batched by design. RSS is a queue, not a stream. You check it when you want to, process what’s there, and close it. There’s no infinite scroll, no pull-to-refresh dopamine loop, no notification that new content has arrived since you last looked. (Unless you configure those things, in which case stop it.)

It works with almost everything. Blogs, news sites, podcasts, YouTube channels, Reddit subreddits, GitHub releases, academic journals — almost anything that publishes regularly has an RSS feed, even if they don’t advertise it.

Current RSS readers worth considering:

Feedly is the closest thing to a mainstream RSS reader. It works, it’s reliable, it has AI features for power users. The free tier handles basic use; the paid tiers add AI summaries, keyword tracking, and team sharing.

Inoreader is the power user’s choice. It has rules, filters, and automation capabilities that let you build sophisticated triage workflows. If you want RSS as infrastructure rather than just a reading app, Inoreader is worth the subscription.

Miniflux is for the self-hosted crowd. It’s minimal, fast, and runs on your own server. If you care about owning your data and don’t mind running Docker containers, it’s excellent.

NetNewsWire is free, open-source, and Mac/iOS native. It does one thing well: present RSS feeds in a clean interface. No AI, no social features, no nonsense.

Feedbin is a clean, well-maintained paid service that works well with various front-end apps. It supports newsletters as feeds, which is increasingly useful.

The pattern that matters more than the tool: RSS is pull-based intake. You choose when to engage. You choose the sources. You process at your own pace. Whatever specific reader you use, these properties are what make RSS valuable.

Setting Up RSS Effectively

A common failure mode with RSS is subscribing to too many feeds and ending up with an unmanageable queue. This defeats the purpose. Here’s how to avoid it:

Start with ten feeds. Not fifty. Ten. Pick the ten sources most important to your professional and intellectual life. Use those for a month. Add more only when you’ve established a comfortable processing rhythm.

Organize feeds by triage priority. Most RSS readers support folders or categories. Create two or three groups: “read everything” (your highest-signal sources), “scan headlines” (good sources where you don’t need every post), and “check weekly” (slower-cadence sources that don’t need daily attention).

Set an unread-count ceiling. If your unread count exceeds some threshold — say, 100 — mark everything as read and start fresh. This sounds brutal, and it is. It’s also necessary. An RSS queue that’s fallen behind is a guilt generator, not an information tool. Reset it. You won’t miss anything that matters — if something was important, you’ll hear about it from another source.

Prune quarterly. Every three months, look at your feed list. Any feed you consistently skip? Unsubscribe. Any feed that’s 90% noise? Unsubscribe. Any feed that’s gone dormant? Unsubscribe. Your feed list should be a living thing, not a monument to your past interests.

Read-Later Services: Managing the Queue of Good Intentions

Read-later services — Pocket, Instapaper, Omnivore (RIP), Readwise Reader, Matter — solve a real problem: you encounter something interesting at a moment when you can’t read it carefully. You save it for later.

The problem is that “later” often means “never.” The average read-later queue is a growing backlog of articles that seemed fascinating at the moment of saving and now feel like homework.

Here’s how to use read-later services without creating a guilt pile:

Treat your read-later queue as a buffer, not an archive. Items should flow through it, not accumulate in it. If something has been in your queue for more than two weeks, either read it now or delete it. Its moment has passed.

Have a designated read-later time. Not “whenever I get a chance.” A specific, recurring time. Saturday morning with coffee. Tuesday lunch. The commute home. A time when you actually open the app and read.

Limit your saves. If you’re saving more than you’re reading, you’re using the service as a guilt-generation engine. Restrict yourself to saving five items per day, or ten per week, or whatever number results in you actually processing them.

Use the reading time for reading, not for triage. When you open your read-later app, don’t start scanning and reorganizing. Pick the first item and read it. Or delete it. But don’t spend your reading time managing your reading queue. That’s a trap that feels productive but isn’t.

Practice the immediate delete. When you open a saved article and your first reaction is “meh,” delete it. Don’t read it out of obligation. Don’t save it for even later. Delete it and move on. The momentary discomfort of letting it go is vastly preferable to the ongoing low-grade guilt of an ever-growing backlog.

Current tools worth considering:

Readwise Reader is the most full-featured option right now. It combines read-later functionality with highlighting, annotation, and a feed reader. It’s expensive relative to alternatives, but if you’re willing to pay for one reading tool, it does a lot.

Pocket is free (with a premium tier) and well-integrated with Firefox. It’s simple and reliable. It doesn’t try to be a knowledge management system, which is either a limitation or a feature depending on your perspective.

Your browser’s reading list. Safari, Chrome, and Firefox all have built-in reading lists. They’re not powerful, but they’re zero-friction. For casual use, they’re often sufficient.

A notes app. Seriously. If your read-later volume is low, just paste links into a note. No special tool needed. Don’t over-engineer this.

Note-Taking Tools: The Practice, Not the App

I’m going to say this once, clearly, and then move on: the choice of note-taking app is one of the least important decisions in your information architecture.

What matters is that you write. Regularly. About what you’re reading, thinking, and learning. The act of writing is the practice that makes information stick. The app is just where the writing lives.

That said, some apps are better than others at not getting in your way. Here’s what to look for:

Speed. How fast can you go from “I have a thought” to “I’m writing it down”? If the answer involves launching an app, waiting for sync, navigating to the right notebook, and creating a new note with the right template, that’s too slow. You need capture in seconds.

Search. Your notes are only useful if you can find them. Full-text search is the minimum. If you’re writing a lot, search quality becomes crucial.

Durability. Your notes should outlast the app. This means either using an app that stores files in an open format (plain text, Markdown) or having confidence that you can export everything if the app dies.

Appropriate structure. Some people think in hierarchies (folders within folders). Some people think in networks (links between notes). Some people think in sequences (everything in one long document). Use whatever matches your brain.

Current tools:

Obsidian stores notes as local Markdown files. It’s fast, extensible, and your data is always yours. The learning curve is moderate, and the plugin ecosystem can be overwhelming. Resist the urge to install thirty plugins in your first week.

Apple Notes is underrated. It’s fast, it’s everywhere in the Apple ecosystem, it syncs invisibly, and it has surprisingly good search. If you’re all-in on Apple and don’t need advanced features, it’s genuinely excellent.

Notion is powerful and flexible to the point of being daunting. It’s great if you need databases and structured data alongside notes. It’s overkill if you just need to write things down. Its offline support has historically been weak, which is a dealbreaker for some.

Plain text files in a folder. I’m serious. A folder of .txt or .md files, synced via your cloud service of choice, searchable via your operating system’s search. It’s zero-overhead, infinitely portable, and will outlast every app on this list. It’s not sexy. It works.

A paper notebook. Also serious. For capture and thinking, paper has zero boot time, no notifications, and a tactile quality that some people find essential for deep thought. The retrieval layer is terrible (you can’t search handwriting easily), but for the “write to think” use case, paper is still viable.

LLM Integration Points

AI tools are everywhere in the information workflow, but they’re not equally useful at every point. Here’s where they help, where they’re neutral, and where they add friction.

Where AI Tools Help

Summarization of long content. This is the most straightforward and reliable use case. Give an LLM a long article, report, or paper, and ask for a summary. The quality is generally good enough for triage purposes. It lets you make informed decisions about what deserves your full attention.

Translation and cross-language content. If you’re trying to read sources in languages you don’t speak fluently (and you should be — see Chapter 20), LLMs are dramatically better than older translation tools. They capture nuance and context that phrase-by-phrase translation misses.

Extracting structure from unstructured content. Ask an LLM to pull out the key claims, evidence, and conclusions from an article. Ask it to identify the assumptions. Ask it to list the sources cited. This kind of structural extraction is tedious for humans and easy for AI.

Generating questions. After reading something, ask an LLM: “What questions should I be asking about this?” or “What are the strongest counterarguments to this position?” This uses AI as a thinking partner rather than a thinking replacement.

Connecting to prior knowledge. “Based on this article and these notes I took last month, what connections do you see?” This is where LLMs’ ability to process large amounts of text becomes genuinely useful — they can hold more context in working memory than you can.

Where AI Tools Are Neutral

Categorization and tagging. AI can suggest tags for notes and articles, and it’s usually fine at it. But manual tagging is also fine and takes about two seconds. The AI isn’t adding much value here unless your volume is very high.

Feed curation. Some tools use AI to recommend RSS feeds or newsletter subscriptions. The recommendations are okay but not better than asking a knowledgeable colleague or searching a community forum for recommended sources.

Reading prioritization. AI can score articles by predicted relevance to your interests. In practice, this works about as well as scanning headlines yourself, which you’re going to do anyway.

Where AI Tools Add Friction

Capture. If capturing a thought requires you to open an AI tool, craft a prompt, wait for a response, and then edit the result, you’ve lost the thought. Capture needs to be instant. AI is too slow for the capture step.

Deep reading. Inserting AI into the process of reading something carefully — asking for explanations, summaries, context — can fragment your attention and prevent the deep engagement that makes reading valuable. There’s a time for AI-assisted analysis, but it’s after you’ve read something, not during.

Simple decisions. “Should I read this?” is a decision you can make in three seconds by scanning the headline and first paragraph. Routing it through an AI relevance assessment adds latency and complexity for no benefit.

Everything, all the time. The biggest friction AI can add is the friction of being one more thing to manage. If you’re using AI at every step of your information workflow, you’re spending more time orchestrating AI than you’re spending on actual information processing. Use AI at specific, defined integration points. Not everywhere.

Email Management Workflows

Email is an information source whether you want it to be or not. Here are workflows that keep it manageable.

The Two-Minute Rule. If an email takes less than two minutes to process (read, respond, archive), do it immediately. Don’t create a task to respond to a one-sentence email. Just respond.

The Triage Rhythm. Check email at defined intervals, not continuously. Three times a day is enough for most people. Once in the morning, once after lunch, once before end of day. Between checks, email is closed. Actually closed. Not minimized. Not “just checking.”

The Inbox Zero Philosophy (Adapted). Inbox zero doesn’t mean you’ve handled everything. It means everything has been triaged: responded to, delegated, scheduled for later, or archived. Your inbox is a triage area, not a storage system. Process it and clear it.

Newsletter Management. If you subscribe to more than five newsletters, they’re drowning your inbox. Options: use a separate email address for newsletters, use an RSS reader that supports newsletters (Feedbin, Readwise Reader), or use a service like kill-the-newsletter.com to convert them to feeds.

The Unsubscribe Audit. Once a month, search your email for “unsubscribe.” Look at everything you’ve received that contains that word. For each one, ask: “Did I read this? Did I want to?” Unsubscribe from everything you didn’t read or didn’t want.

AI Email Processing. LLMs can help with email triage: summarizing long threads, drafting responses, categorizing by urgency. This works well for high-volume email situations. For normal email volume, the overhead of routing emails through AI isn’t worth it.

Social Media as a Deliberate Information Source

Social media is both one of the best and worst information sources available. It’s one of the best because it provides real-time, diverse, expert-populated information networks that don’t exist anywhere else. It’s one of the worst because those networks are embedded in platforms designed to be addictive, rage-inducing, and attention-fragmenting.

The key is to use social media deliberately, not habitually.

Curated lists, not algorithmic feeds. Every major platform supports lists or equivalent features. Create lists of people you specifically want to follow, organized by topic. Read those lists. Ignore the main feed. The main feed is the platform’s agenda. Your lists are yours.

Time-bounded access. Set a timer when you open a social media app for information purposes. Fifteen minutes is plenty for a scan. When the timer goes off, close the app. Without a timer, fifteen minutes becomes forty-five without you noticing. This is by design. Counteract the design.

Separate accounts or profiles for different purposes. Some people maintain a “professional information” social media presence that follows only domain experts and news sources, separate from their personal social media. This sounds excessive until you try it, at which point it feels obvious.

Social media as a discovery tool, not a consumption tool. Use social media to find links, papers, threads, and references. Then read those things outside of social media, in your read-later app or RSS reader. Social media is good for surfacing what exists. It’s terrible for focused reading.

Periodic purges. Every month, review who you follow. Unfollow accounts that have shifted from informative to performative. Unfollow accounts that mainly produce emotional reactions rather than useful information. Unfollow accounts that post more than you can reasonably process. Your follow list is an intake design decision. Treat it like one.

Podcast and Video Triage

Podcasts and video content present a unique triage challenge: they’re linear and time-consuming. An article can be skimmed in thirty seconds; a podcast episode takes an hour. This makes triage strategies especially important for audio and video content.

Speed controls are not optional. Most people can comprehend speech at 1.5x to 2x speed once they’ve acclimated. This isn’t about rushing through content. It’s about matching playback speed to your comprehension speed. Normal podcast speech is slower than most people can process. Speed it up.

Transcripts change the game. Many podcast apps now provide transcripts (Apple Podcasts, Spotify, specialized apps). Transcripts let you skim a podcast episode the way you’d skim an article: scan for relevant sections, read those closely, skip the rest. This is the single biggest improvement in podcast triage in the last decade.

AI summaries for triage. Services like summarize.tech or built-in features in some podcast apps provide AI summaries of episodes. Use these the way you’d use article summaries: for deciding whether to listen, not as a substitute for listening.

The sample-and-commit approach. Listen to the first five minutes of an episode. If it hooks you, keep listening. If it doesn’t, skip it. Don’t force yourself through a mediocre podcast episode out of some sense of obligation to the medium.

YouTube at 2x with chapters. YouTube’s chapter markers (when creators use them) let you navigate directly to relevant segments. Combined with 2x speed, you can extract the valuable content from a thirty-minute video in five minutes. AI-generated video summaries can help you identify which segments to watch.

Subscribe selectively, sample widely. Have a core list of podcasts you listen to regularly (no more than five or six if they’re weekly). Beyond that core, sample episodes from other podcasts based on guest, topic, or recommendation. You don’t need to subscribe to every podcast that has one good episode.

Building Automated Pipelines

Automation is seductive. The idea that information can flow through a carefully constructed pipeline, arriving at your desk pre-sorted, pre-summarized, and pre-categorized, is deeply appealing to anyone who’s ever stared at an overflowing inbox. And unlike many seductive ideas, this one partially delivers on its promise.

But automation can also become its own kind of burden. Every automated pipeline is a system that can break, needs maintenance, and adds a layer of abstraction between you and the raw information. The key is to automate the tedious parts (collection, formatting, delivery) while keeping the important parts (evaluation, judgment, decision-making) firmly manual.

Here’s how to automate wisely.

The RSS-to-Summary-to-Digest Pipeline

This is the single most useful automation for information triage:

1. Your RSS feeds deliver new content to an aggregator
2. An automated process sends article content to an LLM for summarization
3. Summaries are compiled into a daily or weekly digest
4. The digest is delivered to you (email, note, message — wherever you’ll actually read it)

You can build this with tools like n8n, Make, or custom scripts. Some RSS readers (Feedly, Inoreader) have built-in AI summarization. The specific implementation matters less than the pattern: automated intake, automated summarization, human triage of summaries.

Email Newsletter to RSS Pipeline

If you subscribe to newsletters but prefer to read them in your RSS reader, tools exist to convert email newsletters to RSS feeds. Feedbin does this natively. Kill-the-newsletter.com is a free standalone option. This lets you consolidate all your serial content into one interface.

Highlight and Annotation Pipelines

Services like Readwise sync highlights from various sources (Kindle, web articles, podcasts) into a single location and can pipe them to your note-taking system. This is genuinely useful if you’re a heavy highlighter — it means your highlights end up somewhere searchable rather than trapped in individual apps.

The Monitoring Pipeline

If you need to track specific topics, companies, or people, a monitoring pipeline watches for mentions and delivers them to you:

1. Set up Google Alerts, Talkwalker Alerts, or equivalent for your keywords
2. Route alerts to a dedicated email folder or RSS feed
3. Optionally, pass through an AI filter to reduce false positives
4. Review the filtered results daily or weekly

This is particularly useful for professionals who need to know when their company, their competitors, or their research topics are in the news. The key is specificity in your keywords — broad terms generate noise; specific phrases generate signal.

The Weekly Digest Pipeline

For lower-priority information streams, a weekly digest compresses an entire week’s worth of content into a single reading session:

1. Accumulate content throughout the week (via RSS, alerts, newsletters)
2. On a set day, run the accumulated content through an AI summarizer
3. Generate a single digest document with summaries, organized by topic
4. Read the digest in one focused session

This works well for domains you want to track without monitoring daily. It’s the information equivalent of batch processing — less responsive, but more efficient.

Automation Maintenance

Every pipeline you build needs periodic attention. Here’s a quarterly maintenance checklist:

• Are all pipelines still running? (Automated things fail silently more often than you’d think.)
• Is the output still useful? (Your needs evolve; your automations should too.)
• Are any pipelines producing output you consistently ignore? (Delete them.)
• Are there new automation opportunities based on recent workflow pain points?
• Is the total overhead of maintaining your automations less than the time they save? (If not, simplify.)

The last point is crucial. I’ve seen people spend more time debugging their automation setups than they would have spent doing the task manually. Automation should save time in aggregate, not just shift the time cost from “doing the task” to “maintaining the system that does the task.”

What Not to Automate

Triage decisions. Automating what gets your attention is tempting but dangerous. Every automated filter is a potential filter bubble. Automate the delivery of information, not the evaluation of it.

Note-taking. Automated notes from articles are clipboards, not thinking. The value of a note is that you wrote it. Automate the capture of raw material; keep the synthesis manual.

The whole thing. If your information workflow has more automated steps than manual ones, you’ve built a Rube Goldberg machine. Automation should remove friction from specific pain points, not replace the entire process of being a human who reads things.

Relationship-dependent communication. Don’t auto-respond to emails, auto-schedule meetings, or auto-delegate tasks unless you’re very sure the automation won’t damage relationships. People can tell when they’re interacting with a system instead of a person, and they don’t love it.

The Weekly Review

The weekly review is the most important habit in your information workflow, and the one most likely to be skipped. It’s the checkpoint where you ensure your system is actually working.

A weekly review takes 20-30 minutes and covers:

What did I read this week that mattered? Scan your notes from the past week. Identify the two or three most valuable things you encountered. This reinforces good intake choices and helps you remember what you learned.

Is my read-later queue growing or shrinking? If it’s growing, you’re saving more than you’re reading. Either save less or allocate more reading time. If it’s growing fast, you have an intake problem — too many sources, or sources that are too prolific.

Did I miss anything important? Check with colleagues, check industry news, check social media briefly. If you missed something big, figure out why. Was it outside your source coverage? Was it in your feeds but you skimmed past it? Adjust accordingly.

What should I prune? Identify one source that didn’t earn its place this week. Not one that was bad — just one that was consistently mediocre. Unsubscribe or move it to a lower-priority tier. This gradual pruning keeps your intake lean over time.

What am I curious about? End the review by noting one or two things you’d like to explore next week. This isn’t homework. It’s a prompt for your curiosity. If you follow up, great. If you don’t, that’s fine too. The act of noting curiosities keeps your information intake from becoming purely reactive.

The Monthly Source Audit

Once a month, take a broader look at your information sources. The weekly review handles tactical adjustments. The monthly audit handles strategic ones.

Review your complete source list. Every feed, newsletter, podcast, and regular reading habit. For each one, ask: “Is this still serving the purpose I originally added it for?”

Check for drift. Sources change over time. A blog that used to publish insightful analysis might have shifted to hot takes. A newsletter that was focused might have expanded into topics you don’t care about. A podcast might have changed format or quality. Don’t keep following a source based on what it used to be.

Check for gaps. Look at what you’ve been searching for, asking colleagues about, or feeling uninformed about. These are signals that your source coverage has holes. Identify one or two new sources to try.

Check for redundancy. If multiple sources consistently cover the same topics with the same angles, you don’t need all of them. Keep the best, cut the rest.

Review your AI integration. If you’re using AI tools in your workflow, are they still helpful? Have you fallen into any AI-dependency patterns? Are there new integration points that would be useful? AI tools are evolving fast; your workflow should evolve with them.

Sample Daily Workflow

Here’s a concrete daily workflow. Adjust the times and durations to your life; the structure is what matters.

Morning Triage (20 minutes)

Minutes 0-5: Email scan. Open email. Process anything under two minutes. Flag anything that needs a longer response. Archive everything else. Close email.

Minutes 5-15: Feed scan. Open RSS reader. Scan headlines in your high-priority folder. Read anything essential (breaking news, urgent updates). Save two to three items for focused reading later. In your scan folder, skim headlines and save anything that catches your eye. Mark everything as read.

Minutes 15-20: Daily digest. If you have an automated summary pipeline, review the digest. Flag anything that needs full reading. Spend one minute on your social media lists — scan, don’t scroll.

Focused Reading (30 minutes)

This happens whenever works for you — lunch, afternoon, commute. The key is that it’s a defined block, not “whenever I get a chance.”

Read the items you flagged during morning triage. For each one: read it, write a brief note (even one sentence), then move on. If you finish your flagged items with time to spare, process your read-later queue.

Don’t save new items during focused reading time. This is processing time, not intake time. New items you discover while reading go to your read-later queue, which you’ll triage later. Otherwise, reading time becomes an ever-expanding intake session.

Evening Exploration (15 minutes)

This is optional but valuable. It’s explicitly not triage and not catch-up. It’s exploration.

Read something outside your usual domains. Browse a source you don’t normally follow. Read a long-form piece on a topic you know nothing about. Follow a link chain from one interesting thing to another. Take notes if something sparks your thinking. Don’t take notes if it doesn’t.

This is your “curiosity time.” Protect it from the temptation to use it for more triage. The world will not end if you don’t process those three remaining items in your RSS reader. It might get slightly more interesting if you read about mycorrhizal networks or Mongolian throat singing or whatever catches your fancy.

Sample Weekly Workflow

Sunday or Monday: Source Review (15 minutes)

• Review the past week’s reading notes. Star the best two or three things you read.
• Check your read-later queue size. If it’s over 20 items, spend 10 minutes doing a rapid triage: keep five, delete the rest.
• Identify one source to prune and one source to try.

Wednesday: Backlog Processing (30 minutes)

• Process anything that’s been sitting in a queue for more than a week.
• The rule: read it in full, or delete it. No re-saving. No “I’ll get to it next week.”
• Write notes on anything you read. Even briefly.

Friday: Curiosity Time (30 minutes)

• Follow up on the curiosity prompt from your last weekly review.
• Explore a topic, a source, a rabbit hole that’s purely interesting to you.
• No obligation to be productive. No requirement to take notes (though you can).
• This is the most easily skipped part of the weekly workflow and one of the most important. It’s what keeps your information life from feeling like an obligation.

Troubleshooting Common Workflow Problems

Even good workflows break down. Here are the most common problems and their fixes.

“I never get to my focused reading time”

This usually means your triage is expanding to fill all available time. The fix: set a hard stop on triage. When the timer goes off, you stop triaging, even if there are unprocessed items. The unprocessed items will still be there tomorrow. Your focused reading time won’t be, because you just spent it on triage.

Alternatively, your focused reading time might not be scheduled concretely enough. “I’ll read during lunch” isn’t a schedule. “12:15 to 12:45, reading the items I flagged this morning” is a schedule. The specificity makes it harder to skip.

“My read-later queue is a graveyard”

You’re saving too much and reading too little. Either reduce your save rate (be more selective about what deserves saving) or increase your read rate (allocate more time to processing the queue). Or, most likely, both.

The nuclear option: declare read-later bankruptcy. Delete everything in the queue. Start fresh. If something was truly important, it’ll come around again. The psychological relief of an empty queue is worth the theoretical cost of lost articles.

“I can’t maintain the weekly review”

Simplify it. If your weekly review takes thirty minutes, cut it to fifteen. If fifteen is too much, cut it to five: “What was the most valuable thing I read this week? Is my queue growing or shrinking?” Five minutes, two questions. That’s better than nothing, and nothing is what you’re currently doing.

Also consider whether your review day and time are realistic. Sunday evening reviews sound great until you remember that Sunday evenings are for dreading Monday. Pick a time when you actually have energy and willingness.

“I keep switching tools”

Impose a moratorium. Pick the tools you have right now and commit to using them for three months. No reading reviews. No trying alternatives. No “just exploring” new options. Three months of using what you have. If, at the end of three months, you have a specific, articulable problem that a different tool would solve, switch. If you just have vague tool-envy, that’s not a tool problem.

“My system works great for a while, then I fall off”

This is normal. Systems aren’t all-or-nothing. If you fall off for a week, start again. Don’t treat a lapse as a failure that invalidates the whole system. Treat it as a week off. Resume where you left off. Lower the bar if needed — do half the routine rather than none of it.

The people who maintain information systems long-term aren’t the ones who never lapse. They’re the ones who resume after lapsing.

The Anti-Workflow: When Structure Hurts

I’ve spent this chapter advocating for structured workflows, and I stand by that. But I should note: some people, in some phases of their lives, need less structure, not more.

If you’re in a period of creative exploration, rigid triage workflows can kill the serendipity that exploration requires. If you’re recovering from burnout, adding another system to maintain is counterproductive. If you’re genuinely satisfied with your current relationship to information and don’t feel overwhelmed or underleveraged, maybe you don’t need a workflow at all.

The workflows in this chapter are tools. Tools are for people who have a problem the tool solves. If you don’t have the problem, you don’t need the tool. Don’t build a productivity system because the internet told you to. Build one because you have a specific, felt need for one.

And even within a structured workflow, leave room for unstructured exploration. Not everything needs to be triaged, categorized, and processed. Sometimes the best information experience is falling down a Wikipedia rabbit hole at midnight with no purpose whatsoever. That’s not a system failure. That’s being a curious human. Protect it.

Why the Best System Is the One You Actually Use

I’ve given you a lot of specific advice in this chapter. RSS readers and read-later services and daily workflows with minute-by-minute breakdowns. Here’s the caveat: none of this matters if you don’t do it.

The productivity internet is full of elaborate systems designed by people who enjoy designing elaborate systems. There’s nothing wrong with that as a hobby, but there’s everything wrong with mistaking system design for system use.

The best system is the one you actually use. Consistently. Over months and years. This means:

Optimize for consistency over perfection. A simple system you use every day beats a sophisticated system you use when you’re motivated. And you are not always going to be motivated.

Start smaller than you think you need to. You can always add complexity. You can never un-abandon a system that was too complex to maintain. Start with RSS and a note-taking app. That’s it. Add tools and workflows as specific needs emerge.

Give new tools a fair trial. Don’t switch tools every time you see a compelling demo. But do try new tools occasionally, with a defined trial period. Two weeks is usually enough to know whether a tool fits your workflow.

Accept imperfection. You will miss articles. You will fall behind on your read-later queue. You will skip weekly reviews. You will have weeks where your system goes unused entirely. This is not failure. This is being a human with a job and a life and competing demands on your attention. The system is there when you come back to it. That’s enough.

Don’t confuse the system with the goal. The goal is to be well-informed, to think clearly, and to make good decisions. The system is just scaffolding. If you find yourself spending more time maintaining the system than consuming and processing information, the scaffolding has become the building. Step back and simplify.

You don’t need the perfect workflow. You need a workflow. The one you’ll use tomorrow morning when you’re groggy and your RSS reader has 47 unread items and you’re tempted to just open Twitter instead. Make it easy enough for that version of you to follow through.

That’s the real test of a system that works.