AI 101Beginner17 min read

Keep up with AI without burning out: a 20-minute weekly routine

Deciding what counts as news for your work, sources by trust level, a minute-by-minute weekly routine, testing new tools safely, a five-question hype filter, a dated log, and rules that prevent burnout.

Keep up with AI without burning out: a 20-minute weekly routine

Keep Up With AI Without Burning Out: A 20-Minute Weekly Routine

There is a new model, a new feature or a new "everything has changed" thread every day. If you try to follow all of it, you will spend your evenings reading and your days feeling behind. If you follow none of it, you will miss the changes that matter to your work, such as a price change, a privacy setting or a feature that makes a job ten times easier. This tutorial gives you a middle path: a small, repeatable weekly routine of about twenty minutes, a short list of sources you trust and a few rules for ignoring the rest. You will also learn how to test a new tool safely, how to spot hype in seconds and how to keep a simple dated log, so you always know what you know and when you last checked.

Difficulty: ★☆☆☆☆ (no technical background needed)
Tools needed: A notes app and a calendar
Last updated: November 2026

Overview

Nobody can follow AI news in full. The volume is too high, much of it repeats and a lot of it is written to attract clicks. Yet the people who seem to keep up effortlessly are usually not reading more than you. They are reading less, and reading differently. They have decided what they need to know, they use a few good sources and they have a habit, not a feeling of obligation.

The goal of this tutorial is not to make you an expert on everything. It is to make you reliably good enough: aware of the changes that affect your own work, able to try new things without risk and calm about the rest.

The routine rests on four ideas.

1. Decide what you need to know. Your jobs decide what counts as news. A new image model matters a lot to a designer and little to an accountant.
2. Use sources by trust level. Official pages first, a couple of good curators second, and almost nothing from the viral feed.
3. Make it small and regular. Twenty minutes a week, at a fixed time, beats a binge once a month. A habit you can keep is worth more than a perfect one you drop.
4. Write it down. A dated log turns a blur of headlines into knowledge you can rely on, and tells you what has gone stale.

The tutorial has seven steps. First, what you actually need to know. Second, a short list of sources and how much to trust each. Third, the twenty-minute weekly routine itself, minute by minute. Fourth, how to test something new safely. Fifth, how to filter out hype in seconds. Sixth, how to keep a dated log. And seventh, the rules that stop you burning out.

By the end you will have a one-page routine, a short list of sources and a note-taking habit you can start this week.

Who This Is Useful For

  • Busy people who feel they should be following AI but cannot.
  • Professionals who need to stay current enough to advise or decide.
  • Anyone who has been overwhelmed by the pace of AI news and wants a calmer way in.
  • What You Will Learn

  • How to decide what counts as "news" for your own work.
  • How to pick sources by trust level and what to ignore.
  • A twenty-minute weekly routine you can keep.
  • How to test a new tool or feature safely, in ten minutes.
  • A five-question filter for spotting hype.
  • How to keep a dated log, and rules that prevent burnout.
  • What You Need

    1. A notes app, or a document, for the log.
    2. A calendar, for a fixed weekly slot.
    3. The two or three AI tools you use most.
    4. Thirty minutes to set up, then twenty minutes a week.

    Information as of 14 November 2026. The sources named here are kinds of source, such as a company's official release notes or a regulator's page, rather than a fixed list. Which specific newsletters and creators are worth following changes, and you should choose your own. Do your own judgement about any source, including ones recommended in this tutorial.

    Step 1 — Decide what you need to know

    The first filter is yourself. Without it, everything looks urgent.

    List your jobs. Write down the five or six things you use AI for, as in earlier tutorials: drafting, researching, summarising, making images, planning, analysing data. These are your "beats".

    Sort the news into three buckets. Every item you read goes in one of these.

    BucketMeaningWhat you do
    Act nowIt changes how I do a job I do every week, or it affects my money, privacy or legal positionTest it this week, and update my notes
    WatchIt might matter in the next few months, or it is related to my work but not urgentWrite one line in the log, and look again at the next quarterly review
    IgnoreInteresting, but unrelated to my work, or just noiseLet it go, without guilt

    The key habit is the third bucket. Most news belongs in "Ignore", and that is healthy. A new video model is genuinely important news and still irrelevant to your finances.

    Know your standing questions. A few questions are worth asking every week, whatever your work:

  • Has anything changed in the tools I use, such as features, plans, limits or privacy settings?
  • Has anything changed in a rule that affects my work, such as my employer's policy, my industry's guidance or a law?
  • Is there something new that would save me a lot of time on a job I do often?
  • If the answer to all three is no, you can stop. That is a good week.


    Step 2 — Sources, by trust level

    Not all sources deserve the same trust. Use them in this order.

    LevelKind of sourceExamples of the kindHow to treat it
    1. PrimaryThe maker's own pages and the rule-makers' pagesA tool's release notes, help centre and pricing page; a regulator's official page; a court recordThe source of truth for what changed. Remember the maker is also the seller, so benchmark claims are claims
    2. Careful curatorsA small number of people or newsletters who explain and add context, and who show their sourcesA newsletter or channel that links to primary sources and corrects mistakesUseful for context and for finding what you missed. Check the primary source for anything you will act on
    3. News and analysisReputable news outlets and independent researchersA newspaper's technology desk; a research group's reportGood for the bigger picture. Check the date, and whether it is reporting or opinion
    4. Community and socialForums and social postsDiscussion threads, short videos, posts from buildersA source of ideas and early signals. Treat as unverified until you find a primary source
    5. Viral and leaksThreads that claim "everything has changed", leaked-news accounts, and reaction videosAnything built to be sharedMostly noise. Skip

    Three rules for building your list.

    1. Choose two or three, not twenty. Pick the primary pages for the tools you use, one or two curators and one source for your region's rules. That is enough.
    2. Prefer sources that show their evidence. A source that links to the original, gives dates and corrects its mistakes is worth more than a louder one.
    3. Review the list every quarter. Drop anything that no longer earns its place, and add one only when you drop one.

    If you are not sure where to start, the three places that pay off most are: the release notes of the one tool you use the most, the pricing or plan page of that tool, and your employer's or industry's guidance on AI.

    Step 3 — The twenty-minute weekly routine

    Put a fixed slot in your calendar, the same time each week. Treat it like a meeting with yourself. A Friday afternoon or a Monday morning both work. Here is the routine, minute by minute.

    Minutes 1 to 5: Official changes.

  • Open the release notes or news page of your main tool or two. Read only what is new since last week.
  • Check one plan or privacy page if you suspect a change.
  • For each item, put it in a bucket: Act now, Watch or Ignore.
  • Minutes 6 to 10: One curator.

  • Read one good summary from your chosen curator, or skim a headline list. Do not open more than three links.
  • Anything that looks like "Act now" must be confirmed in a primary source before you rely on it.
  • Minutes 11 to 15: Test one thing.

  • Choose one "Act now" item, or one thing you have been curious about. Try it for five minutes on a harmless task, using the method in Step 4.
  • If nothing needs testing this week, use the time to improve one of your saved prompts instead.
  • Minutes 16 to 20: Update your notes and decide.

  • Add dated lines to your log: what changed, what you tested and what you decided.
  • Decide on one action for next week, or none. "Nothing to do" is a perfectly good outcome.
  • Close everything. You are done until next week.
  • A template for your notes:

    Week of ___
    
    Changes in my tools (official sources):
    -
    
    New things I tested (what, result, verdict):
    -
    
    Watch list (one line each, with source and date):
    -
    
    Rules or policies that changed:
    -
    
    One action for next week (or "none"):


    Step 4 — Test something new, safely

    New tools and features are tempting, and trying them is how you learn. Do it in a way that protects you.

    Use a harmless task. Test on something where a mistake costs nothing: a practice email, a made-up example, public information. Never first-try a new tool on a client's confidential document or a message that goes to a real person.

    Compare with what you already do. Take one task you do often, and run it the old way and the new way. Note three things.

    1. Is it better? Quality of the result, in your judgement.
    2. Is it faster or easier? Roughly how long it took each way.
    3. Is it riskier? What does it connect to, what does it store and what can it do without asking you?

    Check the settings before you trust it. Before you connect a new tool to your accounts or give it data, look at its privacy settings, what it stores and how to delete it. The privacy tutorial in this module gives you the five-minute check.

    Limit the blast radius. Start with read-only access and no connections. Give more only when you have a reason and have tested.

    Decide, and write it down. At the end, pick one:

  • Adopt: it is clearly better, safe enough and fits a job I do often. I will use it, with the checks.
  • Keep watching: interesting, not yet ready or not clearly better. I will look again in three months.
  • Drop: not for me. I will not spend more time.
  • Writing "drop" is allowed. A short test that ends in "no thanks" is a success.

    Step 5 — Spot hype in seconds

    You will meet hundreds of exaggerated claims a year. A quick filter saves hours. Ask five questions, in this order, and stop at the first one that fails.

    1. Who is saying it? A maker describing its own product, a neutral researcher, or an account that profits from attention? The seller is not lying by default, but the claim needs checking.

    2. What did they show? A demo, a benchmark, a real example or only adjectives? A claim with nothing shown is hype until proven otherwise.

    3. Can I check it? Is there a link to a primary source, a named test or something I can try myself? If a claim cannot be checked, treat it as unproven.

    4. What changed for me? Does it affect a job I do or a decision I make? If not, it is "Ignore", however impressive.

    5. What is the catch? Cost, limits, who it is available to, what it cannot do, what data it needs. Real announcements usually have caveats, and a post with none is suspicious.

    Some phrases are reliable warning signs, though none proves a claim is false:

  • "Changes everything", "the end of X", "the smartest ever".
  • A big number with no source, such as "100% accurate" or "10x faster".
  • "Leaked" or "insiders say".
  • "You are already behind". Fear of missing out is a tactic.
  • A forecast stated as a fact, such as "will replace half of all jobs by next year".
  • When a claim fails the filter, the right response is neutral: "Interesting if true. I will wait for a source."

    Step 6 — Keep a dated log

    Your memory is a poor place to keep AI knowledge, because the facts expire. A simple dated log solves this.

    What to keep. Four lists in one document or notes app:

    1. My tools. For each tool you use: what it is for, which plan, the date you last checked the plan and privacy settings.
    2. My best prompts and workflows. The ones that work, with the date, and the tool and version they were tested on.
    3. My rules. Your personal policy from the privacy tutorial, your employer's rules and any legal points that affect you, with dates.
    4. My watch list. Things you decided to look at again, each with a date.

    Always write "as of". Next to anything you record about a product, add the date: "free plan allows X, as of 14 Nov 2026". It reminds you, and anyone you share it with, that it may have changed.

    Review on a schedule.

  • Weekly: the routine above, which updates the log.
  • Monthly: look at the watch list. Promote, drop or keep each item.
  • Quarterly: a thirty-minute review. Re-check each tool's plan and privacy settings, retest your best prompts on the current version and prune the source list. Anything older than three months and unchecked is a candidate for being wrong.
  • This log is also the raw material for a personal timeline, which is a better review of the year than any article. It lets you answer "what changed, and when?" in a minute.

    Step 7 — Rules that prevent burnout

    Overload creeps up. These rules keep it away.

    1. One in, one out. Adopt a new tool only when you can name one you will drop, or when it clearly does a job none of your tools does. More tools means more to maintain, more subscriptions and more to learn.

    2. Batch your attention. Read AI news in your weekly slot, not in every spare minute. Mute or unfollow the sources that interrupt you.

    3. Skip a week without guilt. A skipped week costs little, because the important changes will still be there, and your quarterly review will catch what matters. Missing a month is also fine, if you then do the quarterly review properly.

    4. Do not chase the model of the week. The newest model is rarely the one that changes your work. Unless a release addresses a need you have, wait a few weeks and let the early problems surface.

    5. Learn the durable skills. Prompting, checking, giving context, choosing the right kind of tool and judging claims all outlast any product. Time spent on them pays back every year.

    6. Notice the signs of overload. Reading late into the night, a feeling of being behind, switching tools often, or bookmarking more than you read. If you see these, shrink your sources, shorten your slot and go back to Step 1.

    7. Talk to people. A colleague or friend with whom you swap one useful thing a week is better than a firehose. You get filtered signals, and you can compare notes.

    8. Protect what matters. Your attention is better spent doing good work with the tools you have than collecting new ones.

    Worked Example: Rina's routine

    Rina is a marketing manager. She feels constantly behind on AI, with dozens of tabs open and a long list of "to try" tools she never opens.

    She starts with Step 1. Her beats are drafting copy, summarising research, making social images and analysing campaign data. Everything else, such as coding agents and video models, goes in "Ignore" unless it touches those.

    She picks her sources: the release notes of the chat tool she uses daily, the plan page of her image tool, her company's AI policy page and one newsletter that links to primary sources. She unfollows a dozen accounts that mostly post "everything has changed" threads.

    She books twenty minutes every Friday at 4 pm. In her first session she reads the release notes (five minutes), finds a new feature for working with long documents, and puts it in "Act now". She reads one newsletter issue (five minutes) and notes a regulation story in "Watch". She tests the long-document feature on a public report (five minutes), and compares it with her old method. It is faster, and she notes that it needs a connection to her drive, which she does not grant. She writes three lines in her log and decides on one action: try it on one more public document next week.

    After a month she has a log with dates, a short watch list and a calmer feeling. In the quarterly review she drops one source and one tool and re-checks her plan and privacy settings. A colleague asks what she thinks of a hyped new tool. She runs it through the five-question filter in two minutes, and replies: "Interesting, but nothing shown yet. I will wait."

    Common Mistakes to Avoid

    Three habits cause most of the overload.

    Mistake #1: Following everything. Dozens of feeds, newsletters and channels feel like diligence and produce anxiety. Fix: define your beats, choose two or three sources by trust level, put most news in "Ignore" and keep to a twenty-minute weekly slot.

    Mistake #2: Trying new tools on real work. A new tool gets a client's file, an important message or access to an account on day one. Fix: test on a harmless task, compare with your current method, check the privacy settings first and start read-only.

    Mistake #3: Keeping it all in your head. Without a dated log, you cannot tell what is current, you repeat research and you act on facts that have expired. Fix: keep the four lists, write "as of" dates and review weekly, monthly and quarterly.

    Going Further

    Share the routine with your team. Agree who checks what: one person on pricing and plans, one on privacy and rules, one on new features. Share a short weekly note. It spreads the work and builds a common picture.

    Build a personal "how I decide" page. Your five-question filter, your trust levels and your adopt, watch or drop rule on one page. New people on your team will want it.

    Turn your log into a yearly review. At the end of each quarter, summarise what changed and what you did about it. Compare with a published year-in-review, such as the one earlier in this module, to see what you missed and what did not matter.

    What to learn next. This closes the module on AI in 2026. The quest, "hype check", puts the filter from Step 5 to work on an invented news post, so you can practise before the next real headline.

    Key Takeaways

    In this tutorial you learned:

  • You do not need to follow everything. Decide your beats and sort news into Act now, Watch and Ignore. Most of it is "Ignore", and that is healthy.
  • Use sources by trust level. Primary pages first, a few careful curators second, and almost nothing from the viral feed. Two or three sources are enough.
  • Keep a twenty-minute weekly routine. Official changes, one curator, one test and a few lines of notes. Set a timer and stop.
  • Test new things safely. A harmless task, compare with your current method, check the privacy settings, start read-only and decide: adopt, watch or drop.
  • Filter hype in seconds. Who is saying it, what did they show, can I check it, what changed for me and what is the catch.
  • Keep a dated log. Tools, prompts, rules and a watch list, with "as of" dates, reviewed weekly, monthly and quarterly.
  • Prevent burnout. One in, one out, batch your attention, skip weeks without guilt, do not chase the model of the week and learn the durable skills.
  • Before you move on, a challenge. Write your beats, choose your two or three sources and book a twenty-minute slot in your calendar for this week. Run the routine once, using the template, and write three dated lines in a new log. Next week, notice how much calmer it feels to follow a routine than to follow the feed.

    Learn AI, after work

    Track your progress, earn XP, and unlock more free tutorials in the AfterWork Academy app.

    Open this tutorial in the app

    More AI tutorials