October 6, 2026

5 Boring Office Tasks You Can Automate This Week (No Coding Required)

The case for boring automations

Everyone wants to automate something glamorous — an "AI employee" that does their job while they sleep. Ignore that. The automations that actually pay off are aggressively boring: the receipt you photograph, the meeting notes nobody writes up, the follow-up email you keep meaning to send. Each one steals 15–30 minutes a week. Five of them steal half a day.

Here's my rule: automate the task you dread, not the task that impresses people. Dread is a reliable signal — it means the task is repetitive, low-judgment, and exactly what software is good at.

Below are five automations I'd build in a small business this week, ordered by effort-to-payoff ratio. Each one takes under two hours to build on Make, n8n, or Zapier, and each one pays for itself within the month. Pick one. Just one. Build it before Friday.

1. Meeting notes that write and file themselves

The pain: You finish a call, tell yourself you'll write up the notes, then don't. A week later nobody remembers what was agreed, and the action items evaporate. Multiply by every meeting, every week.

The automation: Record the call (Zoom, Google Meet, and Teams all do this natively — just hit record and tell people you're recording). When the recording finishes, a workflow grabs the transcript, sends it to AI with a strict prompt ("Extract: decisions made, action items with owners, open questions. Format as bullet lists."), and files the result where it belongs: the project channel in Slack, a Notion page, or an email to attendees.

Build notes: The trigger depends on your meeting tool — Zoom's "recording completed" webhook, or a scheduled check of a Google Drive folder where recordings land. The AI step is the same pattern as lead scoring: force a strict output format so the result is skimmable, not a wall of text. Always include a link back to the full transcript for the one time a month someone disputes what was said.

Honest caveat: AI summaries are 90% accurate, which means they're wrong about something 10% of the time. For routine internal meetings that's fine. For contract negotiations or anything with legal weight, a human still reviews. Automate the draft, not the sign-off.

Time to build: ~1 hour. Saves: ~1–2 hours/week if you meet regularly.

2. Receipt and invoice collection without the shoebox

The pain: Receipts live in your wallet, your inbox, your camera roll, and that drawer. At tax time (or expense-report time) you reconstruct everything from memory, which is another way of saying you lose money.

The automation: Create a dedicated email address — receipts@yourdomain.com. Any time money leaves the business, forward the receipt there (most email receipts can auto-forward with a filter). The workflow watches that inbox, extracts the vendor, date, and amount (AI vision or simple parsing — most receipt emails have structured data already), and appends a row to a spreadsheet: date, vendor, amount, category, link to the original email.

For paper receipts, snap a photo and email it to the same address. The AI reads the photo. It's not perfect — handwriting and crumpled thermal paper defeat it sometimes — but it beats the shoebox by a mile.

Build notes: This is mostly email filters plus a spreadsheet append — the simplest automation on this list. The categorization prompt should use your actual expense categories, not generic ones. Review the sheet monthly for 10 minutes; that's the whole maintenance burden.

Honest caveat: Your accountant will still want the originals for anything large or unusual. This system organizes the 95% routine stuff so the 5% weird stuff gets proper attention instead of drowning in chaos.

Time to build: ~45 minutes. Saves: 1–3 hours/month, plus real money at tax time from expenses you'd otherwise forget to claim.

3. Follow-up emails that send themselves (politely)

The pain: You send a proposal or quote, the prospect goes quiet, and following up feels awkward — so you don't, and the deal dies from neglect rather than rejection. Every freelancer and small agency owner knows this feeling.

The automation: When you send a proposal (log it in your CRM or a simple spreadsheet — even a "Proposals" tab works), start a timer. If no reply in 4 days: send follow-up #1 (short, helpful — "just checking this didn't get buried, happy to answer questions"). If no reply in 9 days: send follow-up #2 (add value — a relevant case study or a useful link, not "just bumping this"). If no reply in 16 days: send the breakup email ("I'll assume timing isn't right — the door's open whenever"). Then stop. Three touches, then silence.

Build notes: The trigger is "proposal logged + no reply detected." Reply detection is the fiddly part: the simplest reliable version watches your inbox for replies from that prospect's address and cancels the sequence when one arrives. Every platform handles this with a filter + a "cancel" path — draw it on paper first, because the logic has more branches than it looks.

The important rule: these emails must come from you, in your voice, and each one must be genuinely useful or gracefully terminal. Nobody has ever closed a deal with "just circling back!!" for the fourth time. Write the three templates once, make them good, and let the machine handle the timing — timing is the part humans are bad at.

Honest caveat: This automation sends emails as you. Test it on yourself first, and include an easy opt-out. One tone-deaf automated follow-up does more damage than no follow-up at all.

Time to build: ~1.5 hours (the reply-detection logic is the bulk of it). Saves: hard to quantify, but agencies routinely attribute 10–20% of closed deals to systematic follow-up. Even one saved deal a year pays for a decade of automation costs.

4. Review and mention monitoring → one morning digest

The pain: Customers review you on Google, Yelp, industry sites, and social media — and you find out weeks later, if ever. A bad review unanswered for a month is a billboard for neglect. A good review you never thank is a wasted relationship.

The automation: Once a day (morning, before you start work), a scheduled workflow checks your review sources — Google Business Profile alerts, social mentions, whatever applies to your industry — collects anything new from the last 24 hours, and sends you one digest email: new reviews/mentions, each with a suggested draft response written in your tone. You approve, edit, or ignore each one over coffee. Five minutes, done.

Build notes: Google Business Profile has notification options natively — start there before building anything. For the rest, most platforms have "watch" modules, or you can use RSS/Google Alerts piped into your workflow. The AI draft-response step should include your business name, the reviewer's name, and one specific detail from their review — generic "thanks for your feedback!" responses are worse than slow ones.

Honest caveat: Never auto-post review responses. Drafts are fine; publishing without human eyes is how you end up with an AI apologizing to a customer for something that wasn't your fault, in public, permanently. The human approves. Always.

Time to build: ~1 hour. Saves: 30 min/week of checking, plus the incalculable cost of the review you would have missed.

5. The weekly numbers email you never have to compile

The pain: Every Monday (or every month-end) someone — probably you — opens five tabs, copies numbers into a spreadsheet, and writes the "how did we do" summary. It's pure copy-paste work wearing a trench coat.

The automation: A scheduled workflow runs every Monday at 7 a.m.: pull last week's numbers from wherever they live (Stripe revenue, new CRM contacts, website analytics, ad spend), drop them into a template, have AI write two or three sentences of plain-English commentary ("Revenue up 12% vs. last week, driven by..."), and email it to you (and your team, if you have one).

Build notes: Start with just two data sources — the ones you'd check first manually. Most platforms have native Stripe/Google Analytics/CRM modules; where they don't, a simple API call or even a scheduled CSV import works. The AI commentary prompt should include the previous week's numbers for comparison, and strict instructions: "Only describe what the numbers show. Do not invent explanations." AI loves inventing explanations. Don't let it.

Honest caveat: The first version of this email will be slightly wrong or slightly useless. That's fine — you'll tweak which numbers get pulled and how the commentary is framed over three or four weeks, and then it'll be right forever. Budget 10 minutes a week for the first month to review and adjust.

Time to build: ~2 hours (data connections are the slow part). Saves: 1–2 hours/week forever, and — more importantly — you actually look at your numbers every week, which most small businesses simply don't do.

How to actually get one built this week

Here's the part where most "automate your life" articles fail you: they give you ideas but no plan. So here's the plan.

Monday (30 min): Pick ONE from the list — the one you dread most. Sign up for Make's free tier (or open n8n/Zapier if you already use one). Connect the two apps involved.

Tuesday–Wednesday (60–90 min): Build the simplest version. Not the clever version — the dumb version that handles the common case. Test it with real data, not just the happy path.

Thursday (20 min): Break it on purpose. Feed it weird inputs — the blurry receipt, the meeting with no clear decisions, the prospect who replies "maybe later." Fix what breaks.

Friday (10 min): Turn it on and tell one person it exists. Accountability beats willpower.

Then leave it alone for two weeks before building the next one. The temptation is to build all five in a weekend; the result is five half-working automations you don't trust. One working automation you trust beats five you don't. Trust compounds — each success makes the next build faster because you understand the patterns.

What this costs

Per automation, typical running cost:
- Make: $0–9/month (all five together fit comfortably in the $9 Core plan's 10,000 credits; start on the free tier)
- n8n: $0 self-hosted (+$5–10/month server) or from €20/month cloud
- Zapier: $0–19.99/month annual (watch the task math — multi-step workflows burn tasks fast)
- AI usage: $1–5/month total across all five (small models, short prompts)

Realistic total for all five: $10–25/month on Make or n8n; $30–60/month on Zapier depending on volume.

Time investment: ~6 hours to build all five, spread over a few weeks. At even a modest $50/hour value of your time, the payback period is measured in weeks, not months.

The uncomfortable truth

None of these automations is hard. The technology has been ready for years, and the AI step that used to require a developer is now a form field with a prompt box. The reason most small businesses still do all five tasks manually isn't technical — it's that nobody scheduled the two hours to build the first one.

So schedule it. Right now, before you close this tab: pick the task, block 90 minutes on your calendar this week, and build the dumb version. Future you — the one not photographing receipts at 11 p.m. — says thanks.

← Back to all field notes