Skip to content
Little Red Bot AI automation for Singapore SMEs

The real cost of automation: subscriptions, setup time, maintenance

The sticker price is the smallest number in the bill. Credits that expire, AI metered separately, setup hours, and the maintenance nobody budgets — the honest first-year arithmetic for a small automation stack. Checked 15 Jul 2026.

Published 15 Jul 2026 · Facts checked 15 Jul 2026

The short answer: the subscription is the visible fifth of the iceberg. The real first-year cost of a small automation stack is subscriptions + usage meters (increasingly AI credits) + setup hours + the maintenance tax every time a vendor changes an API, a price, or a plan structure. We know that last one is real because it happened while we were writing this: n8n's cloud pricing changed between our 13 July and 15 July checks. Below is the arithmetic the brochures skip — including when the honest answer is a one-time build, or nothing at all.

At a glance — what things cost and how they're metered (checked 15 Jul 2026)

Line itemPriceThe meter (vendor's own definition)
MakeFree (1,000 credits/mo) · US$9/mo (5,000) [1]one credit per module action; credits expire; out of credits → "scenarios won't continue to run", webhooks queue [1]
ZapierFree (100 tasks/mo, two-step only) · Professional from US$19.99/mo [2]per task; AI sold separately: Agents Pro US$400/yr (1,500 activities/mo), Chatbots Pro US$160/yr [2]
n8n cloudno free tier · Starter 20€/mo annual (2.5K executions) · Pro 50€ (10K) [3]one execution = a whole workflow run, unlimited steps; AI Assistant metered (2,300–13,700 credits by tier) [3]
n8n self-hostedsoftware free; ~US$5–10/mo VPS [3]your server, your rules; AI = bring-your-own key, "no credit limits" [3]
Tidio Lyro AIfrom $32.50/mo annual (US$39 monthly) per 50 AI conversations → ~US$0.65–0.78 each [4]per AI-handled conversation
HubSpot Breezeseat + HubSpot Credits (Data Agent: 10 credits/enriched record) [5]per AI action, prepaid credits

The four costs, honestly

1. Subscriptions — the number you compare, and the least decisive one. A modest stack (automation platform + one AI chatbot + a CRM tier) lands at US$50–90/month ≈ S$800–1,450/year at list. Comparing tools on this line alone is how SMEs end up with the "cheap" one that meters everything else.

2. Usage meters — where 2026 pricing actually lives. The industry has converged on credits: Make's per-module credits [1], Zapier's tasks plus separately-priced AI activities [2], n8n's AI Assistant credits [3], Tidio's per-conversation Lyro [4], HubSpot's per-action credits [5]. Two rules of thumb: the unit definition is the price (one n8n execution = a whole run vs one Make credit = one step — a 10-step workflow costs 10× more "units" on Make); and AI is the new stealth line-item — bought "included", billed by the sip. Always ask: what happens at zero? Make's answer, verbatim: scenarios stop [1]. Chatbase's chatbot goes dark mid-customer. Plan for the refill price, not the sticker.

3. Setup hours — real, and paid in your evenings. Our own playbooks state 20–60 minutes per simple build, honestly measured — a first-timer's afternoon per workflow. A consultant charges S$100–200/hour for the same. Either is fine; pretending it's zero is not. Three workflows ≈ one working day of someone's time, once.

4. Maintenance — the tax nobody budgets. APIs change, plans restructure, prices move. Between our two checks this month, n8n went from trial-framing to a 20€ Starter tier [3]; GetResponse's free plan quietly vanished from its pricing page (that comparison); Pipedrive restructured its whole lineup this past year. Budget an hour a quarter per workflow to re-check that triggers fire, and treat any "set and forget" promise as marketing. This page carries its own check-date for the same reason.

The three-way maths at real volume

Take a concrete shop: 300 receipts filed, 100 lead alerts, 30 daily summaries — ~430 simple runs/month.

  • Subscribe-stack: fits Make Free (≈3 credits × 430 ≈ 1,290… just over — so US$9/mo) ≈ S$150/yr + setup day. Cheap while volumes are small; watch the credit line as you add workflows.
  • One-time build (self-host): n8n Community on a US$5 VPS ≈ S$80–100/yr flat, any volume, AI on your own key [3] — if someone can drive Docker. That "if" is the price.
  • Do-nothing: 430 manual touches × 2 minutes ≈ 14 hours/month of somebody's time. At even S$15/hour, that's S$2,500/year — the option that looks free and costs the most, provided the work actually happens. (Sometimes it doesn't, and the cost is invoices nobody chased.)

The flip point is simple: step-heavy or high-volume → per-run pricing (n8n) or self-host; light and simple → Make's free tier wins and n8n cloud can't match it (no free tier) [1][3]. Full head-to-heads: Make vs Zapier, Make vs n8n.

Singapore-specific notes

  • Everything here bills USD or EUR (+ GST where charged) — put the FX line in the budget.
  • PSG does not fund these self-serve subscriptions. Grant-supported automation means a pre-approved vendor's packaged solution — different product, different price, and the reason a S$12k "automation package" quote deserves the markup check and the grant-vs-direct calculator before you sign.
  • PDPA: self-hosting keeps workflow data (customer names in your sheets, chats) on your own Singapore server — for some firms that alone decides the architecture.

Which approach for which business

  • Non-technical, a few chores: Make free tier + our playbooks. Cost: an afternoon.
  • Growing volume, someone slightly technical: self-hosted n8n on a small VPS — flat cost, uncapped AI on your own key [3].
  • Already paying for Zapier at work: stay, but price its AI products separately before turning them on [2].
  • No time, real budget: have it built — with a published flat quote, against the 7 questions.

You might need neither if…

…the chore happens less than weekly, or a checklist fixes it. Automation earns its keep on frequency. A task you do four times a month, taking three minutes, costs you 2.4 hours a year — no stack, no meter, no maintenance beats that. Automate the daily grind; leave the monthly ritual alone.

Your first honest budget, step by step

  1. List your repeated chores with weekly counts (ten minutes, be honest).
  2. Multiply: runs/month × steps ≈ credits (Make) and runs/month ≈ executions (n8n); place yourself on the table above.
  3. Add one setup afternoon per workflow, and an hour a quarter for maintenance.
  4. Compare the total against the do-nothing hours at your real hourly worth.
  5. If the winner is "buy nothing yet" — that's a valid result. Bookmark the playbooks for when frequency rises.

FAQ

Is the AI-credit thing really that significant? It's the fastest-moving line. An AI agent answering 300 customer conversations is ~US$210/mo on Tidio's ladder (monthly billing) [4] and ~US$24–60 in Chatbase credits (why) — a 4–10× spread on the same headline "AI chatbot". Meters, not stickers.

Won't prices in this article go stale? Yes — that's the point of the fetch dates on every table, and why each of our comparisons carries a re-verification schedule. If today's date is far from 15 Jul 2026, click the sources before deciding.

Cheapest possible stack? Free ChatGPT/Gemini for thinking, Make Free for plumbing, the free WhatsApp Business app for customers, Google Sheets as the database — S$0/month, all covered by playbooks. Its cost is your setup afternoons, and it's the right answer for most sub-10-person firms.

Next step

Do the ten-minute chore list, run the table maths, and if a vendor is quoting you a five-figure "automation transformation", bring both to the grant-vs-direct calculator before signing anything.

Sources

  1. Make pricing: make.com/en/pricing — fetched 15 Jul 2026
  2. Zapier pricing: zapier.com/pricing — fetched 15 Jul 2026
  3. n8n pricing: n8n.io/pricing/ — fetched 15 Jul 2026 (changed vs our 13 Jul snapshot — noted in body)
  4. Tidio pricing (Lyro ladder): tidio.com/pricing/ — fetched 15 Jul 2026
  5. HubSpot pricing (Breeze/credits): hubspot.com/pricing/crm — fetched 15 Jul 2026

Sources

Every factual claim above traces to one of these, fetched on the fact-check date in the byline.

  1. Make — pricing (Free plan, credits defined, expiry, out-of-credits behaviour)
  2. Zapier — pricing (free plan limits, Professional; AI products priced separately)
  3. n8n — pricing (cloud tiers, executions, AI credits; self-host free)
  4. Tidio — pricing (Lyro AI per-conversation ladder)
  5. HubSpot — pricing (Breeze agents billed via HubSpot Credits)

How we're funded: some links on this page are affiliate links — if you sign up through one, we may earn a commission, at no extra cost to you. It never changes what we recommend. Full disclosure.

Get the next one in your inbox

The Red Dot Report — one useful automation email a week.