Too cheap to be true? Yes. And the catch is not what you think. Before we touched any AI, we spent time just organising (the boring stuff).

The Situation

Property documents scattered across WhatsApp. Brochures in Telegram. Floor plans buried somewhere in Google Drive. Project updates in email threads nobody reads anymore.

The Steps

1. Getting everything into Drive properly.
2. Name folders by project.
3. Files named consistently.
4. Choosing one place the whole team agrees on.
5. Google Sheet: Projects as rows. Status, unit count, availability — all in columns. Simple stuff.

Once the data was clean, we didn’t need AI to count anything. A formula does that. It’s free, (almost) never wrong.

An Automation Script runs every night, scans the Drive folders, and writes a flat index file — project name, document name, link. Thirty lines of text.

That’s it. That’s the “database.”Then we deployed the AI. It reads the Sheet for numbers. It reads the index file for documents. It handles the conversation.

$12/month. Works for 1,000 users.

Not because the AI is smart.

Because we did the boring work first.

Spent the afternoon with a Mixue licensee and his team managers to talk about Notion and how it can improve their workflow. (Businessowner is already using Notion)

What they care about?

- Google reviews tracking

- Food wastage

- automated comms in telegram

- Staff 1to1 review

- Stock return or missing

- Ingredients price tracking

can Notion solve these?

Yes with simple input and AI for analysis. The key is to keep things simple and not overcomplicate it because at the end you just need data in your system and make decision making based on that. e.g. Utilise form for floor staff, real basic database for managers and that's how it'll make sense for company like Mixue. Not complicated systems.

#notion

Hi entrepreneurs! We will be hosting our monthly Notion Gathering 🇲🇾 this round is for those who run SME businesses (F&B, retails, multi chain stores)

📅 25th May 2026 (Monday)
🕰️ 11am to 1pm
📍 KL
🎟️ Free

You'll learn how to capture sales data, automate reports, and get real AI insights — without expensive software.

Seats are limited to 15pax and approval required 🙏🏼

As a Notion expert, here's what I found after implementing both for clients: the features are honestly kinda equal. The more useful question is what you're actually signing up for when you choose one of them.

They're Built for Different Kinds of Teams

Lark templates are built for operations. Retail checklists, outlet reporting, shift scheduling, procurement flows, attendance tracking. If you run a chain, a warehouse, or any kind of on-the-ground business — the templates feel like someone actually talked to your industry before building them.

Notion templates lean into knowledge-work. For example, product roadmaps, engineering wikis, marketing calendars, OKR trackers. The template ecosystem was shaped by the people who adopted it earliest — startup teams, product managers, designers, indie hackers.

Neither is better. They're just honest signals about who each tool was designed for.

Notion Is a Tool. Lark Is a Migration.

This is the big one and I think it changes the whole conversation.

When you bring in Notion, you're adding it alongside your existing stack. Your team keeps using Gmail, Slack, Google Drive — Notion just becomes the place where your docs and databases live. Low disruption. Easy to try. Easy to adjust.

Lark is a different kind of decision. It's a full platform — chat, video calls, docs, email, calendar, HR, approval flows, attendance, and Base (their database product). The vision is one app for your whole company. And that vision actually works, but it means you're migrating away from Google Workspace or Microsoft 365, not just adding a new tab

So if you're evaluating Lark just for the database — it's worth pausing. The database is good, but the real value shows up when your whole team is already inside the ecosystem. If half your team is on Slack and the other half is in Lark, you're getting a fraction of what the platform can do.

Where Lark Genuinely Shines

Once your team is actually living inside Lark, a few things stand out.

Permissions are more granular. Lark lets you lock down access at the field level — specific columns, hidden from specific roles. So your sales team sees deal values, your warehouse staff doesn't.

Notion's permissions work at the page or database level, which covers most needs but requires some creative workarounds when you need finer control. Notion's native forms are improving, but a lot of teams end up connecting Tally or Fillout to get the experience they want — which works fine, just adds one more thing to manage.

Dashboards are built-in inside Lark. Live charts and metrics pulling straight from your records, no extra tool needed. Notion has added chart blocks but they're still fairly limited for anything beyond simple visualisations.

Notifications actually get read. This one sounds small but it matters a lot in practice.

Lark notifications land in chat — the same place your team is already talking. A record update or an approval request pops up right next to their messages. Notion notifications live in a separate inbox that, honestly, most people stop checking after a while.

In Lark, automation flows naturally. When chat, docs, and database all live in the same platform, connecting them is straightforward. Trigger from a Base record, send a Lark message, kick off an approval — it just works. Notion can match this through Zapier or Make, and those connectors are genuinely capable, but it's a bit more setup and one more subscription to think about.

Where Notion Holds Its Own

Notion's real strength is how it blends data and thinking in the same space.

A database row isn't just a row — it can open into a full page with meeting notes, linked documents, decision logs, whatever context your team needs. For consultants, agencies, or any team where the work involves a lot of writing and referencing (like me), that combination is really hard to replicate elsewhere.

The AI integration is also genuinely useful.

Summarising pages, drafting from templates, asking questions about your workspace — it's woven into the tool rather than added on top. For teams where AI is becoming part of the daily workflow, Notion feels more ready for that.

And honestly, Notion just looks nicer. That's not a shallow thing — if your team finds a tool enjoyable to use, they actually use it. Adoption is half the battle with any new system.

So Which One?

Notion makes a lot of sense if your team is knowledge-heavy, you're already comfortable with your current chat and email setup, and you want something that plays nicely alongside your existing tools. The AI features and the doc-database combination are genuinely strong, and you don't have to ask your whole team to change how they work just to get started.

Lark is worth a serious look if you're running an operations-heavy business — retail, F&B, logistics, manufacturing — and you're open to consolidating your whole team onto one platform. The long-term value is real, especially at scale. But go in with eyes open: it's not just a software purchase, it's a migration, and that takes leadership buy-in and a proper rollout plan.

The honest truth is that if your team is already happy on Slack and Gmail, moving to Lark is a big ask. The tool is great, but the switching cost is real — not just in money, but in habits and culture.

I've been implementing both for Malaysian SME businesses.

Happy to answer questions about either.

Companies aren't replacing call centers because of AI. They're replacing them because they finally can.

AI didn't create the problem. It just made the excuse clean enough for a press release. Work left the US for India in the 1990s. By 2010, IBM's Global Location Trends report named the Philippines the call center capital of the world, overtaking India.

Today over 1.5 million Filipinos work in BPO. The industry is 10% of the country's entire GDP. Companies were never chasing quality. They were chasing cost.

So the real question was never about AI.

It was whether a 24/7 call center was ever truly sustainable in the first place. Or just a cost problem companies kept passing to whoever was cheapest.

If AI never existed, another country would have eventually replaced the Philippines once labour costs rose. This was never a sustainable solution.

Would love to hear if others see it differently

Currently tinkering with OpenWebUI + Ollama.
Got recommendations to try Mistral and Llama 3. For those of you running local AI setups: What stack are you using?

He's running a few F&B stores, with ZUS and Mixue as his main. These 2 brands are already huge in Malaysia. You (& me) probably had 1 today.

So when he said he needed more marketing, I was curious.

Franchisors have provided all the needed brand marketing, brand SOPs and even clientele reporting. What more marketing would you need?

He said he has been searching high and low but to no-avail (unless he also joins the viral music dance trend). He has tried marketing on social media, but it ended up boosting the overall branding instead. It did not lead to more store sales. Rent is hiking, sales is very slowly going downhill.

So then I proposed that what he needed was more personalised data that is catered to his location's community. How to stand out in that location to get more local customers? How to compete with the next-door Aicha?

Towards the end of the conversation, he wanted a system to connect his own project management (e.g. Notion or even pen & paper) with his backend POS that translates to ROI.

How?

- capturing useful data to increase sales (customer preferences, behaviour, demographics)

- customer feedback (how to further standardise their drinks to fight competitors?)

- operation efficiency (less repetitive copy-pasting, standardised reporting formats)

- waste management (tracking actual losses)

We started talking about connecting his POS to his Notion account. An easier way to input waste data for his employees and for himself. Data that is catered to him. Not generated monthly by franchisors.

Naturally, I started wondering how many Malaysian SMEs are in this situation right now? Hundreds of big brands, with hundreds of licensees under them, but lacking a tailored data-reporting for individual businessowners.

Capturing data can be in any forms and ways that can be convenient yet useful. For example, synchronised forms, audio feedbacks from staff consolidated into dashboards and many more.

It IS possible.

But is this gap of system integration worth tailoring to? or is there already something built for licensees?

Like yeah, who wouldn’t want a robot maid to clean my whole house and even clean my cat Oreo.

But reality is that we have Roomba that can clean our floor. Does it work all the time? No.

But is it good enough that it actually provides help?

Definitely.

That’s how I feel AI should be — realistic, and not just fantasizing about how it can replace everything.

To quote my favorite manga Frieren: “In the world of magic, you cannot realize what you cannot visualize. it’s the same with AI.”

If you can’t visualise a process that you can automate with AI then you can’t.

Period.