Notion AI Office Hour
LinkedIn post (draft)
Spent the last hour building a financial tracker in Notion — live, on the free plan — and then layering simple AI workflows on top so it actually earns its keep.
What we built from scratch:
- A clean database with Item / Amount / Date / Category
- Multiple views grouped by category and month
- A simple chart to visualise spend
- A quick way to compare the same category across months
Then we talked about the AI layer (without overcomplicating it):
- Natural-language entry (e.g. “Handphone bill $35.10, log it”)
- Upload a bill (PDF / screenshot) → have AI pull the fields
- A realistic automation path for expense capture: start with transaction confirmation emails (vs. “connect directly to bank APIs”)
Big takeaway: AI doesn’t have to be complicated — start with what feels repetitive, break it into steps, and let the system handle the boring parts.
If you’re building your own “calmer operating system” in Notion and want to see more live builds like this, the full session replay is here:
Session recap
This week's Notion AI Office Hour ran live over Google Meet and was livestreamed on YouTube (Building with Nasri). True to form, the session was kept raw and interactive — rather than a fixed agenda, the attendees steered where we went.
We opened with a quick introduction (five years in Notion, from airport security and GrabFood rider to Notion consultant and ambassador, now working with an AI ops consultancy), followed by two warm-up polls:
- "What's one thing you wish AI could help you with?" — the room landed on "be my PA" and "second brain".
- "Which level of AI user are you?" — most attendees placed themselves at Level 2 (actively using AI for work tasks like writing emails), which is already fairly advanced.
From there, learning goals shaped the rest of the hour: building workflows, using Notion for more than just notes, and improving existing personal systems.
Live build: a financial tracker from scratch
Since the group leaned towards finance, we built a financial tracker live on the free plan, starting from an empty database:
- Core structure — Item, Amount, Date, and Category.
- Multiple table views grouped by category and by month.
- A chart view to visualise spending across categories.
- The sum/calculate function to total expenses.
- Group and subgroup by month to compare spending within a category across different months.
Jia Xuan kindly shared their own template-based expense tracker (currently manual logging, no analysis yet), and Joyce asked a great question about connecting Notion directly to a bank's API. We covered the reality there — direct bank integration is tricky for security reasons — and the practical workaround a fellow ambassador (Jesslyn) built: an agent that watches transaction confirmation emails via Notion Mail and logs expenses automatically (the "Moolah Tracker").
Notion API 101 workshop: Building a custom tracker with AI | Jesslyn S posted on the topic | LinkedIn𝗙𝗶𝗿𝘀𝘁 𝗡𝗼𝘁𝗶𝗼𝗻 𝗔𝗣𝗜 𝟭𝟬𝟭 𝗪𝗼𝗿𝗸𝘀𝗵𝗼𝗽 ✅ In my first Notion API 101 for Students workshop with Christy Zhang, I demoed trackmymoolah 🤠💰 My Telegram-to-Notion integration built to solve a simple pain point: there’s no central, customisable way to track both transfers across banks (like topping up Wise from DBS account) and individual spends -- the best way was to build it myself with full API and hosting control Every part of this build leaned on AI, where I shared both the shortcuts and the blind spots(!!!) via: • 📘 Notion API best practices (safety / capabilities / access ) • 💻 Optimise your AI coder to connect the APIs & host code • 🔑 Leverage AI for categorisation in transactions/spends ✨ (and Notion fun: Workspace Home Page → Home View DB → Feed View — so sleek!) ALL documented in a 📖 Workshop how-to guide so students could follow along at their own pace! Next up! Hosting another workshop with the same AI-first product thinking for my Notion MCP use case, trackmymoolearnlah (thereLinkedIn · linkedin.com
AI workflows that earn their keep
With the tracker in place, we showed how a light AI layer improves quality of life without overcomplicating things:
- Natural-language entry — e.g. "Today's handphone bill is $35.10, add it to the tracker."
- Uploading a bill (PDF or screenshot) and asking Notion AI to log it.
- Notion Mail connected to Gmail, with AI pulling email, calendar, and Slack into a single daily briefing.
- Agent skills — giving the agent character, work context, and reusable skills so it responds in a consistent format every time.
- A daily journal workflow that logs entries into a structured database under set headings.
- A pre-call workflow that loads an upcoming call, drafts discovery questions, and prepares a pre-email — all from the connected command centre.