How Gemini Spark Cut My Google Sheet Trade Entry by Over 90 Percent

TL;DR

For 8 years, every buy and sell in my portfolio got typed in by hand, twice: once into a Google Sheet, once into Yahoo Finance. That's 11 to 15 transactions a month, each one copied field by field from a brokerage screen or an email. Gemini Spark, Google's always-on AI agent announced at I/O 2026, now reads my brokerage's trade confirmation emails automatically and writes the order type, ticker, account type, quantity, price, and date straight into the sheet. The manual work is gone. What's left is a once a month CSV download to check the automation's work, a change that cuts the manual work of building that sheet by more than 90 percent. That figure is for the Sheet side only; the honest number for the whole double entry workflow is closer to 50 percent (the split is worth unpacking below). The same pattern works for any brokerage that emails a confirmation, since the extraction cares about the fields in the email, not the sender.

The Manual Workflow: Logging Every Trade by Hand, Twice

Before Spark, logging a trade meant logging into the brokerage account or opening the trade confirmation email, then reading off the same six fields: order type, ticker, account type, quantity, price, and date. I typed each one into a Google Sheet by hand. Order type, buy or sell, is easiest to skim past, and getting it backwards flips the whole entry. That sheet wasn't the end of it. The same trade got entered a second time into Yahoo Finance, the tool that gives a single glance at holdings across every device I use. Over 8 years at 11 to 15 trades a month, that's roughly 2,000 manual entries, each one a chance for a typo to slip into either record. A single mistyped quantity or price doesn't just look wrong on the sheet; it throws off the running cost basis for that ticker, the number that determines gain or loss when the position eventually sells.

Gemini Spark: How the Daily Scheduled Job Extracts Trade Data

Gemini Spark is Google's 24/7 agentic assistant: it runs on a dedicated Google Cloud virtual machine, working even with the laptop closed, rather than firing off one reply and stopping. Give it a standing instruction once, schedule or trigger condition included, and it repeats the job on its own. In this setup, a Gmail filter isolates every trade confirmation email from the rest of the inbox. A Spark job runs daily, reads matching emails, and writes the same six fields, order type included, into a new row in the Google Sheet. That sheet is now the collected data source: the single record everything else gets checked against.

Any Email Brokerage Works: Why This Isn't Wealthsimple Specific

This setup runs against Wealthsimple, because it emails a confirmation every time an order fills, but nothing depends on its specific format or an API key. Spark reads plain email text for six recognizable fields, so the logic holds for any brokerage that emails those details. A brokerage with no confirmation email, only in app fills, breaks the pattern: nothing for the filter to catch. The bigger shift is in auditing. Instead of double checking every entry as it happens, the review moved to a monthly task: download the brokerage's CSV once a month and diff it against the sheet Spark has been building, which catches an email Spark missed rather than an entry a human mistyped.

The pattern isn't specific to trades. Any manual data entry task that starts with an email and ends in a Google Sheet, receipts, shipment confirmations, invoice line items, rent payments, can be automated with Gemini Spark today.

Yahoo Finance Entry: The One Step Gemini Spark Doesn't Touch Yet

Spark automates the collection leg, not the whole pipeline. Getting each trade into Yahoo Finance, the app that still provides the best glance at holdings across devices, remains a fully manual step today. There's an automation for that half already in place, but it's unreliable enough that it isn't ready to stand behind in writing yet. That's a separate article once it holds up consistently. Stay tuned.

Bonus: A Household Owner Field, Courtesy of a Forwarding Rule

This household runs two Wealthsimple accounts, mine and my spouse's. Her confirmation emails do not land in my inbox on their own, so a Gmail forwarding rule routes them into the inbox mine already uses. That one rule handed Spark a field for free: the filter already has to look at where each message came from to catch the forwarded copies, and that same signal doubles as an owner tag, mine or hers, with no second integration. The sheet stopped being a personal trade log and became a household one, a seventh field at zero extra setup cost. Any household running two brokerage accounts through one inbox gets the same tag for the price of a forwarding rule.

The Results: One Leg Cut by 90 Percent, the Whole Workflow by Half

It's worth being precise about which number applies to what. Spark automated one of two entry legs, so the 90 percent figure belongs to the Google Sheet side specifically: 11 to 15 manual entries a month collapsed into a single monthly reconciliation. Yahoo Finance entry didn't change at all, so measured across the full double entry workflow, the realistic cut is closer to 50 percent, roughly half the original typing eliminated, not nine tenths of it.

For anyone sitting on years of the same double entry, automating the first leg alone is still worth doing, it's the difference between typing every trade and reading a monthly reconciliation, even before Yahoo Finance gets solved.