A read-only window into your money
In May, OpenAI gave ChatGPT a view into your bank account. Through a partnership with Plaid, ChatGPT Pro users in the US can now connect more than twelve thousand institutions and let the model read their balances, transactions, investments, liabilities, and the subscriptions they forgot they were still paying for.
You ask where your money went last month and it answers from your actual statements instead of guessing. You ask how to pay down a card faster and it works from your real balances and rates.
OpenAI says more than two hundred million people already bring money questions to ChatGPT every month, so this was less a new idea than a long-overdue one, and the company bought its way to the front of the line by acquiring the team behind the personal-finance startup Hiro a month before launch.
It is a genuinely good release, and careful in the ways that matter. Plaid hands you a portal where you choose what to share and revoke it whenever you want, ChatGPT never sees a full account number, and OpenAI has said plainly that it does not change anything inside your accounts.
Intuit support is coming, and with it the model will be able to reason about the tax hit of selling a stock or your odds of clearing a credit check. If you have spent years exporting spreadsheets to understand your own money, the dashboard alone is a relief.
So if you just signed up for Sequence, the fair question is why you still need us.
Seeing is not the same as doing
Here is the line that matters, and ChatGPT draws it on purpose. It can read every account and move nothing in any of them. The Plaid connection is read-only by design.
It will tell you that you are three hundred dollars short on what you set aside for taxes, and then it stops, because moving that three hundred dollars sits on the other side of a wall it was deliberately built not to cross.
That is the correct default for a chatbot wired into your life savings. Reading your money and moving it are not the same risk, and you want the second one fenced in with real guardrails: scoped permissions, spending limits, an approval step, a log of everything that happened.
What is striking is that even the people building this layer describe the destination and then stop just short of it. Plaid framed the launch as “seeing your finances and finally knowing what to do about them.” Knowing what to do. Not doing it.
The whole current wave of AI in personal finance, impressive as it is, is a seeing layer. These tools narrate your money beautifully and then leave the moving to you, at your bank’s website, on a Tuesday night.
Where Sequence fits
Sequence is the acting layer. It is the part that takes what your assistant just worked out and carries it through to the money actually moving.
Install the Sequence MCP and the same assistant you are already talking to can top up the tax bucket it flagged, run the savings transfer the day your paycheck lands, or pay the bill it warned you about, each move bounded by limits you set in advance and held for your approval whenever it is unsure.
The point is not to hand a model the keys to everything. The point is to give it a narrow, audited set of actions, so the assistant that spotted the problem can also fix it.
So this was never really a choice between them. ChatGPT and Plaid gave your money a reader. Sequence gives it a way to act on what gets read. You will likely keep using both. And the old distance between noticing a problem on Sunday and finally fixing it on Tuesday night is mostly gone.
You already knew what was wrong with your cashflow. The new part is not that an AI can see it too. The new part is that something can finally do something about it.
Sequence API reference: https://api.getsequence.io/platform/. Knowledge base: support.getsequence.io. Share what you build.