August 2026 • 5 min read
Tags: UX Design, AI, Fintech, Product
TL;DR: Financial companies are starting to integrate AI into their products and connect it to our financial data, so we can ask questions and get insights about our money. This is a major step forward in how we understand our finances, but that is exactly why we need to be careful not to outsource our understanding and decision-making to AI.
This shift is already happening. Calcalist’s article about IBI, Fair, One Zero and AI describes how IBI and Fair allow customers to analyze their investment portfolios through ChatGPT, ask questions about diversification, exposure, performance and risk levels, and receive explanations in plain language.
In my view, this is a great direction.
But there is one condition:
Do not outsource your understanding.
What Actually Changed Here?
Until now, many uses of AI inside financial products were based on a simple idea:
The system identifies what might interest me and shows it to me.
It notices that I spent more than usual and shows me an insight.
It detects a change in my investments and summarizes it.
It finds a subscription and suggests canceling it.
You can build more and more dashboards, charts, categories, alerts and insights, and still not know what the user actually wants to understand.
There is a big difference between receiving an insight the product chose for me and being able to investigate my own information by myself.
If I see: "Your expenses increased by 18% this month," I am still dependent on what the product decided to check. But if I can ask why, in which category it happened, whether it was a one-time expense or a trend, and what the month looks like if that expense is excluded, I can already investigate my own money by myself.
That is a meaningful change.
Maybe This Is the Right Shape for PFM
Personal Finance Management has been trying for years to help us understand and manage our money.
But it is very hard to predict in advance what every person will want to know about themselves.
You can build more and more dashboards, charts, alerts and insights, and still miss the question the user actually wants to ask.
A conversational interface changes the equation.
Instead of trying to predict all possible questions in advance, you can let the user ask.
How much did I spend on restaurants in the last six months? How have my mortgage payments changed? How much of my portfolio is exposed to the U.S. market? These are questions a classic interface struggles to answer without building endless screens, filters and reports in advance.
You can see this direction in existing products outside Israel as well. Finances in ChatGPT lets users connect financial accounts and ask questions about spending, subscriptions, net worth, investments and goal planning. Copilot Money’s Money Assistant also lets users ask questions about their own financial data inside the app.
This is why connecting personal financial information to an LLM is much more interesting than another AI layer that generates summaries.
AI does not need to decide in advance what matters to me.
It can give me a way to reach the question I care about.
But This Is Where the Trust Problem Begins
The more questions the interface lets us ask, the more persuasive it becomes.
An IMF study on LLMs and macrofinancial analysis found that advanced models can assist with financial analysis, but also that they still have meaningful limitations, show certain biases and struggle with open-ended questions that require deep contextual judgment.
Will I know when it is wrong?
When I receive an answer from a conversational model that analyzed my personal data, the answer can feel personalized and therefore more trustworthy.
And when the model sits inside the app of my bank or investment house, another factor enters the equation:
Who gave me the answer?
If I open ChatGPT and ask it a financial question, I understand that I am talking to an AI model.
If I open my bank app and see a chat that knows my account, the experience is different.
Its presence inside the bank’s system can feel like a stamp of approval.
This is not just theoretical. Calcalist’s article about IBI, Fair and One Zero quotes a cybersecurity expert who says that when the bank invites you to use a chat like this, it creates a feeling that it is safe and trustworthy.
I actually agree with that point.
And Shaul Amsterdamski’s column about AI and finance adds another interesting layer: he describes a situation in which AI gets access to our bank accounts, savings, investments and insurance, and can analyze them and recommend what to do.
The risk is that we will simply move from trusting the bank’s recommendations to trusting recommendations from an AI model that appears inside the bank’s system.
This Is Where UX Becomes Critical
The design question is how to make AI give a good answer, and also how to help the user understand the answer, its source, its limits and the degree of responsibility they need to take for it.
If AI makes my financial information more accessible, I gain an ability I did not have before.
If it makes me dependent on it in order to understand my money, I lost something along the way.
AI Can Reduce the Friction on the Way to Understanding
AI can help me ask questions I did not know how to ask.
It can help me find patterns in large amounts of information.
It can help me test scenarios.
It can make complex financial information more accessible.
In the Calcalist article, IBI Smart’s product manager puts it quite precisely: the role of AI is to make the full picture accessible, not to make decisions for the investor. OpenAI also describes Finances as a tool meant to help users understand, plan and evaluate financial decisions, while emphasizing that users are responsible for their decisions and that ChatGPT is not a financial adviser.
But the goal should be that I understand more, not that I stop thinking.
AI can help us understand our money better. But it cannot replace our understanding.
The responsibility for the decisions always remains ours.