The evolution of AI expertise for accountants

The FR sat down with Dave Talach, senior vice president of the money and end-to-end experiences team at Intuit. Prior to working at Intuit, Talach held senior roles at Block, Capital One, and PayPal.

Much of your role at Intuit is about addressing the needs of small and medium-sized business owners though Intuit’s QuickBooks platform, one of the dominant players in that space. What would you say some of the biggest pain points are for small business customers across financial services?

The fintech industry as a whole, is rallying through to solve a plethora of challenges. You go back 10 years, [small and medium-sized enterprises] couldn't get access to capital. Before Square, they couldn't get access to payment processing. With Intuit, it's a source of truth. It's your financial books. It tells you where you stand, and if you actually understand it, you can make better decisions.

What would you say is the biggest pain point for small business owners managing their financial books?

When we survey our customers, they use a tapestry of tools. They have to navigate to connect them together. They have different logins, data siloed in different places. The closest opportunity and the closest platform that can stitch these things together, it’s QuickBooks. 

Fintech as a community has made incredible strides, and we should all be incredibly proud of that: democratizing access to capital and making payments easier. We still have more work to do. What do you do with all that [data]? How do you put it all in one place? That's what we're hearing from customers. They have multiple logins. They have to cut and paste stuff—it just doesn’t connect. I’ve got to cut and paste stuff. 

Where do you see AI helping meet some of the pain points of the customers that you serve?

The big opportunity is how do you get some of that minutia [tasks] that should just be done for them — off their plate so they can get back to speaking with customers. That's what our lesson has been with AI. It’s about how you can automate work.

Can you offer some context on the evolution of Intuit Assist, the AI capability that works across your product suite?

When [generative AI] first broke on the scene, we were all amazed at what ChatGPT could do. We thought it could do everything, and that it could be a digital expert, and it could provide advice and it could help you with reporting. Many of us went into this space feeling very ambitious. Now, the real opportunity is being very pragmatic about what are those places that small business owners spend a lot of time on. Where are they doing minutia and manual work?

When I think of Intuit Assist, I don't think digital expert, I think digital assistant. So you can send it stuff: You can send it job notes. You can send an email. It can construct an estimate. It can construct an invoice. It's almost like a person that works for you.

I take it you’re taking a focused approach to the deployment of the AI assistant?

We've gotten to a point where AI will not do everything for everyone just yet, and we’ve got to start by...really harnessing the tech to focus on [the tasks] that are dragging people down.

Keeping that in mind, does Intuit Assist look like a chatbot, or is it something different?

When we originally started, it was almost more like a bot, almost like a ChatGPT interface, a right rail that popped out and it had a blinking cursor. You looked at it and it almost put cognitive load on you. What am I supposed to ask this thing? What is it capable of doing? Expecting people to engage with AI just because it's AI is not valuable. So what we've done is we've embedded it into the product—into existing workflows.

Now you can send that email to Intuit Assist, and behind the scenes, it will actually create a draft invoice for you, or you can just drag a PDF or drag a file into the invoice form, and it will upload it and go and fill out the invoice.

What are some examples of what the Intuit Assist toolset can do for QuickBooks users?

Using any phone, you can take pictures of handwritten notes. You can upload emails and forward emails. With any set of unstructured data, it will create a draft invoice, a draft estimate and it will do expenses or receipts.

One of the most common places that our power users are is in this transaction categorization table where they're like categorizing [expenses] and so we actually have Intuit Assist now making much more accurate and explainable, transparent recommendations. So it's essentially automating accounts receivable and accounts payable and automating accounting tasks.

We want to be thoughtful, that when we offer insights and advice, it's great insights and it's great advice—it's not just AI for the sake of AI. In the meantime, we have an incredible partner community of accountants that we can lean on. Part of the call to action is that we actually want to work with them. We don't want them to think that AI is somehow going to dilute or undermine [their work] but we want to work with this community so that AI can actually augment human expertise, because we think the two coexist.