Tax season. Just saying it out loud makes most people's shoulders creep up toward their ears. But something genuinely interesting is happening this year—AI tools have gotten good enough that they're starting to make a real dent in the misery of filing, whether you're a freelancer drowning in 1099s or a small business owner trying to figure out what's actually deductible.
Let's talk about what's new, what's actually useful, and where you should still be careful.
The Big Picture: AI Is Finally Talking to Tax Code
For years, tax software was basically just digital forms. You typed in numbers, it did arithmetic. Exciting stuff. What's different now is that large language models are being woven into these platforms in ways that let you ask questions and get reasonably coherent answers back.
TurboTax's AI assistant, for instance, has gotten noticeably more capable at walking users through edge cases—things like home office deductions for hybrid workers, or how to handle crypto transactions that span multiple exchanges. It's not perfect, but it's a lot better than googling IRS Publication 587 and hoping for the best.
H&R Block has been pushing their AI Tax Assist feature hard this year, and honestly it's more impressive than I expected. You can describe your situation in plain English and get guidance that's actually tailored to your specific mess of circumstances. The underlying model is trained specifically on tax law, which matters a lot—a general-purpose chatbot will confidently give you wrong answers about tax stuff if you're not careful.
Document Ingestion: The Quietly Huge Improvement
Here's the thing that doesn't get enough attention: the real time sink in tax prep isn't understanding the rules. It's the data entry. Hunting down W-2s, pulling together receipts, reconciling brokerage statements that somehow span forty pages.
Several tools this year have made serious leaps in document processing. You can photograph or upload a document—a crumpled receipt, a PDF from your brokerage, a handwritten mileage log—and the AI will extract the relevant numbers and slot them into the right fields. Keeper Tax has been doing this for freelancers and it's genuinely slick. Expensify's SmartScan has been around for a while but keeps getting sharper.
For small business owners especially, this is where AI is earning its keep. Not in the philosophical questions about tax strategy, but in the grinding, tedious work of just getting your data organized.
What About Actual Tax Strategy?
Okay so this is where I want to pump the brakes a little. AI tools are getting better at identifying deductions you might have missed—things like the home energy credits from the Inflation Reduction Act, or self-employed health insurance deductions that people routinely overlook. Some tools will proactively flag these based on your inputs, which is genuinely useful.
But "tax strategy" in any meaningful sense—like, should you convert your traditional IRA to a Roth this year, or how should you structure your S-corp distributions—that's still firmly in human CPA territory. The AI tools I've tested will often give you a reasonable starting answer on these questions but they miss nuance, they don't know your full financial picture, and the stakes are high enough that you really don't want to rely on a chatbot.
Think of the current generation of AI tax tools as a very knowledgeable assistant who's great at research and data entry but who you'd still want a senior partner to review before anything goes to the IRS.
Some Tools Worth Knowing About
Keeper Tax — Built specifically for freelancers and gig workers. The AI scans your bank and credit card transactions to find deductible expenses you might have missed. Subscription-based but worth it if you have a lot of business expenses scattered across accounts.
TaxGPT — A newer entrant that's essentially a tax-focused AI assistant. You can ask it questions, upload documents, and get reasonably detailed answers. Still a bit rough around the edges but it's improving fast and the team is clearly paying attention to accuracy.
Intuit Assist — Intuit has been embedding AI across their whole product suite, including TurboTax and QuickBooks. If you're already in the Intuit ecosystem, the AI features are getting good enough to actually change how you work.
Column Tax — A newer tax filing platform that's built AI into the core experience from the ground up, rather than bolting it on after the fact. Worth watching.
The Privacy Question You Should Be Asking
Before you start uploading your financial documents to any AI-powered tool, please, please read the privacy policy. Your tax documents contain some of the most sensitive personal and financial information that exists. You want to know: Is your data being used to train models? Who has access to it? How long is it retained?
Most reputable platforms are pretty clear about this, but "reputable" is doing a lot of work in that sentence. Some of the newer AI tax tools are startups that haven't been stress-tested at scale. That doesn't mean they're bad, but it means you should ask the question.
Where This Is All Heading
Honestly, I think we're one or two product cycles away from AI tools that can handle a pretty significant chunk of the tax prep process end-to-end for individuals and small businesses. Not everything—complex situations will always need humans—but the routine stuff? The W-2 income, the standard deductions, the basic small business expenses? That's getting automated, and fast.
What I'm most excited about isn't even the filing itself. It's the idea of AI tools that help you make better financial decisions throughout the year so that tax season isn't a scramble. Real-time tracking, proactive alerts when you're approaching deduction thresholds, suggestions for retirement contributions before year-end. Some tools are starting to do this and it feels like the right direction.
For now though, the practical advice is simple: try one of these tools this year if you haven't already. Start with something low-stakes—use an AI assistant to answer a question you'd normally google, or upload a document and see how well it extracts the data. Build your intuition for what these tools are good at. Because they're only going to get more capable, and the people who've been experimenting will have a real head start.