When AI Can Prepare a Partnership Return, What Happens to the Business Model?

What happens when an AI agent can prepare a partnership tax return on its own? Not just summarize supporting documents or draft portions of the return but actually complete it.

That’s the claim behind a recent announcement from Basis, the artificial intelligence startup. It built an AI agent capable of autonomously completing a partnership return.

Partnership returns aren’t simple forms. They involve layered allocations, basis tracking, capital account maintenance, multi-tiered entity structures, and K-1 footnotes that can run dozens of pages. For decades, that complexity functioned as a moat around compliance revenue.

If AI can cross that moat, even partially, we need to rethink some assumptions.

We recently discussed this shift on The Accounting Podcast. You can watch the full breakdown in the video below:

This Isn’t Just “Another Tax Tool”

We’ve seen automation in tax before. OCR, data extraction, digital organizers, and e-file systems save accountants time, but this is different.

An AI “agent” implies something more autonomous: software that can ingest source documents, reason through tax rules, make decisions within defined parameters, and produce a completed return with limited human intervention.

If that sounds ambitious, it is. But it’s not irrational.

Large language models already interpret contracts and reconcile structured data. Tax compliance is rule-driven and structured — exactly the type of work where AI systems tend to improve rapidly.

So, AI can draft a Form 1065. But can it draft one accurately, consistently, and defensibly enough to change the economics of compliance work?

Economics Are the Real Story

Step back from the tech for a moment.

Across firms of all sizes, tax compliance accounts for roughly half of revenue, and business tax return preparation is the most profitable service line, according to Thomson Reuters 2025 State of Tax Professionals Report.

If an AI agent can handle even 60-80% of the mechanical work involved in partnership returns, that’s not marginal efficiency. It’s a cost structure shift. And once the cost structure shifts, pricing pressure follows.

The complexity moat protecting profit margins is no longer reliable when AI agents are being trained to handle increasingly complex work.

Judgment is the last line of defense

Now, let’s temper the hype.

A partnership return is rarely “standard.” There are always edge cases, elections, and interpretations. Then there’s that little issue of professional liability.

I’m guessing the IRS won’t recognize “AI prepared it” as a defense under preparer penalty rules.

Don’t blindly automate. Design layered workflows where AI handles structured, repeatable calculations and humans review for judgment, elections, and risk.

Automation without oversight is negligence. Automation with governance is leverage. There’s a difference.

The Bigger Question Is, “What Are We Selling?”

If AI agents can materially reduce the labor involved in partnership compliance, what exactly are we charging for? Time? Headcount? Or expertise?

For decades, the billable hour has been tethered to production effort. If production effort shrinks but pricing stays the same, clients will notice. If pricing drops but firms fail to reposition around advisory value, margins compress. This isn’t theoretical.

We’ve seen this shift before with bookkeeping, payroll, and basic write-up work. Firms that survived moved up the value chain into advisory, structuring, forecasting and interpretation.

If partnership compliance is next, the value shifts from assembly to analysis.

Direction of Travel Matters

I’d love to see a live demo of how Basis prepares a partnership return. But whether Basis’s agent is flawless today is almost beside the point. The direction of travel is clear.

AI systems are getting better at structured reasoning all the time. Investors are aggressively funding automation. Clients expect faster turnaround and clearer value. And finding humans to do this work isn’t easy.

When those forces converge, business models change. So don’t treat this announcement as hype.

I’ll leave you with this: If partnership returns become largely automated over the next five years, what will your firm be known for? And are you building toward that future or defending the past?