Insights

Articles, news and info on Accrual and the AI accounting landscape

Article

Puzzle + Accrual

Accrual has agreed to acquire Puzzle’s accounting-firm technology and business; following close, Puzzle’s team and AI-native ledger and close capabilities will accelerate Accrual’s expansion from tax into client accounting services and a platform for the whole firm.

Read

Article

The IRS just told us what we already knew about AI, and that's the point

The IRS Office of Professional Responsibility's first AI guidance applies Circular 230's existing duties around judgment, verification, confidentiality, billing, and accountability to AI-assisted tax practice.

Read

Article

Arc: How Accrual Builds Accrual

How Accrual built an internal agent platform that lets CPAs, customer success, support, and operations teams build and operate with AI while keeping humans responsible for customer-facing work.

Read

Article

Frontier challenges in AI tax prep: Trust & estate K-1s

The 1041 K-1 has fewer boxes than its 1065 cousin, but they're harder to route correctly on the 1040. This article follows a single trust through the separately stated items, final-K-1 termination rules, and state nexus theories that break partnership K-1 muscle memory — and what firms should ask when evaluating AI tax tools.

Read

Article

Compounding Returns: Learning from the corrections nobody writes down

Tax Season 2026 was Accrual's first full season preparing 1040s end-to-end at production scale. After April 15, we built an autonomous improvement loop that reconciles every Accrual draft against the filed return, classifies preparer corrections, and ships verified fixes — months before TS27.

Read

Article

Why the 'AI Can't Do Accounting' benchmarks are asking the wrong question

A genre of AI benchmarks keeps concluding that AI can't do accounting. They evaluate a frontier model in isolation, one-shotting a task that no human or production system would ever one-shot. This post explains the mismatch between those benchmarks and how AI tax preparation actually works in production today.

Read

Article

Frontier challenges in AI tax prep: K-1 processing

K-1 processing isn't an extraction problem, it's a comprehension problem. This article walks through why extraction alone isn't enough, the three-layer approach Accrual uses for non-resident state K-1s and PTET, and what firms should ask when evaluating AI tax preparation tools.

Read

Article

How Accrual fits into your firm's tech stack

Accounting firms don't operate on a single system, they operate on dozens. This article explains how Accrual's API and developer platform connect to the tools firms already use, why we took an API-first approach, and how a unified platform compounds over time.

Read

Article

How the Accrual platform works

There is no shortage of AI products that can summarize a document or answer a question about a PDF. Preparing a tax return is a fundamentally different problem. This article explains how Accrual is architected specifically for accounting firms, and why the design decisions behind it preserve the trust, transparency, and judgment the profession depends on.

Read

Article

How to run a successful AI tax preparation pilot

Most enterprise AI pilots fail to reach production — not because the technology doesn't work, but because teams lack clear evaluation criteria and execution structure. This guide lays out a structured four-week framework for evaluating AI tax preparation with confidence.

Read

Article

From pilot to production: how to implement AI tax preparation

How leading accounting firms convert successful AI pilots into operational deployments — scoping the rollout, redesigning workflows, training at scale, and managing the organizational change.

Read

Article

Introducing Accrual: the augmented accounting intelligence platform

Accounting is the operating system of the economy. It establishes trust, informs major financial decisions, and underpins long-term growth for businesses and individuals alike. Yet despite its importance, much of the profession still relies on manual, fragmented workflows.

Read