Puzzle's New Bulk Classification Rules, October 2026

Ever spent hours tagging transactions one by one?

Puzzle now offers Bulk Classifications and Classification Rules for all users. These new features overhaul how you manage transaction classifications, automating tedious manual work and bringing real-time financial visibility to startups and their accounting teams.

TLDR:

• Manual transaction classification is one of the highest-volume, lowest-value tasks slowing finance teams down
• Bulk Classifications let you tag hundreds of transactions at once; Classification Rules apply tags automatically to every match, past and future
• QuickBooks and Xero apply rules on a per-transaction review model; legacy tools leave the cleanup work to you
• 46% of accountants use AI tools daily (up from 18% in 2023, per the 2025 Intuit survey), and 80% of routine bookkeeping tasks can already be automated (Intuit QuickBooks, 2025; Wolters Kluwer, 2025)
• Puzzle's Bulk Classifications and Classification Rules are available now to all accounts at no additional cost

The real pain of transaction classification (and why it needs fixing)

For years, accountants and finance teams have wrestled with the same time-draining ritual: manually tagging, categorizing, and re-checking transactions, over and over again. We've heard countless stories of:

"We spend hours each month classifying one transaction at a time."

"We keep a whole spreadsheet just to track our classification logic. It's easy to miss things."

"If I could set it and forget it, I'd never look back."

After closely observing how firms across industries handle this work, the pattern was consistent: finance teams were spending 5 to 10 hours a month on transaction classification alone, clicking through transactions one by one, maintaining side spreadsheets to track their own logic, and still finding misclassifications at month-end. That's a full workday lost every month to a task that's almost entirely rule-based. We saw a clear opportunity to eliminate that repetition and the human error it compounds.

Introducing two new classification features that save hours of work

Classifications (sometimes called classes or departments) are tags that help you track financial data across different segments of your business. They answer questions like:

• How much is each department spending?
• Which locations are most profitable?
• What's the cost breakdown for each project?

At Puzzle, you can already assign classifications to MJEs, Bills, Invoices, and Prepaids. With these new capabilities, transaction-level classification becomes measurably faster and more automated.

Here's what's new:

Bulk Classifications

Select and classify multiple transactions at once, just like you would for a single one. Perfect for clean-up, review, and catch-up work.

Classification Rules

Set it once, apply it forever. Create auto-classification rules that automatically tag transactions based on criteria you define.

Why this matters for accountants and business owners alike

These new features change how you work with classifications in ways traditional accounting software can't match:

• Save valuable time by eliminating repetitive manual work that drains productivity
• Gain deeper business insights with consistent departmental reporting that drives decisions
• Maintain data quality through automation that reduces human error
• Scale operations without proportionally increasing finance headcount
• Close books faster with automated classification workflows that support month-end close automation

Unlike QuickBooks and other legacy systems, Puzzle delivers these capabilities with no add-ons required, intuitive workflows designed for real users, real-time updates across all reports, and a scalable system that grows with your business: all without requiring technical skills or complex setups.

And as accounting moves increasingly toward AI-assisted workflows, these automation features reflect where accounting is already heading: organizational AI adoption has jumped from 22% to 40% in just the past year (Thomson Reuters, 2026), giving you the foundation for more intelligent, real-time financial management while reducing manual intervention. Understanding the balance between AI bookkeeping automation vs. judgment helps clarify where rules-based classification ends and human review begins.

How to use Bulk Classifications: classify multiple transactions at once

Bulk Classifications let you apply the same classification to many transactions simultaneously. No more clicking through them one by one.

Here's how it works:

1. Go to your Transactions list
2. Select multiple transactions by checking the boxes next to each one
3. Click "Classifications" at the bottom of the screen
4. Choose the classification you want to apply (e.g., Department: Marketing)
5. In moments, all selected transactions will reflect the new classification

This feature is perfect for catching up on unclassified transactions or making batch corrections when needed.

How to create Classification Rules: set it and forget it

Classification Rules automate your workflow by creating "set it and forget it" rules that automatically apply classifications based on transaction patterns.

Here's how to set them up:

1. Go to the "Rules" section
2. Create a new rule by specifying a condition (for example, if the description contains "AWS")
3. Choose the classification you want Puzzle to apply when that condition is met
4. Decide if the rule should apply to future transactions only or all transactions (past and future)
5. Click "Save" and watch your classifications happen automatically

This feature shines with recurring expenses like monthly software subscriptions, office rent, or any transaction pattern that follows a predictable rule.

How automated transaction categorization works in modern accounting software

Most modern accounting tools (including QuickBooks, Xero, NetSuite, and Puzzle) share a common foundation for automated categorization, but differ considerably in how deep that automation goes and how much manual cleanup it leaves behind.

At the core, automated categorization works through three layered mechanisms:

• Bank feed rules (pattern matching): When a transaction comes in via bank feed, the software scans its description, amount, or counterparty against a library of user-defined or system-suggested rules. If "AWS" appears in the description, it maps to Cloud Infrastructure. QuickBooks calls these "bank rules"; Xero calls them "bank rules" too; Puzzle calls them Classification Rules: same concept, different execution.
• Machine learning on historical patterns: Over time, the system learns from how similar transactions were previously categorized (by you or by anonymized patterns across users) and surfaces suggestions automatically. The more consistent your classification history, the more accurate the suggestions become.
• Receipt and document matching (OCR): Higher-end systems extract data from receipts or bills (via OCR), match them to open transactions, and pre-fill category and vendor fields, reducing manual entry to a confirmation click.

Where legacy tools stop short: QuickBooks and Xero surface suggestions and apply rules, but default to a one-transaction-at-a-time review model, meaning you still approve each one individually. NetSuite offers more sophisticated matching, but its configuration requirements are sized for mid-market teams with dedicated implementation resources, not early-stage startups.

Puzzle builds the opposite model. Classification Rules apply automatically to every matching transaction (past and future) the moment you save them. Bulk Classifications let you sweep hundreds of uncategorized transactions in a single action. And because Puzzle's books update in real time, your P&L and departmental reports reflect the correct classifications immediately, not at next month-end.

Where the industry stands in 2026

The case for automating transaction classification has only gotten sharper this year. The global AI in accounting market hit $10.87 billion in 2026, up from $4.87 billion in 2024; automated bookkeeping is its fastest-growing segment, expanding at a 46.1% CAGR (Mordor Intelligence). Vendors are far from the only ones investing: 98% of accounting professionals globally now report using AI in some form (Karbon, 2026), with organizational AI adoption jumping from 22% to 40% in just the past year (Thomson Reuters, 2026).

The pattern is consistent with what industry analysts have identified as the defining shift of 2026: accountants are moving from preparers to reviewers. When transaction matching, variance analysis, and classification run automatically, the close stops being a sprint and becomes a continuous flow. Finance teams that build that foundation now, starting with rules-based classification, are positioned to layer in AI agent workflows for month-end close as they scale. Those still approving transactions one by one are slower and may fall structurally behind.

The shift is already happening: don't get left behind

Firms moving fastest on automation are closing their books faster and reclaiming hours every month that used to go to manual data work. See how the best accounting automation tools for startups stack up.

That repetitive work compounds: missed classifications at month-end, side spreadsheets to track logic, and reports that don't reflect reality until it's too late to act. Bulk Classifications and Classification Rules are Puzzle's direct answer to that, built not as a workaround but as a native part of how your books stay current in real time.

Frequently asked questions

What is Puzzle's AI assistant (CoWork), and what can it do for transaction categorization?

CoWork is Puzzle's built-in AI assistant. Beyond surfacing categorization suggestions, it can run repeatable accounting workflows on a schedule: applying classification rules, flagging anomalies, and helping with tasks like revenue recognition, all without you having to log in each time. You direct it through plain-language instructions; it executes against your live books and waits for your approval before anything changes.

How does AI categorization handle vendor invoices with multiple line items?

When a vendor invoice carries multiple line items (say, a SaaS invoice split across infrastructure, support, and licenses), Puzzle's categorization engine processes each line individually instead of forcing a single category onto the whole transaction. Classification Rules can match on description keywords, vendor name, or amount range per line. QuickBooks' bank rules apply at the transaction level only, which means multi-line invoices often need manual overrides; Puzzle's line-level rules eliminate most of that cleanup.

How does Puzzle match a single bulk bank deposit to multiple individual customer invoice payments?

Puzzle's matching logic looks for groups of open invoices whose amounts sum to the deposited total (a common pattern with Stripe payouts or batched check deposits). When a match is found, Puzzle links the deposit to the individual invoices automatically and marks them paid. If the amounts don't add up cleanly (e.g., partial payments or foreign-exchange differences), Puzzle flags the deposit for manual review instead of forcing a wrong match.

How do I categorize individual contractor payments when payroll doesn't send a single lump sum?

Use a Classification Rule targeting the contractor name or payroll platform description (e.g., "Gusto" or the contractor's legal name). Set the condition to match on description, choose your Contractor Payments GL account and the relevant department classification, and apply it to all matching transactions, both past and future. If contractors are paid through different platforms or descriptions vary, create one rule per pattern; they stack and run automatically on every sync.

Why isn't the AI automatically identifying my vendors and categorizing transactions correctly?

AI categorization suggestions improve with volume and consistency: the more transactions you classify correctly, the stronger the pattern signal. If you're early in setup or your bank feed descriptions are cryptic (e.g., "ACH DEBIT 4492817"), the AI has less to work with. The fix: create an explicit Classification Rule for that vendor or description pattern. Once a rule exists, Puzzle applies it deterministically (no guessing) to every future transaction matching that condition. Rules always take precedence over AI suggestions.

How does AI categorization learn and improve over time?

Puzzle's AI learns from your classification history. Every time you correctly tag a transaction (manually, via Bulk Classifications, or by saving a Classification Rule), it reinforces the pattern signal for that vendor, description keyword, or amount range. Over time, the suggestion accuracy improves because the model has more consistent data to work from. That said, rules are always deterministic: if you've created a Classification Rule, Puzzle applies it exactly every time, regardless of what the AI might otherwise suggest. The AI fills the gaps; rules close them permanently. The practical implication: early in setup, lean on rules for your high-volume recurring expenses. Let the AI handle the long tail, and correct it when it's wrong. Those corrections compound into better suggestions over the following months.

How do I set up classifications and view department-level spending in Puzzle?

Go to your Classifications settings and create the departments, locations, or projects you want to track (for example: Marketing, Sales, Engineering, G&A). Once your classifications exist, you can assign them to transactions manually, in bulk via Bulk Classifications, or automatically using Classification Rules. For instance, a rule can tag every transaction where the description contains "Google Ads" as Marketing. Once classified, Puzzle's P&L and departmental reports update in real time, so you can see spend by department without waiting for month-end. If you want to compare department budgets against actuals, run a filtered P&L by classification: you'll see each department's revenue and expenses broken out as its own column.

Ready to start saving time with automated classifications?

Both features are available now to all Puzzle accounts at no additional cost.

Not using Puzzle yet? Start a free trial to see how automated classifications reshape your accounting workflow.

‍