Decimal, a leading modern accounting firm, partnered with Puzzle's AI-native general ledger to enhance operational efficiency and scale their services while maintaining quality across hundreds of clients.
Decimal, a leading modern accounting firm, partnered with Puzzle to complete their vision of building one of the most operationally efficient accounting firms in North America. Through this partnership, they've found a technological advantage for their business including:
"Puzzle is helping us reimagine how we deliver on our promise to clients." Matt, Decimal's CEO
TLDR:
Decimal wasn't a typical accounting firm looking for basic improvements. Through careful optimization of their team and processes, they had built one of the more operationally mature modern accounting practices around. Their focus: process excellence and a clear understanding of what makes accounting firms scale.
"We faced a choice: continue with legacy systems or find a true technology partner who could match our ambitions." Matt, Decimal's CEO
Decimal has always prided itself on using the best technology for its clients, but as they looked at adopting and applying AI, they faced a key challenge: legacy general ledgers were not practically built to work with AI. The options were clear but limited:
After weighing their options, it became clear that Puzzle was building the AI-native ledger needed to scale with their ambitious goals.
Puzzle stood out by offering something different: a true technology partnership that could complement Decimal's existing strengths. By partnering with Puzzle, it allowed them to focus on being the best at their core services while using an AI-native general ledger built for the AI era. The partnership between Decimal and Puzzle is already showing promising results:
| Outcome | What It Means | Result |
|---|---|---|
| 📈 Scalable Operations | Automated quality checks, standardized workflows, and checklists | High-quality books delivered for hundreds of clients |
| 🔬 Rapid Product Velocity | Dedicated communication channel with Puzzle's product team | Feedback turns into real product improvements |
| 💻 Accurate Data Migration | Full client book migration at scale | 100% data accuracy across hundreds of clients, completed in weeks |
The result is improved scalability with Decimal's team able to focus more on service instead of repetitive tasks, with happier clients to show for it.
"Having a direct line of communication with Puzzle has made a real difference. We've seen our feedback turn into real product improvements, and that level of responsiveness has been invaluable." Mitch, Operations Lead @ Decimal
The timing of Decimal's move matters. In 2026, the accounting industry crossed a threshold: roughly 85% of mid-size and large firms now use some form of AI in their workflows, up from 58% in 2024. More importantly, the frontier has shifted from AI-assisted tasks to agentic systems: AI that can detect an anomaly, investigate its source, flag it for review, and draft a corrective journal entry without human initiation. The AICPA's 2026 CPA Firm Top Issues Survey found that change management due to technology and AI ranked #1 across every firm-size group for expected impact over the next five years. A Thomson Reuters 2025 AI adoption report reinforces the urgency: organizations with a clear AI strategy are twice as likely to experience revenue growth as those without one.
Firms that built on legacy ledgers retrofitted with AI are already hitting the ceiling. Firms that partnered with AI-native infrastructure, like Decimal did, are the ones positioned to run agentic workflows without rebuilding from scratch.
The numbers that framed Decimal's decision have only sharpened since they made it. Karbon's 2026 State of AI in Accounting report found that 98% of accounting professionals now report using AI in some form, a number that would have seemed implausible three years ago. But adoption alone no longer separates firms. According to KPMG's 2026 AI in Finance report, the gap between firms is now determined by AI governance: organizations with stronger governance and assurance capabilities consistently report better business outcomes than those that focused only on deploying the technology.
That distinction matters for infrastructure decisions. Firms that layered AI tools on top of legacy ledgers are now finding that governance is hard to retrofit too. When the underlying data model was not built for AI, human checkpoints get inserted at every step by necessity, not by design. Firms reviewing AI bookkeeping software for CPA firms that chose AI-native infrastructure from the start, like Decimal did with Puzzle, are now positioned to deploy governance systematically and avoid patching it in after the fact.
With Puzzle as their technology partner, Decimal delivers on its mission of making it easier for startups to stay on top of their finances. They continue to lead the industry in execution excellence, now backed by technology that can truly scale with their ambitions. The partnership represents more than a software implementation: it's the beginning of Decimal's vision for what a modern accounting firm can be, and an inflection point in their story of what's to come.
The ceiling on firm scale is rarely talent. It's the software. Legacy general ledgers require someone to manually touch every client's books at every step: categorize, match transactions, review, close. That model caps how many clients one accountant can carry before quality slips or hours balloon. Puzzle's AI-native architecture breaks that ceiling. Automated quality checks, standardized close checklists, and AI categorization that handles up to 98% of transactions let a single accountant oversee far more clients without sacrificing accuracy. Decimal moved hundreds of client books onto Puzzle and maintained 100% data accuracy in the process. If your firm is growing faster than your headcount can, that's the model worth looking at.
Monthly PDFs are a byproduct of how legacy software works: the close happens once, the report gets exported, the client gets a snapshot that's already weeks old by the time they read it. AI-native infrastructure changes the underlying rhythm. Because Puzzle categorizes and processes transactions continuously (not in a single month-end burst), cash position, P&L, and balance sheet stay current daily. Anomaly detection runs in the background, flagging unusual patterns before they reach reports. Firms like Decimal can give clients a live view of their financial position at any point in the month, going well beyond a retrospective PDF after the close. The shift from monthly PDFs to real-time visibility is an architectural outcome, not a feature you toggle on.
Firm partners get a meaningfully different experience. Decimal has a dedicated communication channel directly with Puzzle's product team, which means feedback turns into real product improvements instead of disappearing into a support queue. Onboarding for firms is designed around migrating entire client books at scale: Decimal moved hundreds of clients with 100% data accuracy in weeks, not a single-account setup flow. If you're considering Puzzle as a firm, the right conversation is with our firm partnerships team, not a standard sales motion.
Legacy general ledgers were built decades ago to store and report transactions. AI features added on top of them can automate tasks at the surface, but the underlying data model wasn't designed for it, which means there's a ceiling on what the AI can reliably do before human intervention is required at every step. An AI-native ledger is architected from the ground up so that AI operates at the data layer: detecting anomalies, investigating sources, drafting corrective entries, and running month-end close checklists without needing to bridge between old infrastructure and new models. That's the difference Decimal recognized, and why firms building on legacy systems retrofitted with AI are already hitting limits that Puzzle's architecture was built to avoid.
The short answer: it works continuously in the background so your team isn't manually touching every transaction. Puzzle's AI categorizes up to 98% of transactions automatically, learning from your firm's patterns as it goes. When confidence falls below the threshold, it flags the transaction for human review instead of guessing, so errors don't compound silently. AI Accuracy Reviews run proactively, catching misclassifications before they reach reports. For revenue recognition, the AI applies the correct deferral and recognition schedule at the invoice line-item level, automatically. For payroll providers with a native integration (Gusto, Rippling, Deel), journal entries are drafted without manual input. The human accountant's role moves from data entry to review and approval, which is where their expertise actually adds value. Nothing posts until a person confirms it.
Puzzle's AI builds categorization rules automatically as your team finalizes transactions, so accuracy compounds over time and never stagnates. Once a vendor or transaction type is confirmed, future matches follow the same rule without prompting. For revenue recognition, you set the policy by product or subscription type. Puzzle then defers and recognizes revenue over the correct period automatically, applied at the invoice line-item level. Split transactions can be divided by percentage or dollar amount across categories. Bulk editing lets your team handle high-volume periods without clicking through individual records. The No Documentation filter surfaces transactions missing receipts in one view, so nothing gets missed during close. The AI handles the repetitive work; your accountants handle exceptions and sign off on the output.
Yes. For payroll providers that don't have a native Puzzle integration (ADP and Paychex are permanently off the integration roadmap), Puzzle's AI payroll agent accepts uploaded payroll reports in PDF or spreadsheet format and automatically drafts the corresponding journal entries. Your team reviews and approves the entries before anything posts. Native integrations exist for Gusto, Rippling, and Deel; those draft journal entries without any upload required. For Rippling, there's an additional fee of $2 per month per employee and contractor on top of the base subscription. If a client runs payroll on an unsupported provider, the upload-and-draft workflow keeps the process structured without requiring manual entry from scratch.
Puzzle's AI Close agent is agentic bookkeeping automation currently in beta, available to select larger partner firms and not yet broadly available to all partners or general users. For firms that do have access, it automates portions of the monthly close workflow: running checklists, flagging exceptions, and drafting entries without waiting for a human to initiate each step. AI credits apply only to conversational queries in Puzzle's Accounting AI chat, not to the transaction categorization engine or close checklist automation. Credit allocations vary by plan: 25 lifetime credits on Starter and Core, 100 monthly credits on Complete, and 300 monthly credits on Scale. If your firm is interested in AI Close access, the right next step is a conversation with Puzzle's firm partnerships team to understand current availability and eligibility.
Data loss is the biggest fear in any migration, and it's a legitimate one. Historical transactions, closed periods, and audit trails are not easy to reconstruct once they're gone. Puzzle's QuickBooks migration process is designed around that concern. For firms migrating clients off QuickBooks Online, Puzzle's direct integration imports historical financial data through a QuickBooks Connect button, and the migration typically completes in a couple of days. For QuickBooks migration with messy books, the recommended approach is importing a cleaned trial balance as the opening balance in place of raw historical transactions, which preserves accuracy without inheriting legacy errors. Decimal ran this process at scale: hundreds of client books migrated with 100% data accuracy, completed in weeks. If historical books are particularly messy, Puzzle's migration partners can handle the cleanup before go-live. The short version: you keep the history, you lose the friction.
---
Ready to Complete Your Vision? Learn more here
