The close is drafted before your team logs in.
You review. You approve. Done. Nothing posts to the books without your sign-off. Your firm sets the rules. AI runs them.
3h to 30min
50% faster
3x client capacity
10 to 30 clients
2x margin
21 to 3 days
Trusted by forward-thinking accountants.
Built on the only ledger designed for accountants.
Most AI accounting tools give you a chat window or a checklist. Puzzle gives you a full AI Suite built directly into the general ledger. Each tool is connected. Every action is auditable. And because it all lives in the actual ledger, nothing gets lost between systems.
- Ask anything.
- Categorize a batch of transactions.
- Reconcile from a PDF bank statement.
- Get a variance explained in plain English.
- Chat handles one-off requests right inside the ledger.
- Turn any Chat task into a scheduled agent.
- Build your firm's playbook once.
- It runs automatically across every client, every month.
- What your best accountant built belongs to your firm now.
- Your month-end process, managed as a team.
- Assign agents as tasks. Set due dates.
- Review together.
- Full audit trail on every step so partners and auditors can see exactly what happened and when.
What Accountants are saying
Every client is different. AI Close adapts to each one – not a one-size-fits-all checklist.
It is like NetSuite, but without paying for everything. It is an open template. You can build all kinds of agents and establish different rules.
90% of a new client close done in 30 minutes. It used to take 3 hours.
How Debit & Co. Closed All Clients Books by the 4th Business Day with Puzzle
Accountants are the architects.
AI is the engine.
Four steps. Every action logged, sourced, and accountant-approved.
Nothing posts to the GL without your sign-off.
Write the workflow in plain language, the way you'd explain it to your team.
The agent is in learning mode: iterating, reasoning, finding the optimal solution. You don't write code. You write instructions. The agent handles the rest.
AI converts your description into explicit, runnable steps. The logic locks.
This is where probabilistic AI becomes deterministic. The solution is found, locked, and saved. Same inputs, same outputs. Every single run. Your reconciliations and categorizations are not being re-imagined each time. They are executed with the consistency you expect.
You approve. The agent executes. Full audit trail attached.
Nothing posts to the GL without your explicit sign-off. The agent surfaces its findings in a review panel: every proposed entry, every reasoning step, every source transaction. You review, ask follow-up questions in plain English, and approve. Control never leaves the room.
AI runs a final review of the month-end close for accuracy.
After running all your tasks, kick off the Review & Finalize analysis to scan every record across the platform. AI surfaces discrepancies in a review panel: each anomaly flagged, each affected account, each source transaction. You review the findings, ask the agents to continue working on them, and take actions. Nothing slips past you.
Every action logged.
Every entry traces back.
Nothing posts without your approval.
Every action logs the date, source, reasoning, and who signed off.
100%
Full
Zero
Built on three principles.
Designed for the way accountants work.
Customizable
You design every workflow. The agent runs exactly what you built. No code. No consultants. Just plain-English instructions.
Permissioned
Nothing posts to the GL without your explicit approval. Full control at every step. Every action requires a human sign-off.
Aligned
AI executes your judgment at scale. It does not replace it. The accountant is always the architect. The agent is always the engine.
Start here. Your firm's first agents, already written.
Every prompt below is pre-connected to your client's data. Pick one, describe what you need, and let the agent work. Build once, reuse across every client in a vertical.
Bank Statement Transaction Extraction & Reconciliation
Extract the transaction table from this bank statement PDF with the highest possible accuracy. Match each row against the existing ledger entries and flag any…
Match Unpaid Bills to Payments
Identify unpaid bills over the last 3 months and propose the most likely payment matches based on amount, date proximity, and memo. Flag any bills that…
Identify & Draft Unbilled Accruals
Using expense patterns from the last 3 full months, identify likely unbilled costs to accrue this month: subscriptions, recurring services, and vendor…
Journal Entry Review & Anomaly Detection
Review all journal entries posted this period and flag any that appear unusual: round numbers, unusual accounts, entries posted outside business hours…
Prepaids Review
Review the current prepaid and accrual schedules alongside this month's GL activity. Identify any prepaids that should have amortized but haven't, and…
Transaction Audit
Review last month's transactions for the following risk patterns: duplicate payments, split transactions below approval thresholds, and vendors with…
Prepare for a Client Call
Give me a summary of this client's business health and 3–5 talking points for our call. Use plain English for the titles and flag anything that needs…
Variance (Flux) Analysis
Given the trial balance for the current month and the prior month, perform a flux analysis. For each account with a variance over $X, explain the likely…
Cash Runway Estimate
Estimate runway in months based on current cash and last month's burn rate. Produce a summary table showing projected cash over the next 6 months and…
Forecasting & Narrative
Using the client's last 6 months of general ledger activity, build a rolling 13-week cash flow forecast. Highlight the 3 biggest drivers of cash movement…
Find a Key Business Insight
Analyze this client's full GL history and find the single most valuable insight, something actionable that the client likely doesn't already know about…
Profit Margin Analysis
Using the client's revenue and cost data from the general ledger, perform a margin analysis at the product or service line level. Flag any lines where…
How much capacity is your firm leaving on the table?
Adjust the inputs to see your firm's potential capacity recovery.
Scale without headcount.
Stay in control.
Speed is the hook. Control is what closes the deal. AI Close delivers both, without asking you to give up either.
Go deeper.
AI Playbook for Firms
The Next Generation Firm Playbook: How to Adopt AI in Accounting Without Losing Judgment, Margin, or Client Trust
What are AI Agents
What is an AI agent in accounting, and what does it mean for the way your firm works? Learn more here.
Case Study: Debit & Co.
How Debit & Co. Closed 15 Clients by the 4th Business Day with Puzzle
AI Isn’t Taking Accounting Jobs. It’s Exposing How Firms Use Time.
The risk isn’t replacement.
It’s misallocation.
When repetition compresses, judgment expands.
The question is whether your workflow is built for that shift.
How AI Ready Is Your Firm?
Are you compressing repetition
or defending it?
Seven questions.
Instant clarity.
Questions we hear a lot.
Nothing posts to the GL without your approval. Every agent drafts the work for review and approval, citing the finding, reasoning and recommended next steps before you click approve. If something looks off, you edit it, delete it or override it. You're not trusting the AI blindly. You're reviewing its work the same way you'd review a junior bookkeeper's output. The AI has already done the heavy lifting.
Puzzle's categorization is a custom learning model. It trains on your firm's industry mix and your accountants' preferences, not a generic ruleset. Most clients see up to 98% auto-categorization by month 2 to 3 without writing a single manual rule. The remaining transactions can be auto-routed to the client, who responds in natural language by email. By the time you log in, everything is ready for review.
Two differences that matter. The technology: QuickBooks AI is probabilistic. Same input, different output. Puzzle is deterministic. Same input, same output, every time. Your firm sets the rules. AI runs them. The business model: Intuit is using AI to compete for your clients. Puzzle will never compete for your clients, by design. That guarantee is in writing.
That's exactly what the Add Custom Agent flow is for. Beyond the standard templates, you can create custom agents with plain-English instructions. Multi-entity real estate, agencies with class tracking, SaaS companies with complicated revenue recognition: all of these run as custom agents on top of the standard close. Build once, reuse across clients. No code required.
That's exactly what the Add Custom Agent flow is for. Beyond the standard templates, you can create custom agents with plain-English instructions. Multi-entity real estate, agencies with class tracking, SaaS companies with complicated revenue recognition: all of these run as custom agents on top of the standard close. Build once, reuse across clients. No code required.
Your Judgment. At AI Speed.
See it live.
See AI Close in action with a live demo tailored to your firm's workflow.