How Autonomous Accounting Works

Andrew Robinson, CPA

Updated: July 14, 2023

How Autonomous Accounting Works

Autonomous Accounting sounds too good to be true. You’ve thought it, shared it, and we heard it. Curiosity surrounding its feasibility makes sense, considering the complex nature of accounting. Let’s delve into the realm of Autonomous Accounting, exploring the intricacies of how it works in a technical sense, and shed light on how technology is changing the game for accountants and business owners.

Technically, what is Autonomous Accounting?

Autonomous Accounting is Puzzle’s system that automatically generates real-time financial reports, metrics, insights, reconciliations, and tax documents, whenever prompted. It’s a system that combines the logic of rules, automation, and AI into our general ledger. Our unique combination of metadata ingestion, APIs, machine learning, and AI is what ensures the system is learning and improving its accuracy every single day.

In addition to being a system, Autonomous Accounting is also a paradigm shift. It’s a way of thinking about the accessibility of financial information and preparation of financial reports while emphasizing efficiency and effectiveness, in order to amplify human expertise. At Puzzle, this is the foundational framework we use to develop every aspect of our product. Every feature on our roadmap always starts with a question to ensure Autonomous Accounting is built into this core framework: How can remove the barriers between raw financial data and the company's urgent need for this data to enable timelier decision making?

Raw data is not generally useful for decision making. That’s why humans need to aggregate, clean, summarize, and analyze, and transform this raw data into financial information (such as easily digestible financial statements or metrics) so it becomes useful for decision making. Puzzle helps automate the cleaning and converting of this raw data into useful financial information to make decision making easier.  

Of course, many accounting tasks cannot be automated. Puzzle bridges the gap between automation and the need for human input because Puzzle isn't just about being autonomous. Puzzle is about being autonomous first, but not autonomous only. Here’s one example of how this works in Puzzle: 

  • Puzzle auto-categorizes and includes transactions in the general ledger that are immediately populated into financial statements, so financials are useful in real-time, even if they’re not perfect or fully reviewed yet.
  • After rules, automation, and AI has created draft financial statements, we surface exceptions and transactions that cannot be categorized for human review. This system enables easy and efficient review to make sure all the pieces fit together (like a Puzzle, get it?!)

Of course there are much more complex accounting processes than this example. This starts to gets exciting as we extrapolate this automation to more traditionally complicated workflows such as accrual schedules, deferrals, and depreciation.

Our approach of autonomous-first, but not autonomous-only, is important and makes Puzzle unique. Not all accounting systems are built with this framework in mind, as you’ll learn below. Most legacy accounting systems were built with technology that was available at the time of origin and with a different product approach. At Puzzle, we are taking the mindset of using new, available technologies to re-approach accounting software. The ultimate goal is to help users maximize time while retaining the importance of human review when needed.

How does Autonomous Accounting actually work?

Our mission at Puzzle is to focus on the immediate availability of data in financial statements, metrics, and insights, versus other accounting systems that use an “approval” driven approach to bookkeeping (such as QuickBooks), Puzzle is built so that all data is instantly incorporated into essential insights and reports, regardless of human review. We prioritize the immediate incorporation of this data because we recognize that its real-time inclusion into insights is essential for key business decisions. At the same time, we manage the risk that data is materially incorrect by having controls in place that let the reader know how much review and approval has gone into a specific month. With this visibility, the user will know how much confidence to put on the data and either move forward with their decision making and take the time to add additional review into the data. 

To do this, we re-imagined and re-built the core accounting system from the ground up, rather than building on top of an existing general ledger. 

We built a double-entry, real-time, immutable general ledger with a streaming, API-based financial data platform. Our direct API connections with the most modern fintech stack (Stripe, Mercury, Ramp, Gusto, Rippling, alongside other APIs) allow us to ingest robust metadata to represent all of a company’s transactions and payroll events. We connect all other financial transactions in a business using standard integrations via Plaid. Our APIs work together to create the real-time metrics and insights: 

  • Financial transactions we receive via API are turned into accounting entries in Puzzle’s general ledger.
  • Because we have direct APIs, this enables a higher level of confidence compared to other accounting systems when it comes to the categorization of transactions. The risks of disconnection and reconnection are lower. Direct APIs also give us the ability to determine other accounting treatments, such as matching bank transfers, bill payments, and credit card payments. They enable automation around balance integrity checks and reconciliations.

Because we automatically turn data feeds from APIs into first-draft accounting data, a user can log into their Puzzle account at any time and access real-time financial statements, as well as metrics like burn, runway, ARR, MRR, and other insights. This is possible even if the user has not fully reviewed all transactions or transaction categories. an approach that is not always encouraged by other accounting platforms.

Since an early-stage startup typically has not yet reached a level of complexity where it needs robust controls over financial information, we have optimized Puzzle for the real-time recording of transactions without waiting for reviews and approvals to be completed. This allows a founder to instantly access up-to-date financial insights without waiting on formal accounting processes.

What are the benefits of Autonomous Accounting? 

  • Autonomous Accounting empowers business owners and founders to understand their company's financial health and manage accounting all in their natural language without the need to understand complex accounting processes. Learn more about Autonomous Accounting for founders.
  • Autonomous Accounting helps free accountants from manual tasks, allowing them to shift their focus towards high-value work like analysis, strategy, and providing valuable business recommendations. By automating manual workflows, Autonomous Accounting unleashes their potential to drive greater insights to their clients or within organizations.

How does Puzzle scale with startups?

Puzzle was built as an accounting platform to scale alongside a startup as they grow. Here’s how that is possible: 

✅ Our robust data architectures support volume and quality as a company scales 

Our resilient data architectures are built to handle a wide range of transaction volumes from different sources (banks, payroll providers, financial data intermediaries, and more) while seamlessly accommodating a few transactions per month or scaling up to millions of transactions as a company's volume grows. 

Puzzle’s unique ability is in handling volume at scale alongside capabilities for advanced analytics and data visualization by leveraging data within partner API frameworks. Our system also keeps data which allows for a higher degree of confidence when diving into transaction details.

For instance, our users can access summarized general ledger entries, but also receive access to all of the underlying source data in the system, if desired. This provides an efficient way to tie out the data, run analytics, or re-arrange the data in another view.

Puzzle’s data comes with a full audit log and traceable historical details for all data—so you can see where a number came from and always be confident you’re looking at the most up-to-date information. To access this type of data in another accounting platform, you’d most likely need to find the correct version of a spreadsheet you used (e.g.`final-v2-v2` anyone?), which is a waste of valuable time.

Our features are built with a startup-first mentality 

At Puzzle, we stand out from traditional legacy accounting systems with our startup-first approach. Recognizing that there was a gap in the market, our platform was specifically designed to cater to the unique needs of startups, offering a range of tailored features. We prioritize metrics, insights, and reports that matter to startups, such as burn rate, ARR, MRR, and runway, among others.

Our customer-centric approach also keeps us closely connected with our customers and the startup community, allowing us to understand their requirements and challenges. By actively listening and engaging with our users, we continuously build features that address their immediate needs and empower their financial journey. We understand the distinct needs of startups and have crafted our platform to cater to their financial requirements and growth journey.

Do other accounting systems offer Autonomous Accounting? 

Puzzle stands apart from QuickBooks and other legacy accounting software systems with a clear distinction in our core mission and purposeful infrastructure we have established from the very beginning to uphold that mission. Our accounting software was purpose-built from day one to unlock higher-value work and facilitate intelligent decision-making, guided by two fundamental product values:

  1. Instant accessibility to financial data creates timelier and smarter decision-making
  2. Smarter automation leads to time savings and higher-value work

    Learn more about our product values

So Autonomous Accounting is not about AI?

Autonomous Accounting is not an interchangeable term for AI, but recent advancements in AI have paved the way for Autonomous Accounting to be possible. To date, Puzzle has developed AI-augmented workflows, in which we use GPT-4 to run specific workflows. Here’s an example of how an AI-augmented workflow runs in Puzzle: 

  1. First, transactions are ingested and validated. We run data integrity checks on transactions we receive through API. We structure, clean, and de-duplicate all data before automatically feeding the transactions into the general ledger. 
  2. Then, PuzzleBot works its magic: Our PuzzleBot automatically categorizes every transaction to a general ledger account. Auto-tags are based on source data that is set up by Puzzle’s system rules, logic, machine learning, or AI. PuzzleBot has the ability to auto-categorize 90% of transactions. 
  3. AI Categorizer: For the last 10%, that’s where Puzzle’s new AI-powered workflow comes in to help. AI Categorizer notifies the owner of the account about transactions that need extra context to categorize, such as “Check”. The owner will receive a prompt in the Puzzle app or email, like the email below:

AI Categorizer sends an email to the user

The user can reply to the email in plain English (i.e., “This was a new laptop for David”). AI Categorizer will then assign the right category to the transaction using the context that has been provided. 

As AI continues to progress, so will Puzzle. Accountants and founders can anticipate a host of additional benefits and seamless integration into Puzzle’s workflows, empowering them to prioritize high-value tasks and optimize time management.

So you’re not using AI to replace accountants? 

Absolutely not! We have a mandate to empower and enable accountants using AI technology, but not replace. Our software is designed to automate significant portions of common processes, such as transaction categorization to assist humans rather than replace them. 

Autonomous Accounting is not just a mindset, but a tangible system that is available today on Puzzle, relying on the power of technology, rules, logic, machine learning, and AI to enable automation and streamline accounting processes. Human intervention is essential for certain transactions, and Autonomous Accounting may not be perfect, but that's why we have developed tools to facilitate efficient human review and handling. At the heart of our product is the desire to make accounting a less burdensome process, making everybody’s lives easier and enhancing the accounting process and profession as a whole.

We hope we’ve answered all your questions about Autonomous Accounting! This is the just the start. Just because we’ve made the categorization process more efficient, we are not done with Autonomous Accounting. There will much more to look forward to as continue to roll out new features and build new workflows to help user gain efficiencies in more complex processes. We hope you now understand our philosophy to how we’re building Puzzle. We are building a foundational accounting system that will continue to get better, save you and your team time, and make life easier. 

Want to learn more? Read customer testimonials, sign up to try Autonomous Accounting for yourself, or reach out to me anytime.

Share this post
Andrew Robinson, CPA
Product & Accounting @ Puzzle

Related posts