AI bookkeeper Docyt debuts precision-trained AI agents to accelerate accounting automation – News

Docyt AI Inc. is building on the launch last year of its artificial intelligence-powered accountant GARY, unveiling a new AI engine that it says is purpose-built to automate complex bookkeeping workflows.

The Docyt High Precision Accounting Intelligence engine is essentially an AI agent designed specifically for accounting tasks, and its launch follows a fresh funding round for the startup, which has just raised $12 million in a pre-Series B investment led by Pivot Investment Partners.

GARY, which is an abbreviation of “Generative Accounting Retrieval sYstem,” was claimed to be the world’s first AI Bookkeeper when it launched last year, combining generative AI, precision AI and predictive AI algorithms into a single model that’s geared toward tasks such as expense management, revenue accounting and financial reporting.

Docyt says that dozens of accounting firms are using GARY to automate what’s known as the “month-end” closing process – which is where finance teams attempt to account for all of the previous month’s transactions. It’s an arduous process that involves reviewing, recording and reconciling sometimes thousands of accounting records.

Now, Docyt is building on GARY’s capabilities with the Docyt HpAI engine, combining its best large language models with “precision-trained” AI agents that have been tailored for specific accounting tasks, such as reconciliation, categorization, anomaly detection and the month-end close.

These new agentic capabilities are being made available within Docyt’s Accountant Copilot, and they’re uniquely able to simulate the intelligence of human bookkeepers. Because of this, organizations will be able to delegate various bookkeeping tasks to AI agents that would normally be handled by junior accountants. They’re able to tabulate data, prepare analytics, flag any anomalies in financial records, and automate the entire month-end close.

Docyt states that its bookkeeping agents share a revamped foundational AI architecture trained on 128 billion accounting data points spanning more than 20 industries.  It claims that this is by far and away the most “contextual depth” of any accounting model released so far, and enables the Docyt Copilot to not only automate, but also intelligently learn as it completes its work.

Co-founder and Chief Executive Sid Saxena said his company has spent more than five years building high-quality synthetic datasets, using a team of expert accountants skilled in high-quality data labeling to capture the complexity of real-world bookkeeping.

“This dataset is a critical component of our HpAI architecture, enabling it to deliver precise, context-aware automation,” he added. “We are now bringing this foundational AI architecture to accounting firms in the form of Docyt AI Copilot, enabling them to deliver client bookkeeping at scale, and with precision.”

Docyt does not cite any recognized benchmarks regarding the performance of Docyt HpAI, but it nevertheless claims that its growing customer base is seeing a three-times increase in client capacity as a result of the automation it provides. It also claims a 90% reduction in review times per client, and says it enables “enhanced audit readiness” due to its built-in transparency and confidence scoring mechanisms.

J M Keehn Accountancy Managing Partner Christa Wells said Docyt’s AI agents have enabled her company to scale dramatically without increasing its staff headcount.

“Docyt gives us instant clarity and saves us roughly 15 hours per week by showing exactly what needs attention, so there’s no more digging through accounts,” she explained. “It’s also given us the cleanest financials we’ve ever seen, with an unprecedented level of accuracy that requires almost no intervention before moving on to the next steps.”

Image: Docyt