Quillon Secures $1.5 Million for Audit-Grade AI in Technical Accounting

Quillon, an AI platform based in Sofia, has announced the closing of a $1.5 million pre-seed funding round. The company, formerly known as Acclara AI, focuses on technical accounting and financial reporting, sectors demanding extreme accuracy and verifiability. The round was led by 42CAP, with participation from angel investors affiliated with NVIDIA and Roblox, who had previously supported the company before the institutional round.

Founded in 2023 by Nikolay Dakov, Ivaylo Stefanov, and Atanas Dobrev, Quillon operates from Sofia with a presence in San Francisco. Its mission is to address the complexities of technical accounting, a highly specialized area involving the interpretation of complex accounting standards and the preparation of detailed memos designed to withstand scrutiny from auditors and, in some cases, the US Securities and Exchange Commission (SEC). This type of analysis is critical for major financial decisions, such as mergers, debt restructurings, and revenue recognition, where errors can lead to restatements, regulatory inquiries, and significant financial impacts.

The Auditability Challenge in the AI Era

Many accounting professionals have begun integrating general-purpose AI tools into their workflows to manage increasing workloads. However, these tools often present a significant gap in terms of auditability. They can generate unverifiable or incorrect citations and fail to provide a clear, traceable connection between the conclusions produced and the underlying accounting standards. This lack of transparency and verifiability represents an insurmountable obstacle for companies operating in regulated industries, where compliance and data sovereignty are absolute priorities.

For organizations evaluating the deployment of Large Language Models (LLM) in sensitive environments, the ability to audit and trace decisions is a non-negotiable requirement. Adopting AI solutions that do not guarantee these features can expose companies to regulatory and reputational risks. This is a key factor driving many enterprises to consider self-hosted or air-gapped architectures, where control over data and inference processes is maximized.

Quillon's Solution: A Proprietary Knowledge Graph for Compliance

Quillon directly addresses this gap through a platform built on a proprietary knowledge graph of accounting standards, integrated with EDGAR, the SEC's system for collecting, storing, and providing public access to company filings. This architecture allows users to navigate accounting questions step by step, linking each conclusion directly to the source material. The platform combines research, contract analysis, peer benchmarking, and memo drafting into a single, cohesive workflow.

The system is designed to maintain human oversight at every stage, enabling accountants to review, edit, and validate each step. This approach not only increases efficiency but also ensures that every output is fully traceable to its original source, a fundamental requirement for auditability. As highlighted by Nikolay Dakov, co-founder and CEO of Quillon, the platform was developed to provide a workspace where AI performs the analysis, but the accountant maintains control at every step, with every claim traceable back to the exact paragraph in the standards.

Future Prospects and Deployment Implications

Initially, Quillon's platform will focus on technical accounting memos, which underpin nearly all financial reporting decisions within public companies. The company plans to expand into broader financial reporting workflows, including quarterly and annual disclosures. The funding secured will be used to expand engineering and go-to-market capabilities, supporting the transition from a research-oriented product to a platform that produces finalized deliverables and performs end-to-end accounting workflows.

Quillon's approach highlights a growing trend in the AI sector: the development of specialized solutions that meet specific compliance and control needs in regulated environments. For CTOs and infrastructure architects evaluating LLM deployment options, the ability to ensure data auditability and traceability is a decisive factor. This type of solution strengthens the argument for on-premise or hybrid deployments, where data sovereignty and control over inference processes can be managed with greater granularity, reducing the risks associated with using generic cloud services for critical workloads. AI-RADAR offers analytical frameworks to evaluate the trade-offs associated with these deployment decisions, available at /llm-onpremise, to support companies in choosing the architecture best suited to their security and compliance needs.