Stéphan Donzé is the founder and CEO of Aodocs, AODOC is a cloud-based document management platform that converts enterprise content into actionable intelligence. Unlike traditional systems limited to basic storage, AODOC combines powerful document control with workflow automation, allowing businesses to simplify complex processes across industries.
From sales advice and claim management to engineering and manufacturing, AODOCS uses generated AI to extract structured data from unstructured documents (scale).
What prompted you to launch AODOC in 2012? Have you seen a specific issue in document management that was not resolved at the time?
It comes from the concept of bringing the business-to-consumer (B2C) technology pioneered by Google to the enterprise. This includes cloud-based, serverless and automatic scaling technologies that were missing in enterprise software at the time.
At the time, Google advocated applications to adapt to the cloud, thus staying away from thick clients and embracing browser-based and serverless architectures. Unlike Microsoft and Amazon at the time, Google’s position was that cloud applications should adopt a very different architecture from the premise software. I saw an opportunity to create new things by combining old methods.
Early adopters of cloud technology in enterprises that adopt Gmail and Google Drive want to extend these benefits to email and collaboration and to critical business documents. There is a lack of solutions to the needs in the market because traditional document management systems are related to old software and architectures. This gap provides an opportunity to work with these early adopters to build something new from scratch.
Many enterprise SaaS companies struggle with customer adoption. How did you convince major companies like Google or Veolia to trust AODOC early on?
Allowing customers to save documents in their own cloud storage is an important factor in a compelling business. AODOCS manages documents, including shared permissions and business processes, but documents remain in the customer’s environment (originally Google Drive, later Azure, Google Cloud Storage, and Amazon S3). This fundamental feature reduces perceived risk to customers and makes them more comfortable with a small and young company.
Another key reason is close customer relationships. We build and develop together with our customers, especially in the early stages. This proximity and responsiveness builds trust in solving problems and improving and further reduces the perception of risk.
Ultimately, this convinced Google, Veolia, Ascension, Pinnacol Assurance and others to adopt us.
Looking back at your 13-year-old leadership of AODOC, what is the most critical decision that leads to the company’s success?
The decision to keep customer documents in their own cloud storage is a critical factor in our success and remains a key differentiator.
It is also crucial to develop products by being very close to customers. Everything we build is inspired or validated by customer needs. We avoid building features without explicit customer use cases.
We have been good at professional services to better address common and replicable needs across our customer base to identify common and replicable needs that benefit the entire customer base, which is what we are good at.
Born in the cloud, Aodocs has now evolved into an AI-driven solution. What are the biggest challenges in making this transition?
This is not a transition, but a continuity. Born in the cloud intrinsically, making AODOC easy to integrate with new, cloud-based technologies including AI.
AI models such as Chatgpt and Google Gemini are cloud-based, making it natural for AODOC to interact with them and add AI capabilities to our document processing platform.
We think of AODOC as the “spine” or “assembly line” of the document process, where AI is just a new “limb” or “module” that can be inserted to enhance the platform. We did not foresee the specific emergence of AI in 2022, but our cloud-native architecture is designed to scale and integrate, making AI adoption a natural step.
How does AODOC balance automation with human supervision to ensure that compliance and accuracy of human verification are not eliminated?
We provide our customers with tools to use AI in various steps in their documentation process. The customer ultimately decides how much autonomy the customer makes based on the specific circumstances.
Customers can configure their processes in any AI step for complete automation, complete human comment or a mix of the two. We recommend identifying simple cases suitable for full AI automation (e.g., missing documents in detection requests), while for complex cases, AI can often speed up human review by generating summary of complex documents.
Essentially, we give customers 100% control over the level of human supervision by providing configuration tools.
AODOC is moving its platform to an AI proxy focused on document processing. What drives this decision, and how AI agents improve traditional document management systems?
AI Agent is a new feature that can be integrated into our existing documentation “assembly lines”.
The main driver of our excitement for AI agents is their potential to bring huge speed to specific tasks. For simple tasks, such as rejecting incomplete documents or automatically accepting low-value, direct claims, AI can achieve complete automation.
For complex tasks, AI agents can greatly accelerate processing by summarizing long documents, allowing humans to view critical information faster. AI excels in ingesting a lot of information and providing concise summary, saving a lot of time for human reviewers. This increased speed and efficiency enhances the value of the AODOCS platform.
Business search remains a major challenge, and companies are working hard to retrieve relevant documents. How does AODOC’s AI-PARE solution better solve the discoverability problem compared to existing enterprise search tools?
AI chatbots represent significant improvements to traditional enterprise search by allowing users to ask questions in natural language and obtain direct answers. However, they have one major flaw: they cannot distinguish between valid, latest documents and drafts, incomplete or outdated files.
We address this challenge by ensuring that AI agents only use the right information to provide answers. Traditional searches provide users with a list of potential matches, allowing them to identify the most relevant documents (such as distinguishing between signed contracts and drafts).
In contrast, AI chatbots provide confident answers, but hide the basic search process. This increases the risk of using outdated or incorrect information, as the user does not have visible documents for the chatbot to formulate their responses.
Our role is to curate and ensure that all files in the knowledge base are properly verified and up-to-date. This allows them to safely use AI agents as information sources. By ensuring that AI-driven search relies solely on trusted and controlled information, we significantly reduce the risk of inaccurate responses.
With generative AI becoming a key part of document processing, which specific AI technologies or models can leverage AODOC to improve accuracy and reduce manual work?
The LLM model performs well in transforming existing content, such as aggregating complex documents, translated texts, and classified files. The risk of hallucination for these tasks is low, which is why our AI usage strategically focuses on these activities.
AODOC mainly uses AI to re-draw, re-fix and transform existing text, rather than creating new content. When using the correct material, the possibility of errors is kept low. Similarly, our data extraction capabilities (identify deadlines or critical information in a file) are based on existing content to minimize inaccuracy or “illusion.”
These AI capabilities often increase efficiency by accelerating information intake and supporting human reviewers. This creates tangible value while carefully avoiding the risks associated with AI-generated content.
There is another advantage. AODOCS seamlessly integrates with any LLM, giving our customers the freedom to choose their preferred provider. This flexible architecture also ensures that we can benefit from the inevitable AI innovation over the coming months and years, allowing us to quickly merge future models to benefit our customers.
One of the biggest AI challenges in document processing is hallucination or inaccurate responses. How does AODOC ensure that AI-generated insights are trustworthy and reliable?
The key to ensuring trust and reliability is the foundation. We ensure that AI is used to reshape or transform information from trusted sources.
In document processing, AI is required to summarize or translate existing documents to minimize the chance of hallucination. In an AI chatbot or agent, the system provides extracts of controlled documents and indicates that answers are generated only from these extracts – LLM must always refer to the source of their generated answers without using general world knowledge or external information. This prevents AI inventions or provides information that is not explicitly present in the verification document.
With AI-Converting Enterprise Software, where do you see AODOC in the next five years? What role will AI agents play in the future of document management?
Our core business as a document management platform will not change. Our mission remains to ensure trustworthiness, currency, proper business process compliance and traceability of traceability information.
AI is both a new need for controlled information (feeding trustworthy data to AI chatbots) and an accelerator of business processes involving documents. AI can speed up tasks such as summarizing, translating, and sorting documents, making human users more productive.
Over the next five years, we aim to provide companies with tools to easily apply AI so that it makes sense in enhancing their business processes. AI agents will play a key role by amplifying and accelerating our core tasks: providing trustworthy information, ensuring the right version of the document is used, helping to avoid human errors and facilitating access to business processes. We want to help companies effectively and strategically leverage AI to increase productivity.
Thank you for your excellent interview, and readers who hope to learn more should visit AODOC.
