Stéphan Donzé is the founder and CEO of AODocs, a cloud-native doc administration platform that transforms enterprise content material into actionable intelligence. Not like legacy methods restricted to fundamental storage, AODocs combines strong doc management with workflow automation, enabling companies to streamline advanced processes throughout industries.

From gross sales proposals and claims administration to engineering and manufacturing, AODocs makes use of generative AI to extract structured information from unstructured paperwork—at scale.

What impressed you to begin AODocs in 2012? Was there a particular drawback you noticed in doc administration that wasn’t being addressed on the time?

It got here from the idea of bringing Enterprise-to-Shopper (B2C) applied sciences, pioneered by Google, to the enterprise. This included cloud-based, serverless, and routinely scaling applied sciences that had been absent in enterprise software program on the time.

On the time, Google was advocating for functions to adapt to the cloud by transferring away from thick purchasers and embracing browser-based and serverless architectures. Not like Microsoft and Amazon on the time, Google’s stance was that cloud functions ought to undertake a really completely different structure than on premise software program. I noticed a chance to create one thing new by combining previous and new approaches.

Early adopters of cloud applied sciences in enterprises, who had adopted Gmail and Google Drive, needed to increase these advantages past e mail and collaboration to business-critical paperwork. There was a scarcity of options available in the market to fulfill this want, as conventional doc administration methods had been on-premise with previous software program and architectures. This hole introduced a chance to construct one thing new from scratch on the best cloud structure, working with these early adopters.

Many enterprise SaaS firms wrestle with buyer adoption. How did you persuade main enterprises like Google or Veolia to belief AODocs early on?

Permitting prospects to maintain paperwork of their  personal cloud storage was a big think about convincing enterprises. AODocs manages paperwork, together with sharing permissions and enterprise processes, however the paperwork stay inside the buyer’s atmosphere (initially Google Drive, later Azure, Google Cloud Storage, and Amazon S3). This elementary attribute lowered the perceived danger for purchasers, and made them extra comfy working with  a small and younger firm.

One other key cause was shut buyer relationships. We co-built and co-developed with prospects, particularly within the early phases. This proximity and reactivity in fixing issues and making enhancements constructed belief and additional lowered the notion of danger.

This was in the end what satisfied firms like Google, Veolia, Ascension, Pinnacol Assurance and others to undertake us.

Wanting again at your 13+ years main AODocs, what had been a number of the most important selections that led to the corporate’s success?

The choice to maintain buyer paperwork in their very own cloud storage is an enormous think about our success and stays a key differentiating issue.

Growing the product by staying very near prospects was additionally vital. All the things we construct is impressed by or validated by means of buyer wants. We keep away from constructing options with out a clear buyer use case.

Our capability to type by means of buyer requests to establish frequent and replicable wants that profit your complete buyer base, versus very particular particular person requests higher addressed by skilled companies, is one thing we have been good at.

AODocs was born within the cloud and has now advanced into an AI-powered resolution. What had been a number of the largest challenges in making that transition?

It is not a lot a transition however a continuity. Being born within the cloud inherently made AODocs simple to combine with new, cloud-based applied sciences, together with AI.

AI fashions like ChatGPT and Google Gemini are cloud-based, making it pure for AODocs to interface with them and add AI capabilities to our doc processing platform.

We view AODocs as a “backbone” or an “meeting line” for doc processes, the place AI is just a brand new “limb” or “module” that may be plugged in to reinforce the platform. We didn’t foresee the precise emergence of AI in 2022, however our cloud-native structure was designed for extensibility and integration, making the adoption of AI a pure step.

How does AODocs steadiness automation with human oversight, guaranteeing compliance and accuracy with out eradicating human validation?

We offer the instruments to prospects to make use of AI at numerous steps of their doc processes. The choice of how a lot autonomy to offer AI is in the end made by the shoppers on a case-by-case foundation.

Clients can configure their processes for full automation, full human evaluation, or a mixture of each at any AI step. We advocate figuring out simple circumstances appropriate for full AI automation (e.g., detecting lacking paperwork in a request), whereas for advanced circumstances, AI can be utilized to hurry up the human evaluation, sometimes by producing summaries of advanced paperwork.

Basically, we give prospects 100% management over the extent of human oversight of their processes by offering the configuration instruments.

AODocs is shifting its platform to give attention to AI brokers for doc processing. What drove this determination, and the way do AI brokers enhance upon conventional doc administration methods?

AI brokers are a brand new functionality that may be built-in into our present “meeting line” for paperwork.

The first driver for our pleasure about AI brokers is their potential to carry vital pace to particular duties. For easy duties like rejecting incomplete paperwork or routinely accepting low worth, simple claims, AI can allow full automation.

For advanced duties, AI brokers can dramatically speed up processing by summarizing lengthy paperwork, permitting people to evaluation key data a lot sooner. AI excels at ingesting giant quantities of knowledge and offering concise summaries, saving human reviewers vital time. This elevated pace and effectivity improve the worth of the AODocs platform.

Enterprise search stays a significant problem, with firms struggling to retrieve related paperwork. How does AODocs’ AI-powered resolution deal with discoverability points higher than present enterprise search instruments?

AI chatbots signify a significant enchancment over conventional enterprise search by permitting customers to ask questions in pure language and obtain simple solutions. Nonetheless, they’ve a big flaw: they can’t distinguish between legitimate, up-to-date paperwork and drafts, incomplete, or out of date information.

We handle this problem by guaranteeing AI brokers use solely the best data to offer solutions. Conventional search presents customers with a listing of potential matches, permitting them to establish probably the most related doc (reminiscent of distinguishing between a signed contract and a draft).

In distinction, AI chatbots provide assured solutions however conceal the underlying search course of. This creates an elevated danger of utilizing outdated or incorrect data, as customers haven’t any visibility into which paperwork the chatbot used to formulate its response.

Our function is to curate and assure that every one paperwork within the information base have been correctly validated and are up-to-date. This makes them protected for AI brokers to make use of as data sources. By guaranteeing AI-powered search depends solely on trusted and managed data, we considerably scale back the danger of inaccurate responses.

With generative AI turning into a key a part of doc processing, what particular AI applied sciences or fashions does AODocs leverage to enhance accuracy and scale back guide work?

LLM fashions excel at reworking present content material, reminiscent of summarizing advanced paperwork, translating textual content, and categorizing information. These duties have a really low danger of hallucinations, which is why our AI utilization strategically focuses on these actions.

AODocs primarily leverages AI for rephrasing, repurposing, and reworking present textual content somewhat than creating new content material. When working with the best materials, producing summaries, translating paperwork, and categorizing content material all preserve a low likelihood of error. Equally, our information extraction capabilities—figuring out due dates or key data from paperwork—stay grounded in present content material to reduce inaccuracies or “hallucinations.”

By accelerating data ingestion and supporting human reviewers, these AI capabilities usually enhance effectivity by a big issue. This creates tangible worth whereas rigorously avoiding the dangers related to AI-generated content material created from scratch.

There’s an extra benefit. AODocs integrates seamlessly with any LLM, giving our prospects freedom to decide on their most popular supplier. This versatile structure additionally ensures we are able to rapidly incorporate future fashions as they turn into out there, positioning our prospects to learn from the inevitable AI improvements rising within the months and years forward.

One of many largest AI challenges in doc processing is hallucinations or inaccurate responses. How does AODocs make sure that AI-generated insights are reliable and dependable?

The important thing to making sure trustworthiness and reliability is grounding. We make sure that AI is used to rephrase or remodel data from trusted sources.

In doc processing, AI is requested to summarize or translate present paperwork, minimizing the prospect of hallucination. In AI chatbots or brokers, the system is supplied with extracts of managed paperwork and is instructed to generate solutions solely from these extracts—the LLM should at all times quote the sources from which it generated its reply—with out utilizing common world information or exterior data. This prevents AI from inventing or offering data not explicitly current within the validated paperwork.

With AI reworking enterprise software program, the place do you see AODocs within the subsequent 5 years? What function will AI brokers play in the way forward for doc administration?

Our core enterprise as a doc administration platform is not going to change. Our mission stays to make sure the trustworthiness, forex, correct enterprise course of adherence, and traceability of managed data.

AI is each a brand new want for managed data (to feed reliable information to AI chatbots) and an accelerator of enterprise processes involving paperwork. AI can pace up duties like summarizing, translating, and categorizing paperwork, making human customers extra productive.

Within the subsequent 5 years, we intention to offer the instruments for firms to simply apply AI the place it is smart to reinforce their enterprise processes. AI brokers will play an important function by amplifying and accelerating our core missions: offering reliable data, guaranteeing using the right doc variations, serving to keep away from human errors, and facilitating entry to enterprise processes. We wish to assist firms use AI successfully and strategically to spice up productiveness.

Thanks for the nice interview, readers who want to study extra ought to go to AODocs.