Reflections on the launch of AIDA 20.0

Over the past couple of weeks Giorgio (our CEO) and myself have been racking up the miles as we have started spreading the good news of the launch of AIDA 20.0.

Firstly we were present at the AIIM Global Summit 2026 in Baltimore in the US at the end of April where we were announcing the launch of AIDA 20.0. Then at the beginning of May we were over to Copenhagen courtesy of our friends at Selectec for the legendary annual Selectec Nordic Geek Meet!

There was a good deal of engagement and interest in what we had to say, but equally some good insights into how the world of documents is developing, so we thought it would be worth sharing some of our observations.

AIIM Global Summit 2026

AI: from the sidelines to the mainstream

It was as ever a very well attended event with the usual core cohort of very experienced information managers, records managers and knowledge managers, but also a seemingly growing number of information security professionals. The latter was an indication of how the industry in general and this event in particular is evolving. Indeed the initial plenary session was given by the Head of Enterprise AI Governance at AstraZeneca and covered the key areas corporates should be considering when implementing an AI governance framework. 

It was a clear indication, if one were needed, that AI has moved from a niche, peripheral support tool, to one which permeates every element of a company’s operations, as executives seek to benefit from all the upsides that AI offers, whilst managing any associated risks. And for those organisations which try to technically block their employees from using AI/LLMs, there seems to be widespread acceptance that employees just simply use them on their phones. So whether we like it or not, there was broad consensus that AI is here to stay. It’s now a question of how we can best work with it to get the most out of it.

Content is king

And regarding the downsides, one term which kept cropping up was ‘Garbage in. Garbage out.’ I learnt the term ‘backdoor poisoning’ which is a way of skewing the results of LLMs and can be done with an extremely small percentage of erroneous information.

There was also talk about RAG (Retrieval-Augmented Generation), which allows AI to answer questions and perform tasks using an organisation’s own documents and knowledge, combining the reasoning power of generative AI with the reliability of verified business data.

Other buzzwords that were flying about included ‘hallucinations’ and ‘sycophancy’, but we also heard from one speaker who maintained that AI doesn’t have ‘hallucinations’, in fact that AI isn’t intelligent as it doesn’t question its content set. He maintained that LLMs are amazing technology but that the Information Management industry needs to catch up with it.

Humans in the loop

And to do that humans will remain a key cog in the wheel for the foreseeable future, even if their role will likely evolve somewhat over time to one where they will be responsible for curating the information set and reviewing any content which is generated from that basis. Indeed we learnt that there are multiple variants of ‘human in the loop’ (e.g. conditional, in command, on standby, override-only) as well as the importance of a ‘kill switch’ to terminate AI processes which are generating questionable results.

Intelligent Document Processing

IDP came up a number of times throughout the event with Alan Pelz-Sharpe of Deep Analysis saying that “IDP is the low-hanging fruit of AI" with proven results over years both in terms of significant productivity gains and tangible ROI.

That said we also heard of one ‘proof of concept’ IDP project which had taken some two years and seemingly only had one successful use case. To me it was astonishing that a proof of concept could cost so much time - and presumably money! - and still not be particularly conclusive. A two month PoC is already quite a long time for us and in our experience has always been successful.

Unstructured to structured data

One key theme running through the event was the importance of managing unstructured data and more specifically of turning it into structured data, which is obviously something that we know a little bit about at AIDA! Interestingly there was some pretty impressive scanning technology on display (from memory, one device which could scan colour, double-sided, 87 pages per minute!) but a recognition that was only a small part of the document processing value chain. What happens to the document after it is scanned? Where does the data go?

Agentic AI

There did appear to be broad awareness of what an AI agent is and what it does above and beyond an LLM (ie. takes decisions and actions). It was also a regular topic of conversation but what we noticed was that largely people were talking about theoretical use cases. What we think set our presentation apart was being able to show real-life case studies of agentic AI in action. Indeed we had great engagement from audience-members as brows were initially furrowed and, we think, figurative light bulbs started to light up! The key points we closed with were:

• AI agents should build on existing processes, systems and documents.
• That actually the barrier to entry is really rather low…
• …but the return on investment is high - and fast!

which seemed to resonate and reassure the attendees.

The Selectec Nordics Geek Meet 2026

This was a more intimate, targeted affair with our partner Selectec inviting a good few dozen of their closest partners in the Nordics to a couple of days of presentations and conversations broadly around the changing world of print management.

Selectec presented their ‘Efficiency Framework’ which takes a holistic view of document processing beyond the printer/scanner/MFD taking into account capturing, processing, storing, searching and sharing. Stu Brookes from PaperCut highlighted the importance of an effective print management tool to manage an organisation’s printers/scanners/MFDs centrally - and securely. And we presented - and demonstrated - how AIDA does everything it does, seamlessly fitting in with the above.

As with the AIIM event, there was really good engagement from the audience with our session overrunning due to the amount of questions we received - sorry Ben! All the feedback we received both during and after our session was that attendees were really impressed with AIDA’s capabilities. It’s always nice to impress a room full of techies with your technology!

One other thing I personally learnt speaking with the people present was how large Enterprise Content Management (ECM) solutions are in many cases not helping solve a lot of the problems with document management and processing, in spite of their often very high price tag. It seems as though it might be a market which is ripe for disruption!

In summary

Looking back on the two events, there seem to be some common threads we thought it would be good to extract/capture, if you’ll excuse the pun!:

• AI is here to stay but needs careful human intervention in both the set-up and operations.
• The content/data set on which your LLMs operate is of paramount importance.
• Print is not going away as fast as we thought…
• …but companies need to take a holistic view of document lifecycles.
• AI Agents are here, are not such a big deal as they seem and can add rapid and significant value to an organisation's operations.

All of these take-outs are very consistent with how we have designed AIDA over the past years with our hybrid-AI and document validation approach and culminating most recently in our release of AIDA 20.0 with its AI agents. This wouldn’t have been possible if it hadn’t been for our building an architecture - with human intelligence and insight! - which supports agentic AI from the ground up. It is this, together with our approach of continuous deployment by design, which will ensure we continue to bring innovations to our users ahead of the market for years to come.

We hope to take many of you with us as we set out on this exciting new chapter!