Understanding the Organisation, Architecture and Ambition Behind a Different Approach to Enterprise AI

EXECUTIVE DIALOGUE · EDITORIAL CASE STUDY

After three Signals examining AskDiana, one question remained surprisingly difficult to answer. What exactly is AskDiana? It is easy to describe it as an enterprise AI platform. It can interact with users, connect organisational knowledge and support different applications. In some situations, users may experience it through something resembling an AI agent. But after looking more closely at its architecture, that description increasingly feels incomplete.

AskDiana is not another foundation model. Nor is it simply a chatbot or an AI agent competing to provide better answers. Its ambition is broader: to create a governed organisational foundation through which companies and institutions can connect their knowledge, determine who may access it, choose which AI models operate upon it and build applications on top. That distinction matters.

Access to a powerful language model does not automatically give an organisation an AI capability. Organisational knowledge must be connected. Permissions must be defined. Responsibility must remain somewhere. Models change. Information becomes outdated. Regulations differ. And organisations may eventually want to change technology providers without rebuilding everything around them.

AskDiana attempts to build the architecture around those problems.

The three preceding Signals explored that proposition from the outside. For this concluding Executive Dialogue, Altair Media invited the people behind AskDiana to explain the organisation, architectural choices and motivations behind it.

Not simply: what does the product do? But: why build enterprise AI this way?

🟦 Why Does AskDiana Exist?

For Srikanth Gaurav, founder of 4Square Innovations, the starting point predates the current generative AI boom.

His experience in enterprise technology repeatedly exposed the same divide. Large organisations could afford infrastructure, expertise and governance. Smaller organisations were often left with simplified alternatives—or without access to comparable capabilities altogether.

Generative AI appeared to be recreating that divide.

Altair Media — What market failure convinced you that AskDiana should exist?

Srikanth Gaurav:
The market did not need another place where people could type a question and receive an impressive answer. The gap we saw was between access to AI and having an organisational AI capability.

A company can subscribe to a powerful model tomorrow. But that does not automatically connect its approved knowledge, permissions, workflows, accountability structures or operational systems.

For large organisations, those additional layers can be built through significant investment. Smaller businesses, NGOs and public organisations often have fewer options. That was the problem we wanted to address.

The ambition behind AskDiana therefore begins with an important distinction. Using AI is not the same as governing AI.

🟦 So What Is AskDiana?

Perhaps the simplest way to understand AskDiana is not to begin with its interface. Begin underneath it. At its foundation sits organisational knowledge: documents, policies, procedures, internal information and other sources an organisation considers authoritative. AskDiana seeks to create a governed layer around that knowledge.

On top of it sits Genius², the reasoning and verification architecture developed by 4Square Innovations. SDKs and APIs allow applications and integrations to be built around that foundation. Only then comes the interface.

A user might ultimately encounter AskDiana through a conversation, dashboard, widget, specialised application—or an AI agent. The interface can change. The governed foundation is intended to remain.

Altair Media — Which architectural principle is most fundamental to AskDiana?

Tony Larcombe, CTO:
Knowledge Governance.

Models will change. Interfaces will change. Deployment technologies will change. But organisations still need to know which information is authoritative, who owns it, who may access it, when it becomes outdated and what may be shared outside the organisation.

Technology can help enforce those decisions, but it cannot make all of them. People still have to decide what the organisation trusts. That qualification is important.

AskDiana does not claim that architecture can magically govern knowledge. Rather, its proposition is that architecture can make governance operational. Humans establish authority and responsibility. Technology helps preserve and enforce those decisions.

🟦 What Makes AskDiana Different?

Enterprise AI is already becoming crowded. Organisations can use powerful foundation models, enterprise copilots, cloud AI services, retrieval systems and increasingly sophisticated AI agents. So why add another layer?

Interestingly, 4Square Innovations does not argue that every component inside AskDiana is technologically unique. Models exist elsewhere. Databases exist elsewhere. Retrieval exists elsewhere. Identity and permission systems exist elsewhere. The claimed distinction lies in how those components are assembled.

Altair Media — What genuinely distinguishes AskDiana from the broader enterprise AI market?

Srikanth Gaurav:
We are not claiming that we invented every component. The differentiation is in the composition. Can an organisation change the underlying model without rebuilding its knowledge architecture? Can it deploy differently because regulatory or operational requirements change? Can permissions, institutional knowledge and applications remain intact when technology underneath them evolves? That is where we believe architecture matters.

This also explains AskDiana’s interpretation of a word increasingly used across European technology debates:

sovereignty.

For Gaurav, sovereignty does not primarily mean that every component must have been created in Europe. It means choice that can actually be exercised.

Can an organisation decide where its data resides? Can it determine which models reason over that data? Can it change providers? And perhaps most importantly: what would changing providers actually cost?

One of AskDiana’s most practical sovereignty tests is therefore remarkably simple: If your AI provider doubled its prices tomorrow, how long would it take you to leave? If the answer is months or years, technological choice may exist on paper while architectural dependency remains intact.

🟦 Does Conversation Intelligence Change the Equation?

AskDiana’s ambitions extend beyond deployment and control. One of the more unusual concepts developed by 4Square Innovations is Conversation Intelligence. Most conversational AI treats interaction as a sequence. A user asks something. The model responds. More information is added. The conversation becomes longer.

AskDiana’s research asks whether that model is sufficient for organisational knowledge. Because organisations do not merely accumulate information. Knowledge changes. Decisions supersede previous decisions. Policies expire. New information alters earlier assumptions. Different sources carry different authority.

The challenge is therefore not simply remembering more context. It is determining what remains valid.

Altair Media — How does Conversation Intelligence differ from simply building a better conversational AI assistant?

Tony Larcombe:
The problem is not just the size of the context window. If a conversation continues for long enough, you eventually have information that has been corrected, superseded or contradicted. Simply retaining everything does not solve that.

Conversation Intelligence explores how context can evolve while preserving what remains relevant and authoritative.

Importantly, 4Square Innovations describes this as an ongoing research programme, rather than presenting the complete architecture as an already proven production capability. That distinction matters. It separates an interesting research direction from a commercial claim that still requires evidence.

🟦 Does Anyone Actually Need This?

Architecture becomes meaningful only when it encounters real organisations. And this is where the AskDiana proposition becomes easier to understand.

Mohamed Marwan, who works on methodology and sector applications, describes implementation through a framework called Assemble – Ask – Apply.

First, determine what knowledge belongs inside the environment. Then establish how people should interact with it. Only afterwards should AI become embedded into actual organisational processes.

One example came from work with an organisation supporting refugees in Lithuania. The central challenge was not whether an AI model could generate useful language. The questions were organisational. Which information is official? Which information can be shown publicly? What must remain internal? Who is allowed to access particular knowledge? Who remains responsible when information changes?

These are not primarily model questions. They are governance questions. And they illustrate why organisations in healthcare, government, NGOs and regulated industries may require something different from simply giving employees access to a general-purpose AI assistant.

AskDiana remains an emerging platform. According to internal figures supplied by 4Square Innovations, its environment included 121 platform instances across 38 businesses and more than 650 users in July 2026.

The company itself cautions against interpreting those numbers as 121 production deployments or paying customers. That restraint is useful. The real test is not how many environments can be created, but whether organisations continue using them because the architecture solves a problem they genuinely have.

🟦 Why Does Europe Matter?

AskDiana was developed in Lithuania, but its European identity is not simply geographical.

Europe’s approach to artificial intelligence increasingly emphasises governance, accountability, data protection and technological sovereignty. Those principles can sometimes be portrayed as constraints on innovation. AskDiana approaches them differently.

The organisation’s proposition is that governance itself may eventually become part of the infrastructure through which AI becomes usable at institutional scale.

Altair Media — What does European digital sovereignty mean in practical terms?

Srikanth Gaurav:
For me, sovereignty can be reduced to one word: choice. Where does your data live? Which models reason over it? Who controls the knowledge? What does it cost to change direction?

If those choices cannot realistically be exercised, sovereignty becomes theoretical. This interpretation also explains why AskDiana does not necessarily position European sovereignty as technological isolation.

An organisation could potentially use models originating outside Europe while retaining control over deployment, knowledge and the ability to change providers.

The strategic question becomes less about the nationality of every component and more about who controls the architecture connecting them.

🟦 Can AskDiana Scale Without Losing Its Philosophy?

That brings the conversation back to the organisation itself.

4Square Innovations does not envision AskDiana as a closed product used in exactly the same way by every customer. Its longer-term ambition involves developers, universities, consultants, technology partners and sector specialists building different applications around a common foundation.

That creates opportunity. It also creates tension. Because the larger an ecosystem becomes, the harder it becomes to preserve common principles.

Interestingly, when Altair Media asked whether there was a question the team wished had been included in this dialogue, Gaurav proposed one himself:

How do you scale an ecosystem without losing the principles that made it worth building?

It may be the most important question in the entire case study.

AskDiana is attempting to solve a problem created by distributed intelligence: how to allow different models, knowledge sources, applications and people to operate while maintaining clear authority and responsibility.

If its ecosystem grows, 4Square Innovations will eventually confront a similar challenge itself. How much freedom should partners have? Which standards remain non-negotiable? Who protects the architectural principles? And when does openness begin to weaken governance?

There is no final answer yet. Perhaps there should not be. Because these are not simply questions about AskDiana. They are questions about what enterprise AI is becoming.

---

■ Signify

After three Signals and an Executive Dialogue, AskDiana becomes easier to place. Its central proposition is not that organisations need another language model, chatbot or AI agent. It is that they increasingly need an architecture through which those technologies can be governed.

That architecture must connect organisational knowledge with permissions, models, reasoning, deployment choices and responsibility—while allowing the technology underneath it to evolve. Whether AskDiana becomes such an architecture at scale remains to be demonstrated.

Conversation Intelligence remains partly a research programme. Market adoption is still developing. The ability to serve organisations ranging from SMEs to governments will ultimately have to be proven in practice. But the problem AskDiana identifies is increasingly difficult to dismiss.

As artificial intelligence moves from experimentation towards organisational infrastructure, the decisive question may no longer be which model produces the most impressive answer. It may be something much more institutional:

Who controls the knowledge, the reasoning, the permissions—and the ability to change course?

That is ultimately the larger story AskDiana tells.

---

This Executive Dialogue concludes Altair Media’s Editorial Case Study on AskDiana. The conversation builds on three preceding Signals examining the platform from different perspectives: what AskDiana is, who it is designed for, and which architectural claims deserve closer scrutiny.

Signal I — What Is AskDiana?
Can Enterprise AI Be Built Differently?

Signal II — Who Is AskDiana Built For?
From SMEs to Governments

Signal III — AskDiana Under the Microscope
Three Architectural Claims. Three Critical Questions.

---

Photo credit

Illustration: Altair Media / OpenAI (concept visualisation)

Caption

Executive Dialogue with Srikanth Gaurav, Tony Larcombe and Mohamed Marwan of 4Square Innovations. Following three Signals examining AskDiana from the outside, Altair Media turns to the people behind the platform to understand its organisation, architecture and ambition—and why AskDiana sees governed knowledge, rather than the AI model itself, as the foundation of enterprise AI.