Insight
Insight
The AI model isn’t the story anymore
14 Jun 2026
Matt Cull
HumanX revealed where the enterprise AI conversation is heading next. The challenge is shifting from intelligence to orchestration.
Walking around HumanX Europe, one thing became apparent surprisingly quickly. AI is starting to look less like a distinct category of technology and more like the way technology itself is going to work. The event brought together thousands of enterprise leaders, technology companies, investors and AI-native startups, but what struck me most wasn’t simply the scale of the ecosystem. It was how much the conversation had moved on.
For much of the generative AI boom, attention has concentrated on the models themselves. Which is smartest? Which can reason more effectively? Which can code, analyse or generate content better than the last? Those developments still matter, but at HumanX the more interesting conversations were increasingly happening one level above them. The question is becoming less about what an individual model can do and more about how AI connects to the systems, processes and data that actually make a business function.
For me, that was the biggest takeaway from the event. The intelligence is becoming increasingly accessible. Orchestrating it inside a real organisation is becoming the harder problem.
From models to orchestration
One of the clearest themes at HumanX was the movement from standalone models towards agents capable of reasoning, planning and taking action. But businesses do not operate inside a single application. An ordinary process might cross finance, CRM, HR, project management, procurement and several specialist systems before it is complete. Information moves between them, decisions happen at different stages and, in many organisations, people still provide the connective tissue holding the process together.
Giving one part of that chain access to an intelligent model may make an individual task faster, but it does not necessarily make the overall business process work any better.
As agents become more capable, the problem therefore changes. Organisations need to decide what information they can access, which actions they are allowed to take, how they interact with existing applications, how multiple agents coordinate and where human oversight is required. The conversation moves from simply deploying intelligence to orchestrating it across the enterprise.
That is why seeing so many orchestration tools and platforms at HumanX was particularly interesting for us at Vannda. It reflects a problem we have been thinking about for some time with V360: if AI is going to move beyond isolated productivity gains to genuine operational transformation, businesses need a way of coordinating intelligence across the technology estate.
The market has arrived at the problem
Most organisations already have the systems they need to run the business, but those systems rarely operate as one coherent environment. ERP, CRM, HR and specialist applications all perform important functions, yet employees still spend significant amounts of time moving information between them, reconciling outputs and navigating different interfaces to complete a single process.
The opportunity is not necessarily to replace those systems. It is to create an orchestration layer above them that can understand what needs to happen, coordinate the relevant applications and increasingly allow intelligent agents to participate in the process.
That is the thinking behind V360.
Seeing orchestration appear so prominently at HumanX was reassuring, but not because it suggested Vannda had found a category with no competition. In an emerging market, the opposite can be more valuable. When multiple businesses independently begin focusing on the same challenge, it suggests the market itself is beginning to recognise the problem.
The market has arrived at the problem.
The interesting question now is how businesses solve it in a way that works beyond a demonstration or isolated use case.
Enterprise AI has to survive contact with the enterprise
This was another tension running through HumanX. Building with AI has become extraordinarily easy, and the startup presence demonstrated just how quickly new products, agents and services can now be created by relatively small teams.
Deploying AI successfully inside an established enterprise is an entirely different challenge.
A prototype needs to demonstrate that something is possible. An enterprise system needs to work reliably across real data, existing infrastructure, security requirements, permissions, governance frameworks and countless exceptions accumulated over years of doing business.
A model might perform a task perfectly in isolation, but that task may depend on information held across several applications. Data may be inconsistent. Different markets or clients may operate differently. Approval stages may exist for important regulatory or commercial reasons. Some of the knowledge required to recognise an unusual case may not be documented anywhere, but held by experienced people inside the organisation.
This is why enterprise AI increasingly becomes an architecture and operating-model question rather than simply a technology question. The model can be extraordinarily intelligent, but it still has to operate inside an organisation that is complicated.
The interface may disappear before the software does
There is a temptation to look at increasingly capable agents and assume traditional enterprise applications are about to become obsolete. I think something subtler is happening.
ERP systems, CRMs, HR platforms and specialist applications still contain the records, controls and processes businesses depend upon. What may change much more quickly is the need for people to interact with every application directly.
Today, employees often need to know where information lives, which application to open and what needs to happen next. An intelligent orchestration layer changes that relationship. A user could express the outcome they need while the layer above the technology estate identifies the relevant information, interacts with the appropriate systems and coordinates the actions required.
The applications remain underneath, but they become increasingly invisible to the person trying to get something done. This is closely related to the Headless Enterprise idea we have discussed previously at Vannda: the value of enterprise software remains while the traditional interface through which people experience it starts to disappear.
The breakthrough isn’t simply putting an AI assistant inside every application. It is allowing intelligence to operate across them.
Will AI eventually kill the AI conference?
My other major takeaway from HumanX was the amount of talent and startup activity on show, particularly across Europe. But it also left me with one slightly provocative thought: AI may eventually kill the AI conference.
Not because interest in artificial intelligence will disappear, but because separating AI from the rest of enterprise technology may eventually stop making sense. Conversations about AI are already becoming conversations about infrastructure, cybersecurity, data, enterprise applications, software development, operations and organisational design.
We don’t generally talk about businesses having an “internet strategy” anymore because the internet became embedded in almost every part of how modern organisations operate. AI appears to be travelling along a similar path.
Eventually, an “AI conference” may sound as strange as an “internet business conference” does today. AI will simply be part of the technology conversation.
Building for what comes next
For Vannda, HumanX reinforced the direction we have been moving in with AI-Powered Business Systems and V360. The opportunity is not simply to add AI features to existing applications. It is to rethink how the technology estate works when intelligence can operate across it.
The models will change. Agents will improve. New providers will emerge and capabilities that look extraordinary today will become commonplace. Businesses cannot redesign their operating model every time that happens.
They need strong systems underneath, trusted data and clear governance, combined with an architecture capable of coordinating applications, agents, information and actions as technology continues to evolve.
HumanX showed just how quickly that market is developing. For me, the biggest takeaway was not that AI is becoming more powerful. We already knew that.
It was that the next phase of enterprise AI is becoming less about what an individual AI can do, and more about how everything works together.
HumanX revealed where the enterprise AI conversation is heading next. The challenge is shifting from intelligence to orchestration.
Walking around HumanX Europe, one thing became apparent surprisingly quickly. AI is starting to look less like a distinct category of technology and more like the way technology itself is going to work. The event brought together thousands of enterprise leaders, technology companies, investors and AI-native startups, but what struck me most wasn’t simply the scale of the ecosystem. It was how much the conversation had moved on.
For much of the generative AI boom, attention has concentrated on the models themselves. Which is smartest? Which can reason more effectively? Which can code, analyse or generate content better than the last? Those developments still matter, but at HumanX the more interesting conversations were increasingly happening one level above them. The question is becoming less about what an individual model can do and more about how AI connects to the systems, processes and data that actually make a business function.
For me, that was the biggest takeaway from the event. The intelligence is becoming increasingly accessible. Orchestrating it inside a real organisation is becoming the harder problem.
From models to orchestration
One of the clearest themes at HumanX was the movement from standalone models towards agents capable of reasoning, planning and taking action. But businesses do not operate inside a single application. An ordinary process might cross finance, CRM, HR, project management, procurement and several specialist systems before it is complete. Information moves between them, decisions happen at different stages and, in many organisations, people still provide the connective tissue holding the process together.
Giving one part of that chain access to an intelligent model may make an individual task faster, but it does not necessarily make the overall business process work any better.
As agents become more capable, the problem therefore changes. Organisations need to decide what information they can access, which actions they are allowed to take, how they interact with existing applications, how multiple agents coordinate and where human oversight is required. The conversation moves from simply deploying intelligence to orchestrating it across the enterprise.
That is why seeing so many orchestration tools and platforms at HumanX was particularly interesting for us at Vannda. It reflects a problem we have been thinking about for some time with V360: if AI is going to move beyond isolated productivity gains to genuine operational transformation, businesses need a way of coordinating intelligence across the technology estate.
The market has arrived at the problem
Most organisations already have the systems they need to run the business, but those systems rarely operate as one coherent environment. ERP, CRM, HR and specialist applications all perform important functions, yet employees still spend significant amounts of time moving information between them, reconciling outputs and navigating different interfaces to complete a single process.
The opportunity is not necessarily to replace those systems. It is to create an orchestration layer above them that can understand what needs to happen, coordinate the relevant applications and increasingly allow intelligent agents to participate in the process.
That is the thinking behind V360.
Seeing orchestration appear so prominently at HumanX was reassuring, but not because it suggested Vannda had found a category with no competition. In an emerging market, the opposite can be more valuable. When multiple businesses independently begin focusing on the same challenge, it suggests the market itself is beginning to recognise the problem.
The market has arrived at the problem.
The interesting question now is how businesses solve it in a way that works beyond a demonstration or isolated use case.
Enterprise AI has to survive contact with the enterprise
This was another tension running through HumanX. Building with AI has become extraordinarily easy, and the startup presence demonstrated just how quickly new products, agents and services can now be created by relatively small teams.
Deploying AI successfully inside an established enterprise is an entirely different challenge.
A prototype needs to demonstrate that something is possible. An enterprise system needs to work reliably across real data, existing infrastructure, security requirements, permissions, governance frameworks and countless exceptions accumulated over years of doing business.
A model might perform a task perfectly in isolation, but that task may depend on information held across several applications. Data may be inconsistent. Different markets or clients may operate differently. Approval stages may exist for important regulatory or commercial reasons. Some of the knowledge required to recognise an unusual case may not be documented anywhere, but held by experienced people inside the organisation.
This is why enterprise AI increasingly becomes an architecture and operating-model question rather than simply a technology question. The model can be extraordinarily intelligent, but it still has to operate inside an organisation that is complicated.
The interface may disappear before the software does
There is a temptation to look at increasingly capable agents and assume traditional enterprise applications are about to become obsolete. I think something subtler is happening.
ERP systems, CRMs, HR platforms and specialist applications still contain the records, controls and processes businesses depend upon. What may change much more quickly is the need for people to interact with every application directly.
Today, employees often need to know where information lives, which application to open and what needs to happen next. An intelligent orchestration layer changes that relationship. A user could express the outcome they need while the layer above the technology estate identifies the relevant information, interacts with the appropriate systems and coordinates the actions required.
The applications remain underneath, but they become increasingly invisible to the person trying to get something done. This is closely related to the Headless Enterprise idea we have discussed previously at Vannda: the value of enterprise software remains while the traditional interface through which people experience it starts to disappear.
The breakthrough isn’t simply putting an AI assistant inside every application. It is allowing intelligence to operate across them.
Will AI eventually kill the AI conference?
My other major takeaway from HumanX was the amount of talent and startup activity on show, particularly across Europe. But it also left me with one slightly provocative thought: AI may eventually kill the AI conference.
Not because interest in artificial intelligence will disappear, but because separating AI from the rest of enterprise technology may eventually stop making sense. Conversations about AI are already becoming conversations about infrastructure, cybersecurity, data, enterprise applications, software development, operations and organisational design.
We don’t generally talk about businesses having an “internet strategy” anymore because the internet became embedded in almost every part of how modern organisations operate. AI appears to be travelling along a similar path.
Eventually, an “AI conference” may sound as strange as an “internet business conference” does today. AI will simply be part of the technology conversation.
Building for what comes next
For Vannda, HumanX reinforced the direction we have been moving in with AI-Powered Business Systems and V360. The opportunity is not simply to add AI features to existing applications. It is to rethink how the technology estate works when intelligence can operate across it.
The models will change. Agents will improve. New providers will emerge and capabilities that look extraordinary today will become commonplace. Businesses cannot redesign their operating model every time that happens.
They need strong systems underneath, trusted data and clear governance, combined with an architecture capable of coordinating applications, agents, information and actions as technology continues to evolve.
HumanX showed just how quickly that market is developing. For me, the biggest takeaway was not that AI is becoming more powerful. We already knew that.
It was that the next phase of enterprise AI is becoming less about what an individual AI can do, and more about how everything works together.