In my last post, I ended with a thought that I believe captures something important about AI: capability you can buy, conviction you have to earn.
This time, I want to take a closer look at Experion ARC AI Studio, how it works, what goes into each of its three pillars, and how they come together through the course of an AI programme.
First, What is Experion ARC AI Studio?
Experion ARC AI Studio is the framework through which Experion Technologies approaches AI engineering and enterprise transformation. ARC stands for Analyse, Realise and Catalyse.
The idea is straightforward. Identify where AI can create meaningful business value, engineer the right solution, and then measure how well it delivers against the intended outcomes.
While many organisations have made progress in one or two of these areas, connecting all three remains a challenge. A promising use case doesn’t always translate into a successful implementation, and a successful implementation doesn’t necessarily deliver measurable business value.
That’s a gap worth addressing. McKinsey’s State of AI research, for instance, indicates that only 39% of organisations report any enterprise-level EBIT impact from AI. Considering the investments being made, that’s a significant gap between adoption and actual business results.
Three Pillars Addressing Three Common Challenges
One way to understand the thinking behind ARC is to look at where AI initiatives tend to run into problems.
The first challenge is choosing the right use case. It’s easy to get excited about a new technology or an impressive demonstration and start looking for ways to implement it. But the technology being interesting doesn’t automatically make it a business priority. This is the selection challenge.
The second is execution. Even when the opportunity is valid, the solution needs to be engineered around the actual business process, with the right architecture, integrations, security and governance. Otherwise, the result could be technically sound but difficult to use or scale.
The third is measurement. Without an agreed starting point, it becomes difficult to establish whether the initiative has delivered the expected value. This is particularly important when organisations are trying to justify further investments.
These are not necessarily technology problems. More often, they come down to decisions made before implementation, the way solutions are engineered, and how outcomes are tracked.
This is where the three ARC pillars come in:
- Analyse (iThink): Identifying and validating the right opportunities.
- Realise (iBuild): Engineering those opportunities into working enterprise solutions.
- Catalyse (iScale): Scaling adoption, managing operations and measuring business outcomes.
An important aspect of this approach is continuity. The business case and baseline established during Analyse carry forward into Realise and Catalyse. This ensures that the teams building the solution and those measuring its performance are working towards the same objectives.
It sounds straightforward, but maintaining that connection throughout an AI programme can make a considerable difference.
Analyse (iThink): Getting the Opportunity Right
Analyse is where an AI initiative should begin, although it’s often the stage organisations are most tempted to rush through.
There is usually considerable pressure to demonstrate progress, particularly when AI is already on the leadership agenda. But spending time understanding the opportunity and the underlying business process can prevent much larger problems later.
Analyse covers AI opportunity discovery and prioritisation, vision and strategy, UX and experience design, readiness assessments and validated prototypes.
At the centre of this pillar is Rethink, Experion’s module for AI opportunity discovery and ROI assessment.
Rethink starts by understanding how a business process works today. Users can describe the process in plain language and upload supporting documents, including information about the product, stakeholders and intended outcomes. The platform then maps the process, breaking it down into individual steps and sub-steps and identifying which activities are manual, automated or AI-enabled.
The process map can be reviewed and refined through the canvas or conversational interface.
One aspect I particularly like about this approach is that Rethink doesn’t immediately jump to recommending AI. It first looks at whether the existing process can be improved. This could mean removing duplicate activities, changing the sequence of certain steps or addressing inefficient handoffs between teams.
Only after these improvements are considered does it identify where AI could add value.
Each recommended intervention is scored and supported by evidence drawn from Experion’s project experience, research from organisations such as Anthropic, OpenAI, Google DeepMind and Microsoft, and industry research from institutions including MIT Sloan, Harvard Business Review, Stanford HAI and McKinsey.
The next step is establishing the business case.
Rethink collects relevant inputs for each process step, such as volumes, time spent and associated costs. These inputs are reviewed with the process owners to establish an agreed baseline. Based on this, the platform provides conservative, base and optimistic ROI projections, with the underlying assumptions and calculations available for review.
Organisations can then compare opportunities, prioritise them into now, next and later, and build an enterprise-level view of potential value and payback.
The outcome is a prioritised roadmap supported by evidence, along with working prototypes to help validate the opportunities before committing to larger engineering investments.
There is also an interesting learning from Experion’s engagement with a global luxury active-travel company. The opportunities that emerged as the strongest candidates for AI had four things in common: high volumes, significant customer or employee experience involvement, multiple operational constraints and measurable outcomes.
It’s a useful way to evaluate an AI opportunity, particularly when an organisation has a long list of ideas but needs to decide where to begin.
Realise (iBuild): Engineering AI Into Products and Processes
Once the opportunities have been validated, Realise takes them into engineering and implementation.
This pillar covers legacy modernisation, new product engineering, AI feature development, generative and agentic AI, and platform integration.
There are two complementary aspects to the way Experion approaches engineering here.
The first is building intelligent products, where AI becomes part of the product or solution being developed. This could be an individual AI-powered feature or a more complex agentic system, either within a new product or an existing enterprise platform.
The second is using AI to improve the engineering process itself. Rather than treating AI as an occasional productivity tool for developers, the objective is to integrate it into the product engineering lifecycle, from requirements and design through development, testing and deployment.
Experion has been applying this approach internally as well, using its own engineering practices and experience to refine the model before taking it to customers.
The primary module in this pillar is Forge, which focuses on AI-native product management and engineering.
Forge has nine specialist agents working across different stages of the product lifecycle. These include agents for product requirements, user stories, solution architecture, high-level design, UI/UX, backend and frontend development, and infrastructure-as-code and DevOps.
A programme management agent works across these activities, handling deliverables such as project charters, work breakdown structures, RACI matrices and status reports.
Of course, bringing multiple agents into the software development lifecycle also makes governance and human oversight particularly important.
Forge incorporates human approval at key stages. Product managers review and approve plans before user stories are generated. Design and product teams sign off before development begins, and low-level designs are reviewed before coding. Code reviews and final approvals are also built into the workflow before changes are merged or deployed.
These checkpoints are supported by automated quality checks, including a 30-point PRD assessment, a code review agent and an independent reviewer agent.
The intention is to bring AI into existing engineering workflows rather than create another environment that teams need to learn. Outputs can be delivered through familiar platforms and formats, including Confluence pages, GitHub pull requests and pipeline YAML files.
Realise also includes a module for intelligent bug resolution and Lens, which focuses on codebase intelligence.
Lens analyses existing codebases, including configurations, schemas and documentation, and creates a knowledge graph covering services, dependencies, data flows and architectural decisions. This knowledge base is updated as the code changes.
For organisations working with large legacy platforms, this can be particularly useful. Understanding an unfamiliar codebase often depends on a handful of people who have worked on it for years. When that knowledge isn’t documented, onboarding new developers, managing transitions and planning modernisation become considerably more difficult.
Lens helps make that knowledge accessible and easier to maintain, reducing dependency on individuals and giving engineering teams a better understanding of the systems they work with.
Catalyse (iScale): Taking AI beyond Implementation
Getting an AI solution into production is an important milestone, but it isn’t the end of the journey.
Organisations still need to ensure that people are using it, that it continues to deliver value, that operating costs remain under control and that the solution can evolve as business requirements change.
This is the role of Catalyse.
The pillar covers business transformation, managed services and scale-out, ROI tracking and measurement, AI monitoring and observability, and operating model design.
Its primary module, Tower, provides programme control, cost governance and observability across AI initiatives.
Tower gives organisations a consolidated view of their AI use cases, organised by business domain and the value they deliver. It tracks initiatives from the initial idea through to production, while monitoring adoption, delivery performance and associated costs.
Cost management deserves particular attention here. AI solutions, especially those involving agents and large language models, can have operating costs that change significantly with usage and scale.
Gartner has highlighted escalating costs among the reasons why a substantial proportion of agentic AI projects may be abandoned or discontinued by 2027. This makes it important to understand the economics of an AI solution from the beginning rather than treating cost as something to review only after deployment.
Tower helps track platform and operational expenditure at the use-case level, giving organisations better visibility into the financial implications of scaling their AI investments.
Catalyse also addresses the operational questions that come with putting AI into everyday business use.
Who owns a use case after it goes live? Who is responsible for monitoring and tuning the model? How should unexpected agent behaviour be handled? What governance mechanisms need to be in place as adoption increases?
These considerations are essential to moving from individual AI implementations to a sustainable enterprise-wide operating model.
The other important component is outcome measurement.
The value delivered by each initiative is measured against the baseline established during Analyse. This provides continuity between the original business case and the actual results, allowing organisations to assess performance using the same parameters that informed the initial investment decision.
Bringing the Three Pillars Together: A Real-world Example
The three pillars become easier to understand when viewed through an actual engagement.
In my previous post, I touched upon Experion’s work with a global luxury active-travel company, one of the largest in its category. The business manages more than 8,000 trips annually across over 60 countries, serving upwards of 80,000 customers every year.
When the organisation began exploring AI opportunities, some of the initial ideas were fairly familiar: booking chatbots, personalised recommendations, itinerary generation and marketing automation.
These were all reasonable possibilities. However, the Analyse phase took a closer look at the underlying operations before deciding where to invest.
Using Rethink, the team mapped the existing processes and evaluated the opportunities based on business impact, operational complexity and measurable value.
The opportunity that eventually emerged as the strongest candidate was quite different from the initial shortlist.
It involved assigning trip leaders and field staff to individual trips. This was a largely manual process, with scheduling teams having to consider several variables, including availability, visa requirements, language capabilities, gender-related requirements and individual workloads across hundreds of staff members and thousands of trips.
The process was complex, time-consuming and had a direct impact on revenue when planning was delayed.
It also had the characteristics identified earlier: high volume, significant experience involvement, multiple constraints and measurable outcomes.
The Realise phase took the opportunity forward through Forge. Interestingly, the solution wasn’t built around AI alone. It combined AI with operations research and optimisation techniques, which were better suited to addressing the actual scheduling problem.
The initiative began as a proof of concept and has since progressed into a full production project.
Catalyse then provides the framework for managing adoption, scaling the solution and tracking its business impact against the baseline established with the process owners during Analyse.
For me, this example highlights an important aspect of AI opportunity discovery. The best use case isn’t necessarily the most obvious one or the one that attracts the most attention. It is the one that addresses a genuine business problem and delivers measurable value.
Accelerators and Vertical AI Solutions
In addition to the three core pillars, Experion ARC AI Studio includes ready-made accelerators and industry-specific AI solutions.
The accelerators are designed to address common enterprise requirements without having to start every engagement from scratch.
These include:
- Ask: Converts data into answers through natural language interactions.
- Read: Provides document intelligence capabilities.
- Voice: An outbound voice agent.
- Desk: A policy and knowledge assistant.
- Pilot: Supports the journey from autonomous build through to release.
Experion also has vertical AI solutions for Healthcare and Financial Services, bringing industry-specific capabilities and context into the engagement.
Together, these accelerators and vertical solutions complement the three pillars, helping organisations move faster while retaining the same underlying approach to discovery, engineering and value measurement.
Where Should an Organisation Start?
While the three pillars are designed to work together, the starting point will depend on an organisation’s current maturity, priorities and challenges.
There are generally four common scenarios.
Building a new product or solution
For organisations starting with a new idea, Analyse is the natural entry point. Opportunity discovery, ROI validation, prototyping and business case development help establish whether the investment makes sense before moving into engineering.
Modernising a legacy platform
This typically begins with Analyse and moves into Realise. The first step is understanding the existing processes and technology landscape, establishing a baseline and identifying what needs to be retained, improved, retired or rebuilt. Lens can play an important role in understanding the existing codebase and its dependencies.
Taking a stalled AI pilot forward
Here, Realise may be the immediate priority. The initial use case might be valid, but the solution could lack the engineering maturity required for enterprise deployment. Security, governance, integration and observability are some of the areas that need attention before it can become a dependable production system.
Scaling existing AI implementations
For organisations that already have working AI solutions, Catalyse provides the foundation for scaling adoption, monitoring performance, managing costs and establishing the operating model required for sustained value.
Regardless of the starting point, establishing a clear and measurable baseline is important. It gives everyone involved a common understanding of what success should look like.
What’s next in this series?
This post provides an overview of Experion ARC AI Studio and how its three pillars work together. In the coming posts, I’ll take a closer look at each of them.
I’ll begin with Analyse and Rethink, exploring how the platform approaches process improvement, AI opportunity discovery and ROI assessment.
The next post will focus on Realise, including Forge’s nine specialist agents, the human approval checkpoints and Lens’s approach to codebase intelligence.
Finally, I’ll cover Catalyse and Tower, looking at how organisations can manage, govern and measure AI initiatives after they go live.
If there is a particular aspect you’d like me to explore in more detail, feel free to share it in the comments.
A Bit About Experion
Experion Technologies is a product engineering and digital services company with over 1,800 engineers and 500+ customers across more than 35 countries. The company has offices in the US, UK, Australia, Japan and India and has recorded a five-year CAGR of over 30%.
Experion ARC AI Studio has been in use for more than five years, with over 10 customers actively using the platform and a dedicated Data & AI team of around 100 specialists supporting its development and delivery.
One Last Thought
Looking at the three pillars together, I believe the real strength of ARC lies in maintaining continuity across the entire AI journey.
The opportunity identified during Analyse needs to remain connected to what gets engineered during Realise. And what gets delivered needs to be measured against the business outcomes originally agreed upon.
When these connections are lost, organisations can end up with technically impressive solutions that struggle to demonstrate their business value.
AI investments need more than good technology and capable engineering teams. They need the right opportunities, a disciplined approach to implementation and a clear understanding of the outcomes.
That’s the thinking behind Experion ARC AI Studio.
Next post: Analyse.
If you’d like to explore how this approach could apply to your business processes, reach us at info@experionglobal.com or visit experionglobal.com.
Questions I Tend to Get Asked
What are the three pillars of Experion ARC AI Studio?
Analyse (iThink), Realise (iBuild) and Catalyse (iScale). Analyse identifies and validates AI opportunities, Realise engineers them into production solutions, and Catalyse focuses on scaling, governance and measuring business outcomes.
What modules are included in each pillar?
Analyse includes Rethink for AI opportunity discovery and ROI assessment. Realise includes Forge for AI-native product management and engineering, Lens for codebase intelligence, and a module for intelligent bug resolution. Catalyse includes Tower for programme control, cost governance and observability.
How is ROI measured?
A baseline is established with process owners during Analyse. The platform develops conservative, base and optimistic projections using the agreed inputs. Once the solution is implemented, Catalyse tracks actual value against that same baseline.
Can an organisation start with just one pillar?
Yes. The starting point depends on the requirement. New product initiatives generally begin with Analyse, legacy modernisation moves from Analyse into Realise, stalled pilots may begin with Realise, and existing AI implementations that need to scale can start with Catalyse.
Is there human oversight when AI agents are involved in software engineering?
Yes. Forge incorporates human approval at key stages, including planning, design, low-level design, code review and final deployment approvals. Automated quality checks provide additional validation throughout the process.
What accelerators are available?
The current accelerators include Ask, Read, Voice, Desk and Pilot. Experion also offers vertical AI solutions for Healthcare and Financial Services.
Back
