Automating support: why strong foundations matter more than tools
Published March 23, 2026
- CIO & CTO Advisory
Key takeaways
- Support automation doesn’t hinge first and foremost on tool selection, but on the strength of the underlying foundations.
- ITSM processes, data quality, the knowledge base and governance all directly shape the value AI can deliver.
- Use cases need to be prioritized based on the maturity of the support function, business needs and the target level of industrialisation.
- Scaling successfully depends just as much on governance and user adoption as it does on the technology itself.
Artificial intelligence is now emerging as a major driver of change for support functions. Use cases are multiplying, expectations are high, and the first results are already visible.
In a recent article on the 10 AI use cases in support, we already showed how these technologies can automate the handling of simple tickets, speed up triage and help agents make sense of complex incidents faster.
Yet on the ground, we keep seeing the same pattern: support automation initiatives are still too often approached through the lens of the tool itself. Teams focus on which assistant to deploy, how advanced the AI should be or which provider to choose. In most cases, the real challenge lies elsewhere. A more automated support model first requires solid foundations: clear processes, usable data, a reliable knowledge base, a well-defined governance framework and well-supported usage. Only then can AI start delivering real value.
Support automation: the key steps in the transformation journey
The essential building blocks for transforming support
Assess the current setup
Assessing the organization’s current situation is essential in order to gauge its level of maturity. This assessment needs to cover both the business context and user needs, which are often diverse and uneven, as well as the technological maturity of the existing support model. Without this broader view, transformation efforts risk missing real user needs or failing to gain traction.
Define a common ITSM framework
In practice, this framework can take the form of structured documentation for support processes, a clear definition of roles and responsibilities, for example through a RACI matrix, shared ticket management standards and governance rules that guide the evolution of the service catalog, the knowledge base and support tools.
This is what ensures consistency across the support model at organisational level. Once the framework is in place, it can be embedded into the ITSM tool, which then becomes the operational expression of the target model.
ITSM
IT Service Management: the set of processes, tools and methods used to manage IT services and user support.
The foundations that really make the difference
A robust technology backbone needs to be in place before more advanced levers such as automation, self-service or artificial intelligence can be rolled out effectively. That means having structured ITSM processes within a platform, including incident, request and problem management. The service catalog, in particular, needs to be clear, well structured and free from duplication, with categories and items that users can understand at a glance. Once integrated into the platform, it directly shapes users’ ability to submit requests and benefit from smooth support journeys.
A reliable CMDB is another key part of that backbone, providing the visibility needed over assets and their dependencies. Without a well-structured service catalog and clear items, users cannot submit requests effectively. On the support side, a lack of consistent categories makes ticket qualification, routing and management far more difficult.
AI applied to support can only work with what it is able to read, use and learn from. Inconsistent categorization, poorly used statuses, priorities interpreted differently from one team to another or a weak resolution history all undermine the relevance of analysis and automation straight away.
We want to stress this point in particular: even when tickets are correctly categorized, they remain unusable if they are documented only superficially, with little more than a status update. Without that level of rigour, the data may exist, but it remains difficult to use, and AI technologies become hard to activate effectively, if not ineffective altogether.
The knowledge base plays a central role in the future of support and needs to be continuously enriched, particularly through resolution tickets stored in the ITSM platform. Its content therefore needs to reflect the real needs of both support teams and users, and the processes that govern it need to be clearly framed. Outdated or conflicting articles quickly erode trust in self-service journeys and chatbots.
Technology choices that matter
Once the foundations of the support model are in place, the question is no longer whether to introduce AI, but which decisions will shape its success: which use cases to prioritize, how far to scale the solutions and which sourcing model to adopt. These choices involve strategic trade-offs across technology, cost and sovereignty.
The first decision is about which use cases to prioritize. Not all of them offer the same value or require the same level of complexity. Some sit within analytical AI, others within generative AI, and others again within more agentic approaches. In practice, these building blocks are likely to coexist. The key is to shape a phased roadmap with a clear order of priority.
Once the AI direction has been chosen, implementation comes next. Organizations then face several technology options depending on their needs, maturity and the level of control they want to retain.
These choices generally fall into two main approaches: BUY and MAKE.
Scaling starts well before deployment
Prepare the rollout
A credible trajectory requires a structured transition plan. New building blocks need to be integrated, knowledge transfer organized, responsibilities clearly defined, user journeys tested, configurations adjusted and the ramp-up phase secured. Transformation unfolds over a longer timeframe. It involves building the technology foundations, running test phases and implementing the core tools.
Manage the transformation over time
Governance needs to track both the progress of the transformation, the quality of the service delivered and user adoption. Ongoing use of data and analysis helps identify opportunities to improve the service proactively. Robust KPIs make it possible to steer and assess the success of the support transformation. To measure the Shift-Left effect enabled by AI, we can, for example, track the share of tickets handled through self-service or resolved at first contact. To assess request handling more broadly, useful indicators include the automation rate and the average resolution time for tickets handled by AI.
We recommend tracking both operational performance indicators and experience-focused indicators, for example through XLAs or post-interaction scores. This dual perspective matters. A solution can look efficient on paper and still disappoint in day-to-day use.
Secure adoption
One of the most underestimated aspects of this kind of project is user adoption. A change management plan therefore needs to include tailored training, communication and support to secure the transition and foster lasting adoption.
XLAs
Experience Level Agreement: an experience-focused indicator used to measure how users perceive a service, beyond operational KPIs alone.
Shift-Left
An approach that aims to handle more requests earlier in the support journey, ideally at first line or through automated paths.
XLAs
Experience Level Agreement: an experience-focused indicator used to measure how users perceive a service, beyond operational KPIs alone.
Shift-Left
An approach that aims to handle more requests earlier in the support journey, ideally at first line or through automated paths.
What results can be expected?
Support transformation can generate significant gains on both the operational and financial fronts.
By supporting a shift-left approach, AI makes it possible to handle a growing share of requests through self-service, chatbots and automation. This reduces escalation to higher support tiers and helps lower the associated costs. Managed service providers offering AI-enabled support services generally expect the share of tickets resolved through self-service to rise significantly, in some cases leading to reductions of up to 50% in level 1 ticket volumes over five years, alongside a comparable drop in remote support activity and in the number of tickets handled by on-site technicians.
Handling these requests earlier in the journey relieves Service Desk teams of the most repetitive tasks and allows them to focus on more complex and strategic cases. AI also improves the orchestration of support activities through more consistent categorization, more accurate prioritization and shorter handling times.
Some gains are immediately visible in day-to-day operations. In our white paper published with ServiceNow, we show, for example, how a complex incident that previously required more than a minute of manual reading can be understood in under 30 seconds thanks to an automated summary of the incident history. The benefit is twofold: the agent saves time, and the user receives a faster response.
Several of our clients have already seen tangible results. One of them recorded a 40% reduction in Helpdesk requests after a chatbot began handling part of the incoming demand.
Wavestone supports clients at every stage of this transformation, from preparation and operational deployment to performance steering, helping turn ambition into lasting results.
Keeping a clear-eyed view of the limitations
It would obviously be counterproductive to present support automation as a smooth or linear journey. The limitations remain very real.
Tools can hallucinate, misread a request, lack the right context or return inappropriate responses. The quality of the output depends heavily on the data available, the reliability of the knowledge base, the way agents are configured and the surrounding governance framework. A poor start can quickly damage credibility. Sometimes, just a few inaccurate answers are enough to undermine trust in a chatbot or a self-service journey.
There are still barriers among users, many of whom are not yet fully familiar with these changes. Without a sound understanding of prompting best practices and of what AI can and cannot do, the use of AI-enabled support technologies will never reach its full potential. Some employees, for instance, use an AI agent as if it were a search engine, even though it was never designed for that. Usage is also shaped by major cultural differences across organizations, technology environments, business units and regions. That is why change management needs to reflect this diversity of user contexts if the transformation is to succeed at scale.
These challenges highlight two things at once: the need for strong organisational and technological foundations, and the importance of building a robust roadmap that covers the full transformation cycle, from preparation and operational deployment to performance steering and continuous improvement.
