In Part #1 of our ERP Transformation & AI series, operating model design and vendor selection were explored: Navigating ERP and AI – Part 1: Operating Model and Vendor Selection

Part #2 continues with design, build, and test stages of the ERP programmes, and explore how AI can support delivery fundamental and help programme teams move faster, reduce manual effort, and improve execution success.

Once the ERP vendor and system integrator have been selected, the programme can kick-off and enter the design phase. Unknowns and risks are still high at this stage and resourcing ramps up consuming allocated programme budgets. If not started already, before the programme start this is also time to bring data in foreground of activities as a core workstream. AI is already capable to deliver real impact in multiple domains of these programme stages:

AI in the ERP design phase

The design phase is where business requirements are captured / confirmed and translated into solution design / business processes are designed (on and off system). Among other challenges key business decisions, integration decisions, and data model questions need to be surfaced and solved at this phase.

If utilised by Expert Enterprise and Business Architects AI can support design stage work by

Helping teams identify where applying ERP driven capability uplift and any AI capabilities would bring the most benefits for the given business considering sector / technology and process maturity;

Surfacing dependencies / constraints sooner;

Reducing the time spent on manual discovery and any rework at later programme stages;

Reducing effort needed to create key design artifacts, such as As-Is and planned To-Be processes.

In modern ERP implementation programmes that really matters as vendors like Oracle, SAP, Microsoft, Infor or IFS are increasingly embedding AI capabilities directly into their core platforms, and vendor ecosystems are gradually integrating AI capabilities.

Oracle’s latest recommendation is to follow crawl, walk, run strategy starting when introducing AI capabilities: start with predictive AI features e.g., detecting patterns, then GenAI features e.g., summarising financial information, followed by progressing to AI agents that provide services / automations.

When thinking about designing business processes Microsoft CoPilot AI is able to capture meeting memos but is not yet directly integrated with MS Visio to auto-create process maps, competitors in this sphere like Lucidchart stepping up to document workflows and processes by seamlessly transforming text into visuals with AI. All outputs from AI would need expert human validation and enhancement but can be helpful to create first drafts.

Data preparation and migration enhanced by AI

ERP programmes provide opportunity and raise the necessity to significantly improve the quality of enterprise data rather than simply move bad data into a new system. AI can help profile source data, identify duplicates, validate completeness, and flag anomalies before migration begins, reducing reconciliation effort and lowering cutover risk. This is especially valuable where organisations legacy systems contain multiple duplicate supplier records, inconsistent product structures, or fragmented master data across legacy platforms.

Large ERP vendors like SAP promote own solutions like “RISE with SAP” and also there are more niche AI-enabled data migration supporting solutions like Palantir to help the ETL lifecycle and improve verifications and quality checks throughout the process.

In specialised cases AI features might enable extracting and structuring data from unstructured sources, e.g., AI can work on historical legal contracts stored as pdf documents and extract attributes like parties, start date, end date to enable storage and searchability in a contract management module of the ERP.

In practice, AI capabilities can make it easier to move from fragmented / unstructured source data to a stable ERP data foundation that supports both operations and analytics.

Driving the build

During build, teams are typically configuring out-of-the-box ERP functionality, and, where necessary, building integrations, extending the ERP landscape to connect retained legacy systems / any additional applications to be introduced as part of the programme. AI can help here by accelerating development tasks, improving moving configuration across environments.

As ERP vendors continue to expand their AI enabled assistants and copilots, build teams can increasingly use AI to streamline routine tasks and focus more of their effort on build quality and maximising value delivered.

Save testing time and effort by utilising AI

Testing is one of the areas where AI can create immediate value in an ERP programme as AI-enabled tools can

Generate test scripts based on requirements, design documentations,

Suggest regression test suites / support performance testing, and

Automate part of test execution, reducing the volume of high-cost manual effort required to validate changes.

Reduce ongoing testing efforts for regular patches, integrations with pre-scanning release notes, and “self-healing” test scripts.

Enterprise grade platforms such as Tricentis are good examples re how AI can support faster and more scalable testing across enterprise applications, including SAP, Oracle, Microsoft or IFS landscapes. There are multiple alternative tools in this market segment like the official Oracle partner Opkey, an AI-Powered Oracle Lifecycle Optimization with configuration management, testing and training agent functionalities.

By using AI in testing activities, one should keep in mind that ERP testing serves dual purposes:

Confirming that the solution is fit for purpose, and

Helping users to understand and adopt the new way of working.

Focusing on explainability of the work created by AI and keep utilising testing as part of change journey for colleagues is key. AI-generated test assets can improve traceability from requirements to defects, increase confidence in critical business processes, and give programme leaders earlier visibility into risks that could affect go-live. In practice, that means that with the right level of visibility and approvals technology and business side teams can intervene sooner, before issues become expensive to fix or disruptive to live business activities.

Delivery assurance

AI can also strengthen delivery assurance by improving visibility across programme workstreams. Even when plans are well constructed, delivery risk often emerges from fragmented reporting, slow issue escalation, or a lack of joined-up insight across design, build, data, and test. AI-supported dashboards and analytics can help leaders spot emerging threats earlier and make better decisions while there is still time to act.

As tools mature and operating models adapt to AI-enabled ways of working, the potential impact will only grow. The recent movement toward tighter integration across enterprise assistants, such as SAP Joule and Microsoft Copilot, shows how AI is becoming part of the daily workflow rather than a separate layer of technology. That shift creates an opportunity for ERP programmes to deliver simpler, faster, and more connected ways of working.

What this means for ERP programmes

For programme sponsor and leaders, this raises important questions:

Which AI enabled capabilities should be activated and when as part of the ERP roadmap?

Which core ERP capabilities can be utilised and which envisaged capabilities should rely on build or buy specialised tool(s)?

Do I have the technology and even more importantly the human capabilities within my organisation / programme team to utilise state of the art AI tools & features to enhance team output / performance?

In Searchlight’s view the answers largely depend on the value drivers for the business, the maturity of the process / capabilities, and the sector-specific benefits AI can realistically deliver. The right approach is to use AI where it adds measurable value.

AI can accelerate the design, build and test phases of the ERP programmes by fast tracking analysis, reducing manual efforts, and improving decision-making, but it does not remove the need for strong Programme delivery governance and leadership, Enterprise, Business and Data architecture expertise.

Also, one should keep in mind that using AI will have associated cost, e.g., usage priced separately, and colleagues need to be upskilled to use AI effectively and validate outputs at the right time at the right level of detail.

Searchlight’s experts are best placed to help organisations prioritise the right use cases and define how AI-enabled outcomes should be reached in a practical, value-led way.

Get in touch to explore how Searchlight can help you achieve your desired outcomes.

In the next part of this series, we will explore how AI can support service transition, benefits realisation and end state operating model, helping organisations move from go-live readiness to sustainable business value. Follow us on LinkedIn to stay updated.