Delivering GenAI at scale across an organisation: FAQs.
At the Cloud Revolution Summit, 3 experts brought unique perspectives on rolling out GenAI successfully across an organisation.
Mauro Meeuws from NIBC Bank, Allan Chong from Nordcloud and Seda Akdemir from Microsoft brought their unique perspectives to frequently asked GenAI questions.
What is the primary challenge your organisation faces in scaling GenAI across the enterprise?
Cloud Revolution Summit delegates were clear – scaling GenAI isn’t an infra or data challenge, it’s a people, process and use case challenge.
Let’s look at how Mauro, Allan and Seda recommend overcoming them.
How can we address the difficulty in finding skilled professionals to build, implement and run GenAI?
Enterprise perspective.
In the cloud journey, you don’t become cloud native overnight. The same applies to GenAI. There’s no expert with years of experience – you’re in a transformation and need to build within the organisation.
- Choose a partner with experience of upskilling along the GenAI journey
- Look for interested people internally and make them part of the journey
- Organise engagement sessions to get people excited about the possibilities
- Coach and upskill your chosen GenAI champions
Mauro Meeuws, Product Owner – Hybrid Cloud, NIBC Bank
Partner perspective.
- Give people across the organisation GenAI fundamentals training – it builds awareness of the available capabilities and what they can be used for
- Take a multidisciplinary approach – go beyond dev/infra/data and engage people like product owners, legal and security
- For build – rely on partners and upskill internally
- When it comes to management – having a partner run and operate frees up the business to focus on new use cases
Allan Chong, Head of AI, Nordcloud
Hyperscaler perspective.
‘Leadership vision is the most critical thing. You need leadership to encourage people to upskill gradually as part of their daily work. Learning by doing is the only way to keep up with all the changes in the AI world.’
Seda Akdemir, Partner Technology Strategist, Microsoft
How do you get funding for GenAI use cases?
Enterprise perspective
- Develop POCs as close to production readiness as possible – this makes it easier to get funding approved
- Creating a reusable foundation makes the business case for future use cases more compelling – because you’re using what’s already been signed off from a production perspective
- Focus on the run costs when presenting the business case – framing it as, for example, a monthly cost, makes it easier to get funding
Mauro Meeuws, Product Owner – Hybrid Cloud, NIBC Bank
Partner perspective
‘It doesn’t just have to be about getting funding to put POCs into production. It’s ok if – once you start – you discover a third-party provider has the functionality you need. As you experiment, the business learns more about what it needs and what approach offers the best value for money.’
Allan Chong, Head of AI, Nordcloud
Hyperscaler perspective
Don’t just take a business problem and try to address it by using the fanciest tech. Be concrete on expected outcomes and how they will be measured. For example:
- Achieving more – e.g., personalising the customer engagement journey, boosting productivity, facilitating collaboration
- Doing less – e.g., reducing costs, more accurately forecasting demand, automating
- Being more in control – e.g., more intelligent protection, smarter data labelling, better stakeholder engagement
Seda Akdemir, Partner Technology Strategist, Microsoft
How do you overcome data quality issues?
Enterprise, partner and hyperscaler experts agree
Don’t let data quality issues put you off. There’s even trend of using AI to help overcome data quality blockers by using it to validate data and documentation. So start with data you’re more confident in, and leverage the tech to alleviate issues as you go.
How do you move from experimentation to production at scale?
Enterprise perspective.
- Build POCs in a production-lite state – the more POCs are tested and validated towards the organisation’s production standards, the easier it is to get to the next stage. Although this seems like more effort, it’s ultimately quicker because you’ve built something that can go live fast
- Acknowledge that you never leave the experimentation phase – the tech is constantly evolving, so you need to continuously experiment and adapt
- Manage expectations within the organisation – GenAI products are never ‘done’, they need constant tweaking
Mauro Meeuws, Product Owner – Hybrid Cloud, NIBC Bank
Partner perspective.
‘Experimentation is crucial, and the organisation needs to change ways of working and thinking to be comfortable with that (as opposed to a traditional waterfall approach where you deploy and forget). For example, moving from an LLM to a small language model might be an improvement you need to make.’
Allan Chong, Head of AI, Nordcloud
Hyperscaler perspective.
Microsoft conducted research among internal experts, customers, partners and analysts on success factors for moving to production at scale. (Read the full report here.)
Key action points included:
- Assess the number of business units and processes, length of time in production and age of deployments in the organisation to reveal patterns that may point to opportunities or blockers
- Build intelligent apps on data to improve the intelligence and relevance of model outputs
- Consider using a copilot (or building your own) to accelerate learning and time to value
Seda Akdemir, Partner Technology Strategist, Microsoft
Struggling to scale GenAI?
You’re not alone – 44% of tech leaders say this is their most critical challenge right now.
Get in touch to discuss tailored ways that will suit your organisation.