Modernisation of AIML Operations, Development Lifecycles & Data Value in Enterprise Value Chains

In March of 2022 this year I was fortunate enough to be invited to present at the AI Business Summit at the Crowne PLaza in Sydney Australia. My conscientious Data Science colleagues at the time under Aruna Pattam and Phil Hermsen (HCL Data Science champions ... I mean that!) were there to collaborate in terms developing a strong narrative that mirrored the ways in which we had been developing value for clients.

Our approach favours the Activation end of maturing to high functioning Value Chain more effectively utilising Data lifecycle focused approaches to continually improve outcomes for Business needs. This Continual Improvement Management approach having been in development for the last 2 - 3 years and tested widely in global clients. This has been amplified in importance with the reemergence of Digital foundation and product management post COVID lockdowns. Most industries have found cost to operate challenges and a desired uplift in Product Management needing emerging versions of AIML Value Chain activations. Furthermore evident in many clients landscapes all of whom are struggling to achieve their data nirvana due to silos, resources vacuums, skills gaps and legacy operations in situ.

Part 1 : TRUST is why we focus on Value Chain activations for AIML Ops and Business outcomes

As with many cross collaboration approaches they evolve over time and in the shadow of commercial pressures for Platform as a Service sales as well as client preference for handling small Data capability uplifts rather than thinking about value acceleration. The former needing much in the way of trust. Trust is a big player in AIML maturity and delivering on value.

Most enterprises really struggle here. To nurture the business outcomes beyond technology and data implementations means programs of complexity and culture change. What I have witnessed is that effective transformations must go beyond technology requirements to deeply advocate for holistic change across change networks but with the accelerated depth in Data democratisation where ...

• Organisational learning and growth is at the heart of the Program vision.
• Business Value Chain use cases are Insights and Data Value focused not only about Data integration/consolidations,
• Progressive in appetite and feature deployments to a leaner operations future state,
• Change Strategy embracing – Customer/Employee and Process focused,
• Engage the Change network internally and sponsors at from start to finish and at multiple levels,
• Business/LOB integrated Strategies, and product design focused whether in Operations, Manufacturing, Supply or Customer Value Chains. (Noting the huge spectrum of variation here...)

However, all of the above is a mountain to climb for low maturity Digital Organisations. There are many who have spend 10s if not 100s of millions on developing IDM practices from the ground up. Often this slow emerging maturity has been the destroyer of Data or AIML trust. Some key problems in the slow and costly progress in custom AIML Operations rigor to get to the IP and benefits might be,

1. AIML OPs Lifecycles are very hard to coordinate across capability matrices of all Enterprises. differing life-cycles between model maturity and Product Analytics Lifecycles.
2. Often politically difficult due to budget and time competition,
3. Majority of organisations fail to achieve integratable value in this.
4. Many Enterprises spend 10s if not $100s Mn and years getting AIML up and running for only a few implemented business cases.
5. Pockets of AIML capability arise in siloed solutions for Cyber, Performance, portfolio analysis... but have not yet become part of the data life-cycles for the value chains of products and modernising business.

What is AIML Operations activation strategy?

With AIML Operations maturity in the Web 3.0 era the end requirements in terms of Digital workflow and services design go far beyond the Platform as a Service (PaaS) offering that most platforms stitch together for their Platform Proof Of Concepts. In a less polished reality these needs are drawn from the always emerging nature of value in terms of the existing capabilities of client Business Strategy, Product Experiences/Orchestrations, Customer Benefits or Operations Optimisations. All of which will be the downstream recipients for new, leaner and adjusted Service models as part of more integrated Data Lifecycles with future states. Many are however still getting a handle on the value of Data Services for deeper Analytics adoption in across organisations let along the integrated Digital Ecosystem.

Here-in sits the challenge

A Platform as a Service buying approaches seems attractive out of the box with their 'sales slick value sample use cases' but in reality all platform refits require Significant Change in multi-departmental-skillsets with prolific change in granular ops and orchestration use cases. All in all the platform alone does not help accelerate time to value but more accelerate some value at speed like Orchestration or Operations automations. This is often a huge frustration for IP focused organisations, struggling to get to modern Data value and cultures off the ground.

Make the approach more pervasive and make the adoptions cultural

Here is born the need for more integrated POCs driving Insights and AIML model management in extended AIML Ops capabilities end to end. After all, the real long awaited need is in engaging in the value chain uplift (Product or Ops) and the Data Lifecycle management.

"...this is why an informed Procurement approach is critical for low maturity organisations in the Data Value realisation and AIML space ... "

Activation focused is about bringing the End to End piece to life for the organisation, about democratising access to customer feedback, perhaps social platform listening, GEO event inputs triggering understood data lifecycles that both matures the Data Ingres Operations, promotes training of the Language Processing models in the AIML teams, enables meaningful triggers and activations for Experience and Processes.

Example automated intelligence in events driven optimisations outcomes we see in industry might be ...

• Automated edm/sms/callout alerts on market dynamics for high risk portfolios in Superannuation and wealth management,
• Alerts for High risk weather events for house/car owners in certain GEOs,
• Better chat bot customer services handling in B2B services,
• Automated deployments scheduling in Work Order servicing and intelligent work order optimisation for work crews to reduce costs across multiple contractor scenarios.
• Asset Maintenance Automation and procurement activities alignment for warehousing limitations
• SKU preorder capabilities in predictive flow optimisations in sales and supply chain activity
• Logistics timing and re-routing for
• Risk profiling in Image analysis for site and asset maintenance
• Dependency analysis in strategic portfolio management

Honestly, the list of possibilities to optimise organisations and their understanding of engagement seems endless ...

What are the common factors that underpin success?

All these above examples need a strong, top down owned, accessible, democratised view of

1. Flexible Data activations to accelerate value chain insights and automations,
2. Strong understanding of statistical inference in analytics insights !IMPORTANT,
3. Events triggered Data Lifecycle : essentially a view both preferential Digital enablements,
4. Omnichannel Activations in loosely coupled view and engagement states for Experiences and processes,
5. Near real time data supported in the ecosystem with Events driven architecture across platforms supported by the Digital Integration strategies.

____________________________________

Part 2 : Getting your hands dirty in end to end AIML Activations and various use case Digital modernisations

Here we must enter the realm of Design and Digital Strategy influencing Enterprise Architectures ecosystems of Services. So it goes without saying that real on the ground work have many players in the room. Many of whom have P&L interests that don't support alternative innovations concepts where accelerated value could happen in many ways.

It will likely be ....

• Data Consolidation, Data enrichment, Data Cleanse, Data lineage
• Use case flow simulation with Data mining and simulated IOT gap data
• Model evolution and dimension validation and regression analysis

Working on Proof Of Concept's with an end to end nature sounds tough and it is, but with the ideation and collaboration risk comes great rewards! The great reward paying out in groomed strategic approaches bringing together Line of Business, Human Resources, Digital Enterprise Architecture, collaborative Product-Digital Marketing/Operations, improved core platform functionality, more flexible approaches for Data Lake and Data Serving, Learning and business intelligence in the key transactional levers, behavioural insights, Deeper personalisations that are meaningful to value, optimised process that can be automated efficiently and ecosystem flexibility with security. It can be a cornucopia of value if the readiness is there in the other side to train and change push the necessary cultural understanding of the employees and BAU owners for better levels of adoption.

In the lifecycle of Data value use cases quantifiability with qualitative insights is key

The value rewards in all the mentioned use cases are the insights, the activations are the evolutions, but it is also the hard work and how 'as an enterprise' you choose to get there. Ensure that your programs have the right level of thought leadership where

1. Strategically aligned Services and CX leads with strong Data Capability are are the forefront of the conversations. For E.g Personalisation with meaningful inputs from AIML insights on taxonomy of Content relevance, SEO optimisation, personal journey preferences and direct mailing, sms or contact opportunities through preferred channels.
2. That there are Data Strategists who have real activation capability in Data Ops in the practice where your Data Science capabilities live,
3. That pragmatic strategic leadership is at the table and that there is a clear and relevant pathway to the outcomes that will feed further value growth to a well modelled and groomed Growth or Optimisation.
4. Repeatability in the landscape of meaningful intelligence driven goals is key, sometimes looking at new Operations innovations might be the key to breaking the mold.

Lastly reading between the lines I would say this ... the accelerators market for commoditised AIML Models that 'accelerate time to value' is erupting at the moment. Those Products that help organisations jump more deeply into AI and Machine learning value are a key enabler to the organisation who has not yet made the leap to Data Lake or scaled and maintained their own ML models and Model Lifecycles ongoing. By the way, huge costs here.

Commoditization and Democratisation in the applied Machine Learning Model market

Without dropping any names all true Data and Analytics approaches need to be leaning into real time value flow analysis, modelling and simulations optimisations whether customer, supply chain and logistics, mining/resources, workplace optimisation, regulatory processing operations, agri-business product to manufacturing flow, utilities asset management or even superannuation personalisation strategy. The new wave of commodised products is out and they are shaking up the IDM practices of companies that are slowly wading through their Data consolidation and practice maturation issues. Strategic choices on Data Value chain maturity versus accelerated value may in fact hold the key to the CIO Innovation reputation of many data democratising activities.

Use case by use case brave organisations will make these accelerated changes and slower less insightful, more costly legacies will be discarded.