If I have to listen to another technologist promoting “AI as a key component of the CIO’s agenda”, I am going to start getting a little irked… AI is not another app that can be installed and rolled out like a Workday, SAP or a ServiceNow. I even had to listen to an IT executive asking me whether he should “leave AI in the hands of SAP as part of their S4 upgrade”. Not only that, I noticed a well-known analyst firm promoting a webcast last week advising “CIOs how to rollout RPA”. Really?
One of the biggest issues in our industry today is the abject failure of the business teams who design and own the processes, to partner effectively with their IT teams to deliver automation and AI that supports the business vision of where the business leaders want to take it. IT people are not clairvoyant – they can only aspire to deliver what their business colleagues clearly instruct them to do. Otherwise, they’ll just buy all these fancy software suites and say they did their bit for AI… So enterprise leaders have to knock the heads of their business and IT teams together and get them partnering effectively to design a roadmap that takes them and their data where they need to go to stay competitive. There’s no time to keep pointing fingers, we just need to sit down and figure out how to work together in much more effective ways than we have over the past few decades.
Embracing AI is all about crafting the anticipatory organization, one that is hyperconnected across its ecosystem, its customers, employees and partners
The whole purpose of AI in the enterprise is to have business operations running as autonomously and intelligently as possible, which means we need to build enabling IT infrastructure that supports the business process logic and design. People are talking about “re-platforming the enterprise”… this is really about redesigning IT to support the business needs, to help the business respond to customer needs as soon they occur, and have the intelligence to anticipate the needs of their customers before its competitors can.
Enterprises need to be as hyperconnected and as autonomous as possible within their business environments if they want to pinpoint where disruption is coming from, where to disrupt and how to keep reinventing themselves in an unforgiving world when we no longer have time to rest on our laurels:
The problem for IT is that AI doesn’t come packaged in a nice box with an instruction guide
I’m sorry to be mildly offensive here, but AI and automation are only effective when they are designed to solve process and business problems, not check another box on the CIO’s resume. While it is important to keep the IT team in the communication loop so that it is ready to provide the right infrastructure and technology stacks required for operationalizing AI solutions, the steering wheel of any business application of AI must be in the hands of the businesses. Smart businesses know their key pain areas and can identify the most relevant and feasible business cases. They own the data, they know the context, and how a process should run when it is augmented with appropriate AI techniques.
For many firms, the day they implemented their first ERP was akin to pouring cement into their enterprise
The reality is the ERP system of the last 3 decades is no longer the system of record for ambitious, hyperconnected enterprises. It is a rigid suite of standard processes that keep when wheels on a legacy operation. The emerging system of record is the data lake itself, when the business leaders have the ability to extract the data they need to make the right decisions, or have systems that can start to help make intelligent decisions for them.
So let’s examine at the interplay between business and IT with these emerging AI-driven environments with 10 prescriptive activities business leaders and IT leaders need to put into effect, if they want genuinely want to develop AI capability that takes them into this hyperconnected state:
10 AI activities the business teams must lead to ensure AI success
- Prioritize use cases from AI technology availability. The business team must prioritize AI business use cases from the initially identified list of potential AI application opportunities. The team must demonstrate its process knowledge and desired end-state scenario to help the IT team to ensure effective project coordination and outcome-setting. Using external consultants at this phase can be very effective to ensure the best business/technology fit.
- Develop the AI Business case: The most critical step, where the business team must set initial benchmarks, define pre- and post-process improvement metrics, and estimate target benchmarks.
- AI feasibility analysis and specification development: Business teams must solicit help from IT teams for their expertise with items such as technical feasibility analysis, infrastructure requirement specifications, and technology stack selection. Other areas are technology cost estimation, deployment, and production release,
- AI Technology cost estimation: Developing estimates for the cost of technology stacks and solution deployment efforts must be the purview of business teams, but it requires significant and detailed input from the IT team.
- AI Data preparation and identification: Business teams must ensures success by identifying and preparing the data for training algorithms and building models. The team must solicit assistance from analytics and data warehousing teams.
- Coordinate with partners: During design phase of the target process model, the business team should must provide input to implementation partners (both internally and with their consultant/services partner) regarding ontology of the problem domain, the existing process models and rules. Teaming here with IT is essential, but the business team must define and communicate the business and process needs effectively.
- AI Testing: The business team must lead testing the models against the project goals during the early POC and pilot phases
- Manage effective AI feedback loops: To make use cases fir for production release, the business team must provide detailed, regular feedback on the accuracy and performance. Again, they need to work with implementation partners, which may be internal teams from an AI CoE or external partners.
- AI Training: The business team must be responsible for budgeting, planning and executing the training for large AI user teams, encompassing all of the staffing resources, external consultant costs, processes and task owners that are involved in the implemented use case.
- AI Deployment: Deployment doesn’t end once the use case is in production. The business team must continuously monitor the model’s outcomes, maintenance, and updates during the inferencing phase, and if the problem context changes with new rules or data, the team needs to add new dimensions and models and create new clusters. Users may also require retraining, especially as processes may change over time. There will also be the need to monitor change management issues, potential legal issues with data privacy / staffing impacts etc.
The Bottom-line: AI is a business issue that must be directed and managed by business executives, supported by technology experts. CIOs who ignore this will fail
The business team should seek help from IT in terms of infrastructure and tech stack needs, but it needs to own and run the AI projects because it owns the data, context, processes, and rules and understands the pain points.
CIOs will face an existential fight if they don’t start genuinely enabling the business. The world where IT was all about mitigating outages and avoiding risk is being replaced by one that demands speed, agility, and a genuine understanding of the business.
Being tech-savvy isn’t enough anymore… just knowing where to build a data center is pointless if you don’t know what the rest of the business has planned. And this IT obsession of continually trying to upgrade ERP solutions, when most business units these days can handle it. That’s the pitfall of the old traditional IT approach – we have to make sure we never get cemented in like that again.