Aligning AI and Enterprise Architecture for Growth
Navigating the integration of AI experts into traditional IT and data teams poses complex challenges, as outlined in a recent discussion highlighted by Business Insider on the emergence of the Chief AI Officer role. The “right” organizational and reporting structure for Generative AI is evolving, with companies across industries experimenting with the ideal balance of influence and decision making among AI, GenAI, data, analytics, the CIO, Trust officers, the SISO and Line of Business owner including Sales, Marketing, Service, and eCommerce. This post addresses these challenges, focusing on the unclear lines of reporting, coordination of AI efforts across various business units, and line of business owners in orchestrating AI initiatives. Through exploring the shared and unique competencies of technical and AI architects, we offer insights on leveraging existing skills for effective AI implementation within the enterprise architecture.
There seems to be no end of the articles like the BI one above, curriculum, and discussions about the emerging role of AI in the enterprise and as part of the architect’s responsibilities. Read closely, however, and we learn that most of a technical architect’s existing roles and competencies are shared by an AI architect. Unique to AI is a required background in artificial intelligence, including machine learning and large language models. Beyond that the goals are quite the same. Salesforce Architects in particular already have 2 out of 3 of the required skills to help companies be successful. The latest State of IT report reveals that 86% of IT leaders believe generative AI will soon have a prominent role in their organizations. Leverage your status as a trusted advisor to collaborate with the AI experts in your organization to bring it all together.
DATA + AI + CRM
The first step, and shared responsibility between the two architects, is to be close to stakeholders in sales, service, marketing, and logistics where you’ve already had success unifying data sources for analytics and predictive AI to build customer-centric processes. These are the teams that artificial intelligence sets out to help immediately. They are also the teams that architects already have relationships with, and who are already in multi-year, transformational programs to improve business processes across the enterprise.
These two groups of architects must work very closely together to orchestrate how their projects will work side-by-side, and decide which projects need to work together. They will need to work closely together to decide how to prioritize or re-prioritize existing projects, given the reality of the new requirements that AI, and especially AI using a GPT, will necessitate.
When there is no close coordination between the two groups, the door is open for wasted resources and project redundancy. Consider where both work to refresh software, introduce new process innovation, acquire new talent, obtain budget approval, and define the new KPIs for business improvement.
Areas Salesforce architects & AI overlap
- Data architecture: Data cleansing and modeling
- Data Analytics
- Data visualization
- Business Process Optimization
- Software engineering
- DevOps
- Data and system security
- Privacy and data governance
Areas where AI architects bring unique skills
- Deep Learning and Machine learning and specific algorithms such as TensorFlow
- OpenAI, LLMs (BERT, LaMDA, GPT-3, 3.5, 4)
- Reinforcement Learning
- Natural Language Processing
- AI Cloud Services from AWS, Google Cloud AI
The Innovation Window
55% of IT leaders say their organization needs accurate, complete, and unified data to use generative AI successfully.
Innovation mostly happens when people with different skill sets and points of view brainstorm around a common problem. The timing is most important. We call this the innovation window. The Salesforce architect and enterprise architects are perfectly positioned between the business and the tech to identify the areas for collaboration and innovation to capitalize on the work done in data and CRM and hop curves at the right moment during this rapid growth phase in AI.
Generative AI is a great example. An architect may be working with a group from Product Development or Customer Service or with HR. They will be intimately familiar with the key processes, content, data sources, and user endpoints. An AI Architect would be brought in to see what value GenAI could bring, and the payback given the degree of difficulty.
To illustrate the advantages of the two architects working together, focus on specific initiatives inside of your company. We will use the example of Customer Service. For this case, we have a Salesforce architect who talks to the head of customer service. After examining the existing processes, the architect concludes that AI, including Service Cloud Einstein, could increase the speed of case resolution or problem resolution (where there may be no formal case creation). Collaborating with an AI architect to leverage existing customer data sources, knowledge articles, and existing flows could speed up case resolution by 30% while enhancing customer satisfaction and loyalty delivered by using AI to understand customer sentiment, automate responses, and personalize interactions.
Overcoming Resistance
The integration of AI into a mature IT landscape parallels the historical introduction of Salesforce into established businesses. Resistance is inevitable, stemming from fear of change and attachment to legacy systems. Based on insights from the Generative AI in IT Survey, over 70% of IT leaders have concerns about bias, security, and sustainability. The Beckhard-Harris Change Model can guide this transition, where ‘D’ represents the risks if the organization remains static, ‘V’ encapsulates the compelling vision for AI integration enhancing customer service, and ‘F’ delineates the initial tasks crucial for early success. The architects must work together to skillfully amplify dissatisfaction with the status quo (D), create a compelling vision for all stakeholders (V), and specify the first steps towards action to create momentum (F) to ensure this formula tips the scales towards change and overcomes the inertia of resistance. The momentum of generative AI this year has ignited dissatisfaction with the status quo with 99% of IT leaders believing they need to take measures to equip themselves for generative AI. Creating a clear vision by translating AI terminology for stakeholders and presenting the first steps will help overcome resistance to change.
To create a clear and compelling vision let’s return to the customer support example. There are typically 25–30 software applications running inside of any customer support team. The world is Omni-Channel, and that alone necessitates upwards of a dozen technologies. A short list of areas where an architect would leverage AI to create a clear and compelling vision are:
- Content and knowledge management required to solve questions and feed chatbots
- Self Service customer portals
- Messaging systems to engage customers either via the website or through a mobile application
- Email to initiate a case (inbound) or to proactively notify the customer
- Agent desktop to which service requests are distributed, and where issues are managed, customer profiles are viewed, and subsequent notes are entered.
Architects can also identify the first steps to business leaders that the support team could most quickly and efficiently use GPT for three tasks:
- Writing service replies
- Finding and Creating knowledge articles
- Generating post-call work and case summaries
- Predicting where errors might occur
This last point surfaced recently in a case where a chatbot provided incorrect information to a customer, due to errors in the knowledge base from which the answer was derived. This situation of faulty chatbot answers is widespread, and will only be exacerbated when Generative AI is added. Architects are in a position to deliver the guidance that will allow teams to streamline their AI efforts and avoid pitfalls.
The combination of the business functional leadership working with the architects allows for the building of a skills inventory, a process map, KPI definitions, and project prioritization. Back to our example, the business lead knows that they will need to decide which agents to select for the initial trial and rollout and be able to estimate the training, incentivizing, and balance in compensation that will be required when working with the agents.
The core architect will know which service channels will be impacted, and have the experience in leveraging existing work, as well as understand that there will be an impact on handle times, a need to better understand agent availability, and a requirement to track agent activity to ensure that the new technology and processes are working correctly.
Advice on how to begin
Some resources to get started
