image copyright: Salesforce
Written by Ian Gotts
Salesforce’s AI vision is compelling. It is not a recent knee-jerk reaction to ChatGPT. Their journey started back in 2016, when Einstein launched. ChatGPT has simply captured people’s imagination and made AI real for everyone. But as we all know GPT is far more than ChatGPT.
AI is way bigger than that, and is really the next big thing after mobile and social changed the tech world over the past five years.2016, MARC BENIOFF, CEO SALESFORCE
It doesn’t matter what industry you operate in, customer expectations are climbing. 85% percent of customers expect to be dealt with consistently across all channels. Surprisingly 62% of customers expect you to anticipate their needs. This is now becoming possible when your employee and customer experiences are powered by AI.
AI is not a trend but a seismic shift and we are at the very start of it. Remember back (depending your age) to the first calculator taken into school, the start of the internet, or launch of the first iPhone, or when Salesforce showed what could be done with cloud computing. And then think how far each of these evolved. Before each of these technologies became mainstream there were years of prior development. The key point in time is when they hit the mainstream. So, remember that we are at that time now, so you can say, “I remember the start of AI”
Over the last 5 years AI has proven to have a strong ROI: 67% increased revenue growth and 79% reduced costs. In the last 6+ months ChatGP, given a glimpse of what is possible with AI and today has become a priority for 67% of IT leaders.
The coming wave of generative AI will be more revolutionary than any technology innovation that’s come before in our lifetime, or maybe any lifetime. It is reshaping our world in ways that we’ve never imagined. This AI revolution is just getting started2023 MARC BENIOFF, CEO SALESFORCE
Delivering AI + Data + CRM: AI Cloud
When we think of AI, it is normally in the context of what can be done with customer data. That is at the heart of Salesforce’s AI go-to-market strategy. Aggregate your data from CRM and other sources into Data Cloud and then the suite of AI-enabled clouds can do their magic.
If we think about the ability of AI to accelerate time to value for Salesforce implementations the data is very different. That data is information about the business (process maps and architecture diagrams), the configuration of org / apps (metadata), design standards (Well Architected Framework), and regulatory standards (ISO9001, GDPR, FDA etc).
As Clara says, garbage in, garbage out. The challenge is to get clean source data and then put in governance and processes to prevent those sources being polluted as the business changes.
Data is fuel for AI — without high-quality, trusted data, it becomes ‘garbage in, garbage out.’ AI pulling from data sources that are irrelevant, unrepresentative, or incomplete can create bias, hallucinations, and toxic outputs.CLARA SHIH, SALESFORCE’S CEO OF AI
This program is a major transformation initiative that will require analysis of your current market and then redesign of your business processes. You need to evaluate the sources of data and the downstream dependencies, wading through the inevitable technical debt. Once you have architected the solution you need to implement it.
A Change Intelligence Platform will enable you to deliver the program more quickly by providing the intelligence on the risk of making changes. It will help you understand which data you want to use, and what is redundant. It can reduce this org analysis by 50% and development rework by 80%. It delivers an automated analysis of the org configuration, a centralized source of all business analysis documentation and a tight integration with DevOps tooling.
Elements.cloud is a multi-cloud Change Intelligence Platform, supporting transformation programs across the entire IT landscape. Multi-cloud is key. Your changes business processes span business units and apps. The data feeding Data Cloud will come from different apps. All of this can be documented, managed and implemented with Elements.cloud.
Now Elements.cloud is augmented by AI/GPT. It can create user stories from process maps and then recommend solutions personalized to a client’s Salesforce org and aligned with the Salesforce Well Architected framework. What used to take 8 hours is delivered in less than 5 minutes with greater accuracy. The better the process maps and the org metadata documentation, the better the recommendations. But many (most) organizations do not have great business or org documentation. But you are creating it as you deliver the AI program.
Elements.cloud will allow your team to develop and collaborate on the business analysis and change documentation. You can accelerate the solution development. You can reduce rework.
You don’t need to wait for EinsteinGPT to be available to launch your “Salesforce AI program”. EinsteinGPT is just going into a pilot phase and there is no GA date yet. The work to understand the new processes and the data streams will take time. You need to get started now.
The core capabilities of Elements.cloud that you need for the org preparation are already available. The ElementsGPT pilot starts this month and GA is before Dreamforce23. These features will help you accelerate the implementation of the changes to the org. But you don’t need to wait for ElementsGPT.
The great news is that even if you never implemented AI, the work you are doing has a positive ROI, that is measured in weeks, not months. Just use the promise of AI to convince your senior management to get started.
Delivering the program
This Salesforce vision is compelling and the entire organization is now focused on delivering it for customers: R&D investment, research and marketing content, and sales. The potential benefits of data-driven automation is that data insights can be transformed into actions that will grow revenue, improve collaboration, and boost productivity. This is driving clients to invest in Data Cloud which is at the heart of the strategy.
Today, 49% of companies are using automation to provide better integration of customer data and improved customer experiences through the coordination of different business units. Not surprisingly, 72% of organizations have automation as their strategic priority, with 95% of technical leaders focusing on process automation.
“Automation only comes before Simplification in the dictionary”IAN GOTTS, CEO ELEMENTS.CLOUD
This is how we see the scope of the program and the detailed steps. We’ve highlighted where Elements.cloud can support you. And not surprisingly, we are with you every phase of the way.
Agree AI-enabled vision
Work with stakeholders to establish the vision for the business, enabled by AI.
- What are the objectives and measures for success, and over what period of time.
- How does this dovetail with other strategic business initiatives?
Design how operation works
- Working with the key business units, run workshops to brainstorm the art of the possible with AI Use Elements.cloud to capture the brainstorm results.
- Redesign the business processes from top level down in UPN format so that AI can read it. Do this in live workshops This is across the business to deliver the program, not just those related to Salesforce Use Elements.cloud Process
- Capture business requirements, linked to process activities Use Elements.cloud to document requirements
Establish the data requirements
- Build metadata dictionaries for the current org(s) Use Elements.cloud to automatically build and analyze
- Analyze the data sources that are required to deliver the vision and the new business processes. Document the data flows that shows how the data is populated through to how it is streamed into Data Cloud. Use the Salesforce Diagrams Standard and link metadata items in the metadata dictionary. Use Elements.cloud to capture as Architecture Diagrams
- Document why metadata items were created as you come across them in the MDD (Metadata Description Definition) format so that AI can read it Use Elements.cloud Metadata Dictionary Description field
- Identify optimization / clean up opportunities as you come across them Use Elements.cloud Metadata Dictionary optimize tab
- Design data governance processes Use Elements.cloud Process
Design the solution
- Create User Stories from the activities on the process maps Use Elements.cloud GPT to autocreate
- Validate/update the generated user stories Use Elements.cloud Change – User Story view
- Create Salesforce org metadata changes / recommended solution Use Elements.cloud GPT to autocreate
- Create the non Salesforce solution and document in the user story Use Elements.cloud Change – User Story view
- Validate/update the generated recommendations Use Elements.cloud Change – User Story view
- Group user stories into releases Use Elements.cloud Releases
- Pass User Stories to development Use Elements.cloud integration with Jira / DevOps tools
Deliver the solution
- Create Data Cloud integration to sources Use Elements.cloud to sync Data Cloud and build metadata dictionary and create metadata dictionaries for key 3rd party systems
- Use development tools to make build/updates Salesforce and apps to support the business processes. Include new Salesforce AI Cloud (e.g. EinsteinGPT)
- Test solutions. Use the acceptance criteria and test conditions in the user stories. Use Elements.cloud Change – User Story view
- Document new or changed metadata using Metadata Diagram Definition Use Elements.cloud metadata dictionaries
- Run UAT against the business process diagrams including data sync to Data Cloud Use Elements.cloud – Processes
- Confirm sync of data to Data Cloud
- Use the process maps to design the change management program [use Elements.cloud Process]
- Deliver the change program
- Go Live
- Capture end user feedback [use Elements.cloud Process and Salesforce In-App Help]
- Spot check results from AI: “Human in the Loop”
- Survey employees and customers
- Scope next phases
AI relies on clean sources to be able to deliver consistent results. That data is aggregated in Data Cloud from sales, service and external systems. Analyzing the data streams is challenging in complex orgs, where a field could be populated by automation, code or 3rd party integrations. High levels of technical debt mean that there are probably several fields that look and sound similar, and require detailed analysis.
Then there is the issue of cleaning up data. The numbers are staggering. By 2025, there will be 100 zettabytes of data in the cloud and this is forecasted to double by 2026. Currently data is being created more quickly than teams are able to integrate it, let alone harmonize it and drive out meaningful insights. So it’s not surprising that 36% of IT teams say that data is a major challenge for transformation programs. This knocks the confidence of business leaders, 33% of whom say that they are not getting the expected insights from all their data.
New (old) skills
The core skill is business analysis to deliver the project. Arguably this is more important than Generative AI skills, as this side of AI is being dealt with by the apps like EinsteinGPT and ElementsGPT. The AI apps that help you configure Salesforce are driven from an understanding of what changes are required. You won’t be typing all that information into a ChatGPT window. You won’t need prompting skills.
The apps decide what information they need and get it for themselves. And this information is discovered and documented by business analysts. It is stored as requirements, process diagrams in the Salesforce UPN standard, architecture diagrams in the Salesforce Diagrams standard, and a metadata dictionary with metadata documentation and all the dependency and impact analysis.
If your org has very little of this in place, then you cannot take advantage of the staggering productivity gains that AI can give you. If you do not have the business analysis skills to develop documentation in each of these areas, then you will fall behind.
The great news is that we are not expecting you to learn skills in technologies that are still evolving. That work is being done by the development teams who are building the AI apps. They are getting to grips with training LLMs, tuning semantic databases, developing agents and stuff that hasn’t been invented yet!!!
The skills you need are core business analysis. Skills that have been proven and developed over nearly 20 years. UPN was developed in 2004, but the principles behind it date back to the 1960s. The scope of the training and documentation you create is in the Business Analysis Certification. So pass that Cert and you are in good shape. To help you learn, the Elements Academy has great training courses and the Elements app has a free Playground so you can hone your skills.
Tech debt is an inevitable result of agile systems that evolve to support constantly changing business needs. As organizations accelerate their digital transformation plans in order to accommodate hybrid working and respond to market forces, the systems that underpin their operations need to change. This often leads to tech debt. Salesforce is also investing heavily in its platform, so customer functionality that was previously developed in-house may now be integrated into the core platform, making existing configuration unnecessary.
The real issue is that in many (most?) orgs the levels and hotspots of tech debt are not understood. This makes it very difficult to scope and estimate changes with any level of confidence. This program requires confidence that the correct data sources are being used to populate Data Cloud. Time needs to be factored in for the analysis based on the level tech debt.
Luckily Elements.cloud provides automated insights in the shape of org impact analysis, dependency analysis, and custom list views of metadata. And whilst the program may not be explicitly removing technical debt, highlighting opportunities to clean up that are discovered during the program will make future changes quicker and easier.
You may come across clients that believe the levels of tech debt are so high that they wan to throw the org away and start again. In our experience it is rarely necessary, and it is often a reaction based on little understanding of what is in the org. Once you use Elements.cloud to build the metadata dictionary and the analysis, you can have a facts-based discussion with the client.
These are natural concerns when AI is being applied to customer or employee data. IT leaders are the most concerned with 79% worried about security risks, and 73% worried about biased outcomes. Generative AI giving wrong or misleading results in a personal setting, such as writing a poem or a complaints letter. This is very different than when AI is making investment recommendations or staff scheduling decisions. This has reputational, operational and regulatory compliance risks.
Senior IT leaders need a trusted and secure way for their employees to exploit AI. It cannot, and should not, be banned. 79% of senior IT leaders reported concerns that AI brings the potential for security risks, and another 73% are concerned about biased outcomes.
In 2019 the Salesforce AI Research Team published their trusted AI principles (transparency, fairness, responsibility, accountability, and reliability), intended to support the development of ethical AI tools. They have just (June 2023) revised them and they are discussed in this Harvard Business Review article. The principles are accuracy, safety, honesty, empowerment, and sustainability.
When considering using AI for org management, some of these risks are lower. You are not dealing with customer data, but instead it is app configuration data. There are still security risks, but the issues of safety and honesty are less of a concern.