Did you know that your role as a Salesforce professional is on the brink of a seismic transformation… Alarmist nonsense or an exciting future? Soon your job could look almost unrecognizable compared to what it is today. This isn’t a distant reality, but a sweeping shift that’s happening right now. Are you prepared for this change?
Generative AI is the force behind this revolution. This technology is not just redefining the parameters of problem-solving, but also opening up entirely new patterns of innovation. The capabilities of generative AI go far beyond what people have experienced in playing with ChatGPT. When applied to org management, generative AI has transformed the speed and quality of the change decisions that can be made. And this is redefining every role in the Salesforce ecosystem.
Are you ready to embrace this change? Your journey starts here.
How Generative AI / GPT Works (And Doesn’t!)
For the last couple of months the world has been taken by storm by one specific generative AI: ChatGPT. The GPT LLM (Large Language Model) is trained on huge amounts of data, most of which is publicly available. The first GPT is over five years old. GPT-3 was trained on 570 GB of data from snapshots of the internet from 2016-2019 at an estimated training cost of $4.6M.
It used 1.3 gigawatt-hours of electricity – enough to power 121 homes in America for a year! Training GPT4 was estimated to be $100M, which gives you some indication of the step change in its capabilities. Advances in GPT, the use cases, and supporting technologies are changing almost weekly.
But there’s a catch: while this data enables it to understand and respond in a natural language, it doesn’t know anything about how your company operates, what specific configurations have been made in your Salesforce org, or what your business’s unique practices and standards are.
Prompting
This is where ‘prompting’ fits in. A prompt provides the information that GPT does not know about and the format in which you want the result of your query delivered to you. It’s like saying: This is what I want, and here is the specific context you need to know to produce it.
Here’s what I have discovered by actively using ChatGPT and other apps in every area of my business and personal life for the past five months… The best way to get a great answer from ChatGPT is to spend the time creating a comprehensive prompt. This prompt could be almost a full page of text in length – far better than a loose question followed by a long conversation back and forth.
In the Elements Product Management team, we have developed a library of prompts we use in different parts of our process. The prompts have places where we substitute text. For example, we have an internal website where we post information about new functionality ahead of a release for Marketing, Customer Success, and Sales. The pre-built prompt is updated with the key points of the functionality and it produces the webpage for us.
But for org management, the task of providing sufficient context can be a real challenge. Firstly, you shouldn’t need to type in all of the context data. Secondly, the volume of context data in the prompt rapidly overwhelms apps like ChatGPT.
Apps, Not Prompts
So, how do we resolve this? The answer lies in automated prompting. Apps know the context and the background data, and use it in one of their highly-tuned prompts. This is then sent to GPT to use its reasoning capability to get the result. The art of prompting is within the app, but as the app is focused on a narrow set of use cases, these prompts can be fine-tuned to optimize the results.
Self-prompting AI agents are becoming the norm. Imagine an AI that’s not just responding to prompts but is proactive, interpreting limited input, seeking additional information, and finding its way to the right result. This isn’t just a wild dream. Apps like ElementsGPT are merging their domain-specific knowledge with the reasoning and creative prowess of GPT models, automating tasks in a way that would have seemed like magic just a few years ago.
Jobs, Not Tasks
Let’s take a moment to understand what this means for jobs. A Goldman Sachs report recently predicted that around 300 million jobs worldwide could be automated through AI. That’s a staggering number! But here’s the twist: AI is brilliant at automating tasks, not entire jobs.
When you break down any job, it’s a collection of tasks. Some of these, like analyzing large amounts of data or generating text, are right up AI’s alley. As these tasks become automated, it doesn’t mean the job disappears. Instead, it frees up time for the human in the job to focus on tasks that have been sidelined, tasks that require a human touch like creativity, empathy, strategic thinking – all things AI isn’t good at.
The AI revolution isn’t about replacing jobs; it’s about redefining them. It’s about taking the mundane and repetitive tasks off your plate and allowing you to dive deeper into areas where you can truly make a difference. And when you wield the power of AI, you’re not just doing your job, you’re transforming it.
Far from being a threat, Generative AI is your secret weapon. Think about it… It’s not AI that poses a risk to your job, but someone armed with AI who knows how to wield it definitely will be. The key takeaway is clear: Get ready to team up with your AI sidekick and let the transformation begin!
Future of Salesforce Work
The future of work won’t be about grinding through data or drafting content. It will be about orchestrating AI agents and refining the art of input and context creation. It will mean harnessing human communication skills and documenting insights with clarity and detail, because remember, with AI, it’s always ‘garbage in, garbage out’.
In this brave new world, Salesforce roles are set for a significant transformation. AI tools will absorb best practices in coding and Salesforce org configuration, taking care of documentation and freeing up human roles for higher-level tasks. Routine will give way to strategic analysis.
The most sought-after skills of the future will be business and architectural analysis. Being able to convert a stakeholder’s requirement into a detailed blueprint, revealing hidden assumptions and stakeholders, will be essential. It’s these inputs that will enable AI agents to suggest and implement effective solutions. Get ready to turn problem-solving into an artform, because in the AI era, the quality of the context you provide will be ‘king’.
“This is the most exciting time to be a business analyst that I have seen in my 30 years of BA work.”
– Ian Gotts, Elements Cloud
Introducing ElementsGPT
Elements.cloud is a Change Intelligence Platform that provides the org intelligence required to make change decisions faster. This intelligence is business analysis and org analysis. ElementsGPT adds an AI layer, which enables you to make better quality decisions, even faster. This brings us to the introduction of ElementsGPT, a tool that perfectly embodies the future we’ve just described.
ElementsGPT combines:
- Knowledge of your Salesforce org’s configuration
- Insights from your business analysis
- Understanding of best practices from Salesforce’s Well-Architected Framework
- Updates from the recent release notes
- The full power of GPT
ElementsGPT is not a ChatGPT add-on, but a sophisticated AI platform using agents, semantic databases, and a query engine that goes to GPT for reasoning. ElementsGPT creates the prompts by understanding the context – where it has been launched from. And then it queries resources (org configuration, business process data, Well Architected frameworks, and governance standards) to get the right data they can feed to GPT in bite-sized chunks to reason and collate the results. This is the future of org management.
Click here to watch our recent webinar with Salesforce Ben: The Ultimate Guide to AI & GPT for Salesforce Org Management on Demand
Currently, ElementsGPT has three primary features.
1. Writes User Stories
Highlight one or more steps in a UPN Process Diagram, click Create user stories. It will create a user story for each resource, complete with acceptance criteria, and it will link the user story to the related process step.
The user story is stored in Elements so that it can be managed through the lifecycle and added to releases. It means it is also visible in Jira, Salesforce DevOps Center, or Copado through the integration. The quality, accuracy, and consistency is amazing.
2. Recommends Solutions
For any user story in Elements, it recommends the changes that need to be made to Salesforce (new metadata or update existing metadata) based on your org metadata and Well Architected framework. User stories don’t have to be automatically generated. They could be manually entered, synced from Jira, Salesforce DevOps Center, or Copado. Alternatively, they can be imported from a spreadsheet
Not only are the recommendations staggeringly fast and comprehensive, but they can also help to reduce technical debt. ElementsGPT knows about every metadata item, so will recommend updating existing metadata items where possible, rather than simply creating new duplicate items.
3. Ask the Org, Well Architected, or Release Notes
ElementsGPT has a conversation interface in your org as it has an intimate knowledge of all your metadata – even the seemingly unimportant details. You can also query the Salesforce Well Architected Framework or the Release Notes.
This is free format prompting, so the more specific and directed the prompt, the better the answer. What’s interesting is that the results depend on good quality documentation (clear, with no assumed knowledge, unknown abbreviations, or euphemisms). A feature that is EOL in the release notes could be described as EOL, end of life, not available, or disabled. This is worth testing; if you read it and have to ask someone for clarification, it is not clear enough.
Preparing for ElementsGPT
ElementsGPT is entering a closed pilot with GA due before Dreamforce ‘23, but you shouldn’t be waiting for GA. To take advantage of ElementsGPT, you can start preparing right now:
- This is the time to start documenting your org. The quality of your org documentation will directly feed into the quality of outputs you will get from ElementsGPT. Your first step should be providing detailed descriptions on your Objects, fields, and automations.
- Document your existing business processes using UPN methodology. There are free resources, including a Trailhead course, the Elements Academy course, the Ryan Reynolds workshop, and the Elements Playground.
- For each step in your business processes and consider which Salesforce metadata items are used, or which support the process. These are important and need to be documented first. Use the “description” field in Salesforce.
- Look at which metadata items have been most recently changed, or are due to be changed in future projects. Again, document them using the description field.
- Learn how to write high-quality user stories – the recommendation engine will only provide recommendations for the solutions based on the quality of the details you feed it. Do your stories follow the standard format? Do they have acceptance criteria?
I know I’ve said it before, but the results really do depend on good quality documentation, which should be clear, with no assumed knowledge, unknown abbreviations, or euphemisms. The test is this: If you read it and have to ask someone for clarification, it is not clear enough.
Final Thoughts
In conclusion, the AI revolution is not just a change; it’s an evolution. It brings with it a myriad of possibilities for innovation, efficiency, and the transformation of roles within the Salesforce domain and beyond. Embrace this change – improve your business analysis skills – and the future of org management is yours to command!
Find out more about ElementsGPT.
Comments: