Months to Minutes. Documentation on demand

TL;DR

Salesforce customers have built complex systems, “black boxes”, that hinder project delivery and reliable AI Agent deployment.

A lack of value for Salesforce is a concern for senior management, who feel powerless to change the status quo. Again, this is a customer quote, but it echoes what they all feel:

“What keeps me awake at night as Vice President is that we have so much stuff we’ve built over time on Salesforce, and I’m super worried that the next person who walks out the door is going to leave us even more vulnerable than where we are today. We don’t have a knowledge base that is accurate. And I don’t want a knowledge base based on what somebody thinks is in their head. I want a knowledge base based on what’s really in the system.“

This is from a recent call with a Salesforce Product Owner, but over the last 8 years, we’ve heard the same thing from every customer.

“People are requesting things for Salesforce. They think they know Salesforce best and they’re they’re coming to us with the solution. We know they are not considering xyz because they (and we) can’t do the impact analysis. And I am about to pull out my hair. This is madness. We’re no longer in the wild west. We are well past drawing Lucidcharts.“

And the question is always, “Can’t you take all the metadata and draw the process diagrams that explain how my org works so we can see the implications of making changes?“

Matthew Morris (Salesforce CTO & DCX Innovation Director at Capgemini UK), highlights why this has even more importance.

“Every Salesforce customer is about to rewire their business processes. Whether your org is 2 years old or 20, it’s full of assumptions that won’t survive the AI era. Before you can transform, you need to understand what you’ve got.“

The release of Configuration Mining from Elements.cloud is a pivotal milestone that changes the way we think about documentation. It provides an on-demand “blueprint” to visualize, understand, and take action.

Configuration Mining is a family of documentation that can be generated automatically from metadata; Process Diagrams, Data Models, and Order of Execution Diagrams.

This has been the holy grail of BPM (business process management) and metadata management for the last 20 years.

State of the Org

SaaS applications are incredibly easy to configure, “clicks not code”, and this often leads to building the wrong things fast. Salesforce’s power and flexibility is a double-edged sword, resulting in complex, undocumented “Jenga towers” of tech debt. We are seeing 50% of the objects that are built are never used. 41-80% of custom fields are never populated.

Most customers operate in a “black box” – they can’t see what they’ve built, how it connects, or if it’s right. This lack of visibility means they can’t trust the underlying data or processes. This kills agility and increases risks for every project. To mitigate the risk, there is expensive, manual, and time-consuming org discovery that documents the org. This is critically important at the start of every project, particularly if the team is new to the org. Now this can be automated, there is a huge ROI –  “months to minutes”.

A large insurance company showed how much poor documentation costs.

“We’ve just spent $500k and 6 months with (major consulting firm) documenting our system. Configuration Mining has done it in 5 minutes.“

Agentforce: Not every project is an agent. But Salesforce is betting its future on AI Agents (Agentforce). You cannot deploy reliable, governed AI Agents on top of an unknown, messy foundation, the “black box”.  This isn’t an engineering challenge AI alone can fix. It requests a deep technical analysis of metadata and dependencies that explain how metadata relates to each other. Trusting AI Agents requires trusting the underlying “black box”. Currently, that trust is impossible for most.

The true cost of poor documentation

It’s not just the cost of creating and maintaining the documentation. The impact is significantly larger than that:

• Kills agility: Every change needs time-consuming analysis to ensure that it won’t impact other functionality and break the org.  That slows down delivery and the ability of the business to stay agile. But has to be done as Salesforce is a strategic application.
• Collaboration and communication: Without a shared view of the state of the org, teams are inefficient or even working at cross purposes. This further delays projects, and the cost is even greater if the teams are expensive consultants. 
• Operational risk: There is a technical cost of making a change that breaks the org; the work to roll back and fix. But the operational impact is many times greater; user downtime, corrupted data, and development effort wasted that could have driven the business forward.
• Increased tech debt: If it is 100% clear what a metadata item does and where it is used e.g. Flow, then it is risky to reuse it. Instead, a new metadata item, potentially duplicating existing ones, is created. This further adds to the org complexity.
• Expensive to maintain: Organizations in highly regulated industries have little choice. They need documentation to support their compliance regimes. This comes at a huge cost in the maintenance of documentation.

On-demand documentation. Changing the documentation game

In a recent article by Matthew Morris (Salesforce CTO & DCX Innovation Director at Capgemini UK & Salesforce MVP Hall Of Fame) highlighted this transformative shift in thinking about documentation. New tooling enables “documentation on demand”, rather than painful manual documentation that is created and gathers dust and gets out of date.

At TrailheadX in San Francisco last month, one of the most oversubscribed sessions came not from a big name, but from Elements.cloud, a small ISV with a tool that struck a nerve. 

Elements describe it as AI-powered, and yes, there’s a natural language interface that’s impressive. But the real breakthrough is the reverse engineering of business processes from Salesforce metadata. This could have been built at any time in the past decade ago using standard APIs.

That’s the part that stole the show.

And the fact that it still feels groundbreaking? That says more about the state of Salesforce tooling than the technology itself.

So why has it taken so long?  Elements.cloud Chief Product Officer, Xavery Lisinski explains:

“It has taken 300 FTE-years of effort to build the application that syncs and analyzes all the metadata. This is foundational for PCM. There are so many metadata types, and more are added every release: Industry Clouds, Data Cloud, and most recently Agentforce. For every metadata type we need to analyze every possible metadata dependency. But not just if it exists, but how it is used. And then to build the process diagrams and data models we need to run a huge sequence of analysis programs. The specification for this was 60 pages, and it was 6 months of development by our most senior development team. This is so much more than a clever AI prompt.“

Configuration Mining: a documentation family

Configuration Mining is a family of documentation that can be generated automatically from metadata:

• Process Configuration Mining: draws UPN process diagrams that describe how Salesforce is configured
• Data Model Configuration Mining: draws ERDs using the Salesforce Diagrams notation
• Order of Execution Configuration Mining: draws UPN diagrams that show the order that automations are triggered

Process Configuration Mining uses our AI model to take a user’s request for a specific view of the system. It then takes the 100,000s of metadata items in the Salesforce org – objects, fields, record types, security permissions, automations, validation rules, metadata dependencies, field analysis, and complexity scoring. It runs a sequence of deep analysis to build the data that our AI model uses to draw a very detailed process diagram and data model, putting all the metadata in context.

This is changing how we think about documentation. Instead of creating documentation and leaving it on the shelf to gather dust, you create it on demand when you need it, from a specific perspective.

“Process Configuration Mining” is different (and more powerful) than Process Mining.

• Process Configuration Mining tells you every possible path that any user could take based on the Salesforce configuration. The ability to quickly visualize the operational, compliance, and adoption impact of changes is huge.
• Process Mining tells you what paths users have been taking through application screens based on tracking users when they are updating fields over a set period of time. It is an incomplete picture which limits it value.

The difference is that Process Configuration Mining is the map of the city with all the dark alleys and cut-throughs. Process Mining shows the most popular routes (where there are sensors) over a set period of time.

How Salesforce ACTUALLY works

Every customer we talk to is struggling to make changes to their Salesforce org at the pace that the business is demanding.  A lack of understanding about the org increases the cost of ownership and delays projects that the business demands to stay agile. There is a huge demand for automated documentation that is more than superficial “AI-generated descriptions”. 

PCM is a current, accurate view of how the org works. 

• Not how you all thought it worked. 
• Not how the documentation (if you have any) claims it works. 
• Not how people remembered it works. 
• Not how the design documentation says it works.

This is HOW IT ACTUALLY WORKS.

Richard Parker, Co-founder and Chief Customer Officer, reflects on discussions with customers over the last 8 years. 

“Ever since we showed people how to map UPN process diagrams in workshops at Dreamforce in 2017, they have been asking, “I assume that these diagrams are drawn automatically?”  That has been our north star. I’m so excited we are now able to say “Yes”

Matthew highlights in his article that agents are suddenly increasing the importance of understanding the org.

“Every Salesforce customer is about to rewire their business processes. Whether your org is 2 years old or 20, it’s full of assumptions that won’t survive the AI era. Before you can transform, you need to understand what you’ve got.“

Here is an excerpt from a recent customer conversation that echoes what we hear from customer after customer.

“People are requesting things for Salesforce. They think they know Salesforce best and they’re they’re coming to us with the solution. We know they are not considering xyz because they (and we) can’t do the impact analysis. And I am about to pull out my hair. This is madness. We’re no longer in the wild west. We are well past drawing Lucidcharts. I cannot think of a better way to do it personally than Process Configuration Mining. They will ask for an enhancement. We will take this autogenerated diagram and pinpoint where they’re asking for changes, and then visually show them during the conversation that this is the impact that it will have, and the other changes we need to make.“

How will you use it

There are some very obvious use cases.  But as we dig deeper into the PCM diagrams, more sophisticated use cases are emerging. And as the diagrams are built from the massive data analysis, it is possible to use AI to drive out even deeper insights or put an agent on top of the detailed analysis so you can simply “ask the diagram”.

And the diagrams are actionable and editable. Change the current boxes and lines. Add sticky notes. Comment and collaborate. Add new boxes, lines and attachments.

Here are some of the use cases, but plenty more will emerge as customers and consultants start using it in anger:

• Highlights potential scope for agents: Once the processes are understood, the diagrams make it very easy to see what could be outsourced to an AI Agent, what permissions it needs, and which actions (Flow, Apex Prompt Template, API) it needs to access. This is the time to optimize the process before you apply an agent to it. The diagrams make it obvious where the simplification opportunities are. 
• Provides data to estimate the risk / effort of any changes: The dependencies between metadata are easily understood. A change to one metadata may require changes to others, which can impact how that metadata affects other areas of the business. This makes it easier to estimate the risk of making the changes and the effort involved. Too often, a simple change has far-reaching, unintended effects because dependencies were overlooked.
• Visibility of data governance/data quality issues: You can see the end-to-end lifecycle for an object and how and where data can be created and updated. This enables you to redesign the security profile and the workflows to provide more robust data governance.
• Uncovers security/compliance vulnerabilities: Each process step shows the teams that have access. Suddenly, you can see the impact of the complete profiles and permissions and how access has inadvertently been given to teams that don’t need it.
• Shows process improvement opportunities: Whether you are implementing agents or not, the diagrams clearly show where there are process opportunities. They also make it easier to estimate and communicate to teams the changes that need to be made.
• Identifies need for critical documentation: The Salesforce org probably has integrations with 3rd party apps. These are often undocumented. Reviewing the diagrams identifies missing documentation that is critically important, and that can now be manually added as metadata dependencies.
• Prioritizes Salesforce optimization activities: Teams often embark on tech debt or data cleanup activities. The diagrams help you understand what will make the most difference. It is frustrating to spend valuable time in areas that are changed very little, are not used, or will be eliminated by process improvement or agents. The diagrams can help prioritize the clean-up activities.

Based on your role…..

CIO / VP / Platform Owner

You can reduce the risk of your org. You want to drive up team productivity and collaboration. You can see where to prioritize your resources to get the biggest bang for your buck. If you are working with an SI, then they no longer need to do the weeks of manual analysis of your org – at your cost. If you are asking SI’s to bid for work, give them access to Process Configuration Mining so they are able to give an accurate proposal.

Here is another customer excerpt:

“What keeps me awake at night as Vice President is that we have so much stuff we’ve built over time on Salesforce, and I’m super worried that the next person who walks out the door is going to leave us even more vulnerable than where we are today. We don’t have an accurate knowledge base. And I don’t want a knowledge base based on what somebody thinks is in their head. I want a knowledge base based on what’s really in the system. How do we figure out the impact of what we’re trying to do in a much more visual manner? I think we’re spending millions of dollars unnecessarily on meetings after meetings because we don’t have a live knowledge base. I want one that stays alive with every single change we implement.”

Architect / Admin / Developer

Your role is to optimize your Salesforce org and plan future changes. The diagrams give you unparalleled insights into the current state of the org so you can make informed decisions. What is unique is that you can ask natural language questions to drive the scope and perspective of the diagram.  You can architect, design, and implement better solutions, more quickly, with confidence. 

Consultant

Consultants are paid vast sums to manually document client orgs. Org discovery is a necessary phase at the beginning of every project. So for them, PCM might be seen as a threat. But no consultant ever had a client who was excited about the fees they paid for documentation, and would write a glowing success story. If the client gives their consultant access to Process Configuration Mining, then the consultant can focus on driving more value for the client for their fees. There is a far larger benefit for the consultant who is forced to bid based on wildly inaccurate assumptions. Suddenly, with PCM, it is easier to scope projects more accurately. No more costly overruns or embarrassing conversations with the client about scope creep. Recommendations can be made more confidently as they are based on accurate data about the org. 

Final Word

I’ll leave the final word to Matthew.

Why This Matters Now.

“Salesforce orgs are complex. Years of configuration, integration, customisation, not all of it well documented. Modernising this isn’t just a clean-up job. It’s essential. And history tells us that the teams who adopt tools early move faster and smarter.

Here are some powerful tools from my past that were used to great effect; Think 4GLs. Think integrated CASE tools. Those didn’t just make development faster—they made it better.

We’re at that moment again.

This is the foundation of modern Salesforce delivery—and the future of how we build. Advanced tooling isn’t just an upgrade. It’s a strategic capability.“

Want to know more or get on the waitlist to trial it?

Photo by Lieu Cap on Unsplash