Most ERP systems do exactly what they were designed to do. They capture transactions, enforce structure, and serve as a reliable system of record.
Unfortunately, that’s where the support usually stops.
The actual work behind those transactions—drafting, collaborating, reviewing, handing off—still happens somewhere else. And users are left bouncing between tools, chasing approvals, and navigating systems that expect them to come find the work instead of bringing the work to them.
This article explores why that gap still exists and how forward-thinking organizations are shifting from reactive ERP experiences to proactive ones where work shows up with context, in the flow of how people actually work.
Why ERP Still Feels Like Work — And Not in a Good Way
You’ve probably seen this scenario play out before. An approval is sitting there waiting in the system, but no one is quite sure where. So, a follow-up email gets sent, someone else checks a dashboard, and a spreadsheet gets updated “just in case.” Before long, there are multiple parallel threads trying to move the same task forward, and by the time the transaction is finally entered into ERP, the real work has already happened elsewhere.
That’s the disconnect.
ERPs are exceptionally good at recording what happened, but they provide far less support for the work required to make it happen in the first place. Creating, coordinating, reviewing, and passing work between teams tend to live outside the system, which forces users to operate in multiple places and then return to the ERP system to record the outcome.
Built for a Different Era
ERP systems still operate on an outdated assumption that users will come to them, logging in to check worklists, monitor dashboards, and go looking for tasks that need attention. That model made sense in a more centralized, system-driven world, but it no longer reflects how work actually gets done in the modern workplace.
Figure 1: The Perils of Being Tethered to ERP Worklist Transactions
Today, work flows through conversations, shared documents, and collaboration tools like Teams and email. Work moves across roles and systems, often asynchronously, and increasingly with AI agents helping to surface what matters at the right moment. Compared to all that, ERP user experiences can feel static and reactive, a place you go to update the system rather than a system that actively helps move work forward.
That tension is becoming harder to ignore. To some extent, ERP vendors have tried to paper over these deficiencies by modernizing user experiences—replacing legacy desktop clients with responsive web apps, refreshing interfaces, and improving accessibility. Those changes matter, but they largely address how ERP looks and feels, not how work actually flows.
Figure 2: Evolution of User Interfaces in ERP Systems
At its core, the model remains the same: users are still expected to come to the system, navigate to the right place, and take action there.
Built for Transactions, Not Processes
ERP systems were designed to be systems of record, and they excel at recording transactions. These transactions provide structure, enforce controls, and create a reliable audit trail of what happened.
What they don’t do nearly as well is support the work required to get there.
The reality is that most business processes don’t unfold neatly inside a single system. They involve coordination across teams, informal checkpoints, and iterative steps that rarely fit into a predefined transaction. As a result, much of the actual work ends up happening outside the system, whether it’s aligning on details, resolving issues, or preparing inputs in email threads, spreadsheets, and side conversations.
By the time a transaction is entered into the system, the process is already complete. The ERP captures the outcome, but the system remains largely disconnected from the execution that made it possible.
That gap between system of record and system of execution is where friction builds, and where the biggest opportunity for improvement still exists.
Disconnected Workflows
Most ERP platforms have built-in workflow capabilities, but these features often come with a catch: the work lives inside proprietary inboxes that are disconnected from where users actually spend their time.
Instead of surfacing tasks in familiar tools like Teams or Outlook, most systems require users to drop what they're doing, log in to a separate system, navigate to a specific worklist, and manually check for updates. If they don’t, approvals stall, tasks sit untouched, and someone eventually has to follow up outside the system to keep things moving.
Figure 3: Working with SAP's Proprietary Workflow Inbox
The result is a workflow model that depends on users going to find the work rather than the system bringing work to them. In a world where most collaboration happens in shared conversations and real-time tools, that disconnect creates unnecessary friction and slows down even the simplest processes.
From Systems of Record to Systems of Action
If ERP systems are where work gets recorded, then the obvious question is: where should work actually happen?
The answer isn't to force process flows back into the ERP system. It's to introduce a complementary layer around the ERP system that supports process execution while letting the ERP do what it does best: serve as the system of record.
This new collaboration layer is made up of extension apps, intelligent data platforms, and increasingly AI-powered agents that sit alongside your ERP system and bring work to users in context. Instead of requiring people to log in, navigate, and search for what needs attention, this layer pushes the right work to the right person at the right time, within the tools they already use.
That shift may sound subtle, but it changes everything.
In this scenario, work no longer starts in your ERP system. Instead, it flows through collaboration tools, mobile devices, and lightweight applications that are designed for how people actually operate. Your ERP system becomes the place where outcomes are finalized, not the place where every step of the process has to occur.
None of this replaces ERP. It extends it.
In the next section, we'll take a closer look at what this responsive operating model looks like.
What “Proactive ERP” Actually Looks Like
Proactive ERP isn’t about a single product, tool, or feature. It’s about reshaping how work flows across your environment so that people can take action without constantly stepping out of their natural workflow.
In this model, the experience shifts from one where users are responsible for polling systems and chasing tasks to one where the system actively surfaces what needs attention, complete with the context required to act on it. Instead of navigating menus, dashboards, and inboxes, users receive work where they already operate, whether that’s in collaboration tools, email, or on mobile devices.
The result is a more responsive, connected experience where work moves forward with less friction, fewer handoffs, and far less guesswork about what needs to happen next.
Work Comes to the User
In a proactive ERP model, one of the most noticeable shifts is where work shows up.
Instead of forcing users to dig around in an ERP system to find tasks to work on, tasks, alerts, and approvals are pushed directly into the tools they already use every day. This includes collaboration tools such as Microsoft Teams, email, push notifications on their mobile devices, and increasingly AI-powered agents. In this scenario, the system meets users where they are, rather than expecting them to go looking for what needs attention.
To put this concept into perspective, consider the approval request shown in Figure 4 below. Here, a credit limit request has been routed from the backend ERP system to multiple approvers in their Outlook inbox. While the approval form itself is straightforward enough, there are a couple of notable things worth mentioning about it:
The approval form is rendered using Microsoft's adaptive cards technology, which means that it's a "live form" that acts like a little miniature app that can run in Outlook, Teams, Copilot agents, etc.
The form has a live connection to the backend ERP system, enabling the approvers to process their approvals without ever having to leave their Outlook inbox.
Figure 4: A Live Approval Request in Outlook
Of course, these interactions aren’t limited to simple notifications with a link back to ERP. They include the context required to actually take action — possibly in multiple systems. From the user's perspective though, it's all one single pane of glass to see what the task is, why it matters, and what decision(s) needs to be made without having to open another application and piece things together.
This may seem like a small change, but it fundamentally alters how work moves. When tasks are visible, actionable, and delivered in context, processes accelerate, bottlenecks are easier to identify, and far less time is spent chasing down information or waiting on the next step.
Process Lives Outside the Core (But Stays Connected)
During the early stages of work, users are in information gathering mode: coordinating across teams, refining inputs, and passing the baton back and forth between relevant process contributors. In these early stages, it's much easier for users to work in modern tools designed for collaboration and iteration. These flexible environments make it easier to move quickly, loop in the right people, and flexibly adapt as details evolve.
In this scenario, the ERP system still plays a critical role, but it enters the picture at the right moment, not the first moment.
Instead of forcing users to shoehorn half-baked transactions into the system too early, it's preferable to delay posting until key decisions are finalized and the data is ready to be recorded. This keeps the system clean, reduces rework, and ensures that what gets captured in the ERP system reflects the outcome of the process, not the noise along the way.
The key to making all this work is introducing a shared workspace where users can capture information in "draft transactions" that guide them through the process step by step. Instead of forcing everything to be completed and validated upfront, this approach allows key validations and gate checks to occur at appropriate points in the process flow, rather than relying on the all-or-nothing model that most classic transactions enforce.
Figure 5 illustrates these concepts by showing how Dynamics 365 Sales integrates with Microsoft Teams to bring sales work into a shared, collaborative space. By linking a sales opportunity to a Teams channel, the entire deal team can work together in one place—combining chat, audio and video meetings, shared documents, activity management, and CRM updates into a single experience. Instead of jumping between systems, sales team members can collaborate, progress the opportunity, and record key transactions without ever leaving the flow of their work.
Figure 5: Collaborative Authoring with Teams and Dynamics 365
The key is making sure work doesn’t become disconnected. With thoughtful design, it stays linked to the system through integrations, shared context, and structured handoffs, so even when the process happens outside the core, it remains fully connected to it.
AI as a Coordinator, Not Just a Chatbot
Agentic AI is taking these concepts to the next level.
While earlier AI experiences focused on chat—answering questions, generating content, or retrieving information—the emerging model is far more proactive. In a proactive ERP environment, AI doesn’t just respond to requests; it helps coordinate how work moves across the organization.
In this model, system events become triggers. A delayed order, an exception in a transaction, or a threshold being crossed doesn’t just get recorded—it initiates action. Autonomous agents can interpret these events, determine what needs to happen next, and reach out to the right stakeholders at the right time using conversational interfaces.
Instead of users checking systems for updates, the system engages them directly—asking for input, providing context, and guiding the next step in the process.
Figure 6 illustrates this “weave” between ERP systems, autonomous agents, and conversational agents. In this scenario, the ERP system remains the source of truth, autonomous agents handle orchestration and decisioning, and conversational agents serve as the interface through which users engage. Together, they form a connected loop where work is continuously monitored, surfaced, and advanced without requiring constant manual oversight.
Figure 6: Orchestrating Workflow Using AI-Powered Agents
How to Get There Without Ripping Everything Out
For many organizations we work with, this is where hesitation sometimes sets in.
There’s a natural concern that introducing extension apps, collaboration layers, or AI-driven experiences around ERP might violate clean core principles passed down like scripture from ERP vendors. After years of being told to reduce customization and avoid reinventing the wheel with transaction screens, anything that feels like “building around it” can raise red flags.
In reality, the opposite is true.
The approach we've outlined here is directly aligned with how modern ERP platforms are designed to evolve. Leading vendors are actively encouraging side-by-side extensions, API-driven integrations, and externalized workflows precisely so the core system can remain stable while innovation happens at the edges.
Rather than modifying ERP itself, the goal is to extend the reach of it, thereby increasing its value. By moving process execution, user experience, and orchestration into a complementary layer, you reduce pressure on the core, avoid unnecessary customization, and create a more flexible foundation for change.
Start with High-Friction Processes
As is the case with any enhancement initiative, it’s important to determine if the juice is worth the squeeze. Not every process is worth optimizing, especially those that are rarely performed or have minimal impact on day-to-day operations. While it may be tempting to clean up edge cases or infrequent workflows, the real value comes from focusing on processes that are executed often, involve multiple stakeholders, or consistently create delays.
The best place to start is with processes where the friction is already obvious. For example, anywhere work gets stuck or leaves the system. Approvals that require multiple follow-ups, order management steps that rely on email coordination, and service workflows that live partly in ERP and partly in spreadsheets are all strong candidates.
These are the areas where the gap between system of record and system of action is most palpable. Work leaves the system because it has to, not because anyone designed it that way, and once it does, it becomes harder to track, slower to move, and more dependent on manual effort to keep things progressing.
By focusing on these high-friction processes first, you can introduce targeted improvements that bring work forward—surfacing tasks, simplifying interactions, and reconnecting execution back to the system. The payoff is typically immediate, which helps build momentum without taking on unnecessary risk.
Extend, Don't Replace
Most ERP modernization projects fail before they start because they're solving the wrong problem.
The assumption is that the system is the constraint. That the only way to move faster, serve users better, or integrate new tools is to rip out what's there and replace it. So, companies spend years and millions on a rip-and-replace upgrade project, only to discover that the new system has many of the same limitations.
The constraint was never the ERP system itself; it's the architecture around it.As noted earlier, the better approach is to leave the core alone. Let ERP do the one thing it's actually good at—maintaining an authoritative record of transactions, inventory, financials, and master data. Then build around it.
Side-by-side extensions sit alongside the core rather than inside it, handling the kind of unstructured process flows the ERP was never designed to handle well. These aren't bolt-ons. They're purpose-built, loosely coupled extensions that communicate with the ERP system through APIs and events.
A warehouse team gets an interface designed for a warehouse. A field technician gets something that works on a phone, offline, in a facility with no patience for slow transactions. A procurement analyst gets a workflow that pulls from ERP without living inside it. Each application can be built, tested, and changed independently—without touching the core, without adding upgrade risk, without making the next S/4 migration harder than it already is.
This is what "clean core" actually means in practice. Not a constraint. A discipline. You're not prohibited from customizing—you're protected from the kind of customization that turns a system of record into a system of debt.Surround the core with a layer that moves fast. Keep the core itself boring, stable, and invisible. That's the architecture that lets you innovate without starting over.
Layer in Data and Context
As you extend your ERP and begin to shift toward a system of action, the next critical layer is data. Without shared context, even the best-designed workflows and experiences fall short because users are still forced to piece together information from multiple sources before they can act.
Modern data platforms like Microsoft Fabric help solve this by bringing relevant data together under one roof. With a unified foundation like OneLake, organizations can connect ERP data with CRM, operational systems, and external sources to create a more complete picture of what’s happening across the business. Instead of chasing data, users and applications can operate from a shared, consistent context.
That unified context becomes even more powerful when paired with capabilities like Real-Time Intelligence. Rather than simply reporting on what has already happened, these tools introduce AI-powered rules engines that can monitor events, detect patterns, and trigger actions as conditions change. This is where data starts to move from passive insight to active participation in the process.
Figure 7: Building Proactive Workflows Using Fabric Real-Time Intelligence
When you combine unified data with real-time awareness, decision-making becomes faster and more informed, and automation becomes far more effective. Work doesn’t just move based on static workflows; it adapts based on what’s actually happening in the business, enabling a more responsive and intelligent operating model.
Introduce AI Where it Actually Helps
It’s easy to get caught up in the bigger vision of AI and agents, but the most effective way to start is much simpler.
As we discussed in our article on building practical business cases for AI automation, the most successful initiatives focus on practical, high-impact use cases that deliver measurable value quickly. Things like summarizing information, routing work to the right person, flagging exceptions, or recommending next steps may not sound revolutionary, but they can dramatically reduce the manual effort required to keep processes moving.
These are the moments where teams tend to spend time on low-value tasks: chasing updates, gathering context, or figuring out what to do next. Introducing AI in these areas frees up capacity and allows people to focus on more meaningful, higher-value work.
Starting small also helps build momentum. By proving value early, you create a foundation for broader adoption and more advanced, agent-driven scenarios over time—without taking on unnecessary risk upfront.
Closing Thoughts
For most organizations, the challenge isn’t that ERP lacks capability. It’s that too much time is spent chasing work instead of moving it forward. Approvals stall, follow-ups pile up, and critical context lives in inboxes or in people’s heads, making progress slower and more dependent on individual effort than it should be.
A proactive ERP model changes that dynamic. By bringing work to users in context and supporting execution outside the core system, organizations can reduce cycle times, minimize manual follow-up, and improve user adoption without requiring everyone to become a system expert. Processes become more transparent, less reliant on tribal knowledge, and more consistent in how they operate.
This approach also creates a strong foundation for scalable AI. When work is structured, connected, and context-rich, it becomes far easier to introduce automation and agent-driven coordination that actually delivers value.
Ultimately, this is how you unlock capacity without increasing headcount. The goal isn’t to get better at working inside your ERP. It’s to make your ERP work wherever your people already are.
