For the better part of two decades, ERP systems have promised one thing above all else: control. A single source of truth for your finances, your inventory, your customer records, and your operations. For organizations across Kenya and East Africa, adopting a system like Microsoft Dynamics 365 Business Central has always meant moving away from spreadsheets, disconnected tools, and manual reconciliation, toward one platform that tells you, accurately, what is happening in your business right now.

That promise has not changed. But what is changing, quickly, is what the system is able to do with that information once it has it. For most of ERP's history, "automation" meant rules. If a purchase order exceeds a certain amount, route it for approval. If stock falls below a threshold, generate a reorder. If an invoice matches a purchase order and a goods receipt, post it automatically. These rules-based workflows removed a huge amount of manual effort, but they were static. They could only do exactly what they were configured to do, and nothing more. A human still had to notice the exception, interpret it, and decide what to do next.

Microsoft's 2026 update cycle for Dynamics 365 marks a genuine shift away from that model. Rather than simply adding more automation rules, Microsoft is introducing AI agents directly into the ERP environment, agents that can interpret a goal described in plain language and carry out the multi-step work needed to achieve it. This is not a chatbot bolted onto the side of the system, and it is not another dashboard. It is a shift in what the software itself is capable of initiating.

From Rules to Reasoning

The easiest way to understand the difference between traditional ERP automation and agentic ERP is to think about what each one needs from a human being.

A workflow rule needs a human to define every condition in advance. A purchase approval workflow, for instance, only works because someone sat down and mapped out every scenario: this amount goes to this approver, this category needs two sign-offs, this vendor is pre-approved up to a certain limit. The system executes faithfully, but it cannot handle anything outside those predefined paths. The moment a situation falls outside the rule, it stops and waits for a person.

An agent works differently. Instead of being told exactly what steps to take, it is given an objective, expressed in ordinary business language, and a defined scope of authority. It then plans and executes the steps required to reach that objective, checking data, cross-referencing records, drafting communications, and flagging anything that falls outside its permitted boundaries for a human to review. Where a workflow rule can only follow a path someone already drew, an agent can construct a reasonable path toward a goal it has not been explicitly walked through.

This matters enormously for organizations that do not have large IT departments to maintain complex custom automation. A SACCO finance officer, for example, does not need to know how to configure conditional logic to get help reconciling a batch of member contributions. They can describe what they need, in the way they would explain it to a colleague, and the system does the interpretive work that used to require a developer.

What This Looks Like in Practice

Microsoft has already begun rolling out a set of purpose-built agents across the Dynamics 365 suite, and the direction of travel is instructive even for organizations that will adopt these capabilities gradually.

A finance-focused agent, for instance, is designed to sit inside the day-to-day work of a finance team rather than in a separate reporting tool. Instead of a finance officer manually pulling data to reconcile accounts or investigate a variance, the agent can surface the relevant transactions, propose an explanation for a discrepancy, and prepare draft communications to stakeholders, all within the tools the finance team already uses, including Excel and Outlook. The finance officer's role shifts from data-gathering to judgment: reviewing what the agent has prepared, correcting it where needed, and deciding what action to take.

On the commercial side, a sales-focused agent can help a small sales team punch above its weight. Rather than a sales representative manually researching a prospect, drafting an outreach message, and updating the CRM afterwards, the agent can handle the groundwork, enrich records with relevant information, and recommend the highest-impact next action, leaving the salesperson to focus on the actual conversation and relationship.

For organizations running Business Central specifically, Microsoft's roadmap goes further still. Beyond the built-in agents, businesses will be able to design their own custom agents for scenarios specific to their operations, defining what the agent should do in natural language rather than code, testing it safely, and then granting it a defined set of permissions before it goes live. For a distribution business, that might mean an agent that monitors stock levels across multiple locations and automatically prepares purchase orders for review. For an NGO managing donor-funded projects, it might mean an agent that checks project expenditure against budget lines and drafts the variance notes a donor report requires.

Why Governance Matters More, Not Less

It would be a mistake to read any of this as ERP systems that now run themselves unsupervised. If anything, the introduction of agents raises the importance of governance and oversight, not lowers it.

ERP data is financially consequential. A workflow rule that misfires produces a predictable, bounded error, because it can only ever do what it was configured to do. An agent that is poorly scoped or inadequately supervised can, in principle, take a wider range of actions, which is precisely why Microsoft has built permissioning, execution logs, and approval checkpoints into the architecture rather than treating them as an afterthought. Every action an agent takes can be traced back to a defined scope of authority, and organizations are expected to define clearly where a human must remain in the loop: large financial commitments, adjustments above a set threshold, or any communication touching a sensitive matter.

For regulated or donor-funded organizations in particular, SACCOs answering to the Sacco Societies Regulatory Authority, NGOs reporting to international donors, academic institutions managing public and private funding side by side, this is the detail that matters most. The value of an agent is not that it removes the need for internal controls. It is that it can do the preparatory work faster and more consistently, while the accountability structure that already governs financial decision-making stays firmly in human hands.

What Kenyan and East African Organizations Should Take From This

For many businesses across the region, the immediate, practical question is not whether to adopt AI agents in Dynamics 365 tomorrow. Most organizations are still in the process of consolidating their core financial and operational processes onto a single ERP platform, and that foundational work remains the right place to start. But there are three things worth doing now, ahead of that curve.

The first is to treat this as a reason to get the fundamentals right. Agents built into Business Central are only as good as the data and processes they operate on. An organization with clean chart of accounts structures, well-defined approval hierarchies, and consistent data entry practices will get far more value from agentic capabilities, sooner, than one still wrestling with inconsistent records.

The second is to start thinking now about where human judgment is genuinely irreplaceable in your organization, and where it is simply filling a gap that better tooling could close. Most finance and operations teams in the region are stretched thin, doing reconciliation, reporting, and follow-up work that adds little strategic value but consumes a disproportionate amount of time. That is exactly the category of work these agents are designed to absorb.

The third is to choose an implementation partner who understands both the software and the governance context you operate in. Configuring an agent safely, with the right permissions and the right checkpoints, requires the same discipline as configuring any other financially consequential system. It is not a feature you switch on; it is a capability you design.

Preparing Your Organization for What Comes Next

Getting ahead of this shift does not require a large budget or a dramatic overhaul of how your organization works today. It requires a handful of deliberate steps taken early, so that when agentic capabilities do become part of your everyday ERP experience, your organization is ready to use them well rather than scrambling to catch up.

Start by auditing where your current processes still rely on manual, repetitive work: end-of-month reconciliations, donor or board reporting, vendor follow-ups, and stock checks are common candidates. These are the tasks most likely to benefit first once agentic tools become part of your day-to-day system, and mapping them now gives you a clear picture of where the returns will be greatest.

Next, invest in the people who will supervise these agents, not just the technology itself. The skill that matters most in an agentic ERP environment is not configuration, it is judgment: knowing when an agent's recommendation looks right, when it needs a second look, and when it should be overridden entirely. Finance officers, procurement leads, and department heads across SACCOs, NGOs, and academic institutions will increasingly need to develop this reviewing instinct, and it is far easier to build that habit gradually than to introduce it all at once alongside a major system change.

Finally, resist the temptation to wait for a finished version of this technology before engaging with it. Microsoft is rolling these capabilities out in stages, through regular release waves, refining governance and permissioning as adoption grows. Organizations that start with small, low-risk use cases, a single reconciliation task, a single reporting workflow, will be far better positioned to expand thoughtfully than those that wait and then attempt to adopt everything at once.

The Shift Is Structural, Not Cosmetic

It is tempting to treat every new AI feature as marketing dressing on top of software that works the same way it always has. That would be a mistake here. The introduction of agents into Dynamics 365 changes the operating model of ERP itself, from a system that faithfully records what happened, to one that can, within clearly defined boundaries, help make things happen.

For organizations across Kenya and East Africa still working to get the basics of digital transformation right, that shift is not an immediate priority. But it is a strong signal of where the platforms they are investing in are headed, and a good reason to build ERP foundations today with an eye on the capabilities that will sit on top of them tomorrow.

If your organization is planning an ERP implementation, or reviewing an existing Dynamics 365 or Business Central setup, it is worth having that conversation now, while these capabilities are still emerging, rather than later, once they have become the standard everyone is trying to catch up to.

                                                                                                 Contact US

Office Number 718, 7th Floor, KU PLAZA, Haile Sallassie Avenue, Nairobi CBD

Postal Address: P.O Box 25063-00100, Nairobi, Kenya

Phone: +254743313103

E-mail: Info@fanisitech.com

                                                                                      Website: www.fanisitech.com