Microsoft Copilot in Excel: Faster Fire Engines for Spreadsheet Planning

Excel has always had a strange role in supply chain planning. Officially, most companies run their supply chains through ERP systems, planning platforms, MRP engines, APS tools and reporting environments. Unofficially, many of the real decisions still happen in Excel.

It is where planners correct system output, work around constraints, prioritise exceptions, and turn “what the system says” into “what we are actually going to do.” For years this has been treated as a necessary evil. Everyone knows it happens. Few businesses fully admit how dependent they are on it.

Now Microsoft Copilot is being embedded directly into Excel. That may be useful for productivity. It can help people analyse data, generate formulas, create summaries, filter tables, identify trends, build PivotTables and get answers from workbook data. But in supply chain planning, there is a bigger question: if Excel is already the unofficial planning system, what happens when AI is embedded into it?

The risk is that companies do not eliminate spreadsheet firefighting. They make it faster, easier and more legitimate.

Excel is not outside the planning landscape. The ERP may hold the official transactions. The planning system may produce the official output. The reporting tool may show the official view. But Excel often contains the actual decision logic.

When a planner exports data, changes assumptions, adds formulas, applies judgement, reshapes the answer and sends the result back into the business, Excel is no longer just a productivity tool. It has become a decision engine. It may be unofficial, uncontrolled and invisible to senior management, but operationally it is part of the planning process.

This is why Copilot in Excel needs careful thought. If spreadsheet planning is firefighting, then AI in Excel is not fire prevention. It is a faster fire engine.

That can be useful when the fire is already burning. Better tools, quicker analysis, faster formulas and easier summaries can help planners react more efficiently. But a business should not confuse a faster fire engine with a safer operating model.

The real objective should not be to get better at putting out fires every week. It should be to reduce the number of fires that happen in the first place.

If companies put AI into Excel without changing the planning model, they may create AI-assisted firefighting. The planner moves faster. The spreadsheet becomes more powerful. The analysis looks more intelligent. The workaround becomes more convincing. But the process remains reactive: wait for a problem, export the data, manipulate the answer, apply local judgement, and push the result back into the operating process.

That is not a resilient planning model. There will always be moments when a business needs emergency response. Real disruption will not disappear. But when firefighting becomes the weekly planning model, the issue is no longer speed of response. The issue is system design.

Copilot can make spreadsheet work easier. That is not the problem. The problem is what kind of work it makes easier. If it helps a user summarise a table, build a formula, identify an outlier or explain a data trend, that is useful. If it helps a planner manipulate the unofficial supply plan faster, the implications are different.

A planner could use AI-supported Excel to analyse MRP outputs, build alternative order scenarios, explain stock risks, classify exceptions, identify shortages, model supplier or lead-time changes, create local planning summaries and reshape data for action. Some of that may help. But it also risks reinforcing the behaviour most businesses should be reducing: critical planning decisions being made in local spreadsheets outside governed planning logic.

The management trap is that fast fire engines feel like progress. A business can say its planners can analyse faster, build scenarios faster, summarise exceptions faster and manipulate the plan faster. All of that may be true. But the harder question is: why is this decision still being made in Excel at all?

If the decision affects orders, inventory, suppliers, capacity, service or working capital, it should not depend on a local spreadsheet workaround, however AI-enabled that spreadsheet becomes. The question is not whether Copilot can make Excel planning faster. The question is whether Excel should remain the place where critical supply planning decisions are made.

Excel is not the enemy. It is flexible, familiar and useful. It has earned its place because formal planning systems often fail to reflect operational reality quickly enough. The danger is not that planners use Excel. The danger is that businesses allow Excel to own decisions that should be governed, repeatable, auditable and systemised.

There is an important distinction between AI that explains and AI that decides. Copilot-style tools are useful for explaining data, summarising patterns, helping users ask questions and supporting analysis. That does not automatically make Excel the right environment for supply chain decision automation.

Once the output becomes a decision — what to order, what to change, which supplier to use, what inventory to hold, what capacity to commit — the model needs more than analysis. It needs controlled data inputs, defined planning logic, constraint awareness, auditability, explainability, governance, approval workflows, integration to the system of record, performance measurement, scenario control and clear ownership.

The biggest risk is that businesses mistake AI in Excel for AI transformation. They may say, “We have AI embedded into our planning process.” But what they really mean is: “We have made the spreadsheet workaround smarter.” That is not the same thing.

Before encouraging AI-supported Excel planning, leaders should ask: which planning decisions are currently being made in Excel? Which of those affect inventory, capacity, suppliers, service or working capital? Which spreadsheets contain business-critical planning logic? Who owns that logic? Is it audited, repeatable, integrated and measured against outcomes? Would AI support make this process safer, or simply faster?

These questions matter because AI does not automatically improve a bad process. Sometimes it simply accelerates it.

The better route is not smarter spreadsheet firefighting. It is fire prevention. In supply planning, that means moving repeatable decision logic out of local spreadsheets and into focused, governed AI/ML decision modules.

Instead of using AI to help planners manually rebuild the answer, companies should use AI/ML to systemise the decisions that repeatedly create planning fires: what to order, which orders to change, which exceptions matter, how to respond to lead-time changes, how to model supplier disruption, how to adjust routes, costs or capacity assumptions, and how to protect service without creating unnecessary inventory.

That is a different use of AI. It is not a faster fire engine. It is a better alarm system, better building design and fewer preventable fires.

The real opportunity is not to make Excel the permanent AI planning cockpit. It is to identify the decisions that have fallen into Excel and move them into focused, governed AI/ML modules. Examples include AI-enhanced MRP, mechanised ordering, order change recommendations, production scheduling, transport planning, disruption simulation, supplier switching scenarios, exception prioritisation and inventory positioning.

O8 Organic Planner is built around this distinction. It is not designed to make the spreadsheet workaround more efficient. It is designed to reduce the need for the workaround. Organic Planner uses AI/ML to support supply ordering, replanning and disruption response, helping answer questions such as what to order, which orders to change, and how to respond when demand, supply, lead times, suppliers, routes or capacity assumptions change.

With O8’s visual planning front end, users can manipulate network assumptions, costs, lead times, order patterns, routes, supplier choices and other planning parameters, then re-run Organic Planner against the revised scenario. That is different from building another spreadsheet. The objective is controlled scenario-led decision making, not AI-assisted firefighting.

Microsoft Copilot in Excel will make many Excel users more productive. That is not in doubt. But for supply chain planning, productivity is not the only question. If Excel is already the unofficial planning system, embedding AI into Excel risks legitimising the workaround and giving businesses a reason to avoid confronting the real issue: critical planning decisions still live outside governed planning architecture.

The future of supply chain AI should not be smarter spreadsheets. It should be focused AI/ML decision modules embedded into the planning landscape, with clear logic, governance, auditability and integration. Excel Copilot may help people understand data. But supply chain leaders should be very careful before allowing it to become the next generation of unofficial planning control.

The goal should not be to build faster fire engines for spreadsheet firefighting. The goal should be to design a planning model where fewer fires need to be fought.