Conversations with Your ERP System

What would you ask if you could have a conversation with your entire business, and receive the answers before the meeting ended?

“How profitable were our service contracts last quarter once we include the time spent fixing problems after delivery?”

It sounds straightforward. In practice, answering it may require accounting records, contracts, timesheets, project data and support tickets. Someone must gather and reconcile the information, allocate the costs and prepare a report.

By the time it arrives, the original question has usually produced several more. Which contracts caused the problem? Was it poor pricing, underestimated work or excessive support? Is the pattern confined to one service or appearing elsewhere?

Some answers may be in the report. Others require another analysis and another wait.

This is how management reporting has always worked. It is functional, but frustratingly slow: ask, wait, review and ask again. Dashboards help, but they can answer only the questions someone anticipated when building them.

Management teams have accepted this because there was no practical alternative. They fill the gaps with experience, make assumptions and hedge when the evidence is incomplete. Sometimes that is enough. Sometimes a competitor moves first, an emerging risk is recognised too late or a valuable contract is lost before management has established what changed.

The information may already exist inside the business, or in the events and trends developing outside it. The limitation has been the time required to bring it together, test what it means and turn it into a decision while there is still time to act.

Strategy at the speed of conversation

Management traditionally has to decide in advance which reports it will need, how frequently they should be produced and how much work is reasonable to invest in them. Some are prepared annually, others monthly or quarterly. They are usually used to check whether assumptions were correct and decisions were effective, not to explore strategy while it is being formed.

With an AI agent connected to the company’s systems and the internet, that changes.

The company’s accounting, sales, project, HR, inventory, marketing and service data provides a concrete picture of the business. At the same time, the agent can research industry trends, competitor activity, acquisitions, regulatory changes, emerging technologies and events that may affect the market.

Management can explore practical decisions against the company’s actual position:

“What happens to margin if we increase prices by 5%?”

“If we discontinue our slowest-moving products, how much cash would we release?”

“What happens if we add two people to delivery rather than sales?”

“Which customers would be affected if we changed our service model?”

“If conversion improves but payment times remain unchanged, what happens to cash flow?”

“Which combination of changes produces the best result without increasing our working-capital requirements?”

The same conversation can respond to changes outside the business.

“A competitor has acquired a smaller service provider. Which of our customers are most likely to be affected?”

“If they reduce prices by 10%, where are we vulnerable?”

“Which customers value service more than price?”

“What would happen to margin and cash flow if we matched the price only for vulnerable accounts?”

“Now model the effect of increasing our support capacity instead.”

The agent can test each scenario against the company’s costs, customers, capacity and financial position, then present the results as spoken answers, tables, graphs, forecasts or spreadsheets.

Assumptions can be tested as quickly as they arise. Assertions can be challenged against company records and developments in the market. Adjustments can be simulated before they are implemented.

A few late-night conversations with the business could produce a strategy containing proactive moves, defensive responses and contingencies for several possible outcomes. Work that once required months of reporting, research and repeated meetings can begin to take shape in hours, grounded in what is happening inside the company and in the world around it.

Making the conversation possible

Taking advantage of this requires more than giving an AI agent a company login. It needs controlled direct access to the relevant systems, an understanding of how the company’s information fits together and clearly defined limits on what it may see and do.

Crucially, the company’s existing access controls must be preserved. Someone who cannot view payroll, employee records or detailed financial information through the ERP should not be able to retrieve it by asking the agent. Access must continue to reflect each person’s role, with sensitive actions restricted, approved or recorded where appropriate.

A financial director might explore detailed costs, margins and cash flow. A sales director may work with customers, opportunities and forecasts. Senior management can bring those views together without making confidential information available throughout the organisation.

The agent also needs business context. It must understand that time recorded after a project closes may represent rework, that a signed sale is not yet cash, and that revenue without delivery cost does not reveal profitability. Those definitions turn access to data into useful management understanding.

Once connected, questions raised in meetings can be investigated immediately. Strategic assumptions can be revisited as conditions change. Monthly reviews become active discussions with the underlying information rather than presentations prepared weeks earlier.

Hatton Locks creates these controlled connections between AI agents and the business systems we implement, while preserving the permissions and boundaries the business depends on.

Management can then explore, test, simulate and decide while the opportunity, or the threat, is still in front of it.