Connecting decisions from quote to cash is the next manufacturing challenge

Manufacturers have spent decades digitizing individual parts of their businesses. ERP manages transactions, PLM supports engineering, MES provides visibility into production and CRM connects commercial teams with customers. Yet the process of turning a customer request into a deliverable order can still expose how difficult it is to coordinate decisions across those systems when engineering requirements, inventory, supplier availability and production capacity all must support the same commercial commitment.

For Rama Kolukula, Solution Head, Topaz AI Next Platform, at EdgeVerve, a wholly owned product subsidiary of Infosys, working directly with industrial customers, the problem is particularly visible in engineer-to-order environments. The individual systems may be capable, but the information required to make a reliable customer promise often remains spread across different functions and applications, forcing people to coordinate the process manually.

“The biggest gap we are seeing in ETO environments is around the promise date and the commitments being made to the customer,” Kolukula says. “A manufacturer needs to understand inventory, supplier availability, capacity and other constraints quickly enough to close the deal, but the forecasts are not always sufficiently reliable. What customers are looking for is greater foresight when that commitment is made, but also an ongoing radar that tells them whether everything is still on track or whether something is likely to cause a delay.”

The challenge is therefore not simply creating a quote faster. A quotation in complex manufacturing can represent a commitment involving engineering feasibility, material availability, sourcing, production capacity and delivery. If those decisions are made independently, problems that remain invisible at the quotation stage can surface later as order changes, production disruption, delayed fulfillment or a revised customer promise.

The gaps between digital functions

The complexity increases quickly in configure-to-order and engineer-to-order businesses. Rama points to work with a manufacturer supplying systems for data center applications, where a customer requirement can expand into a multilayered bill of materials containing hundreds of related items. An apparently familiar configuration may contain a variation that requires additional engineering or a component that has not been accounted for in the original estimate.

This makes quoting a cross-functional operational process rather than a task owned solely by the commercial organization. Engineering may need to establish whether elements have already been designed or require modification. Planning must assess capacity, while procurement considers supplier lead times and inventory teams determine whether the expected materials will still be available when production begins.

“The stakeholders all have to come together before the business can make a commitment on timeline, engineering complexity and therefore cost,” Kolukula says. “The problem is that the systems do not necessarily facilitate that coordination. Everybody can do their part in isolation. Planning looks at forecasting and inventory, buyers look at the supply base and supplier lead times, but nobody necessarily connects all of it to the degree required for a complex quotation. The work we are doing is intended to provide that coordination, breaking the request into its constituent parts, identifying what has already been designed and what needs engineering input, and making sure those handoffs are aligned before the quote is submitted.”

The consequences of getting that coordination wrong extend beyond the time required to generate the initial quotation. A manufacturer may promise a delivery date based on an inventory position that subsequently changes, while materials expected to be available for one order are consumed by another. Supplier performance or a production constraint can alter the picture again.

Kolukula identifies customer trust as one consequence when commitments repeatedly have to be revised, but there are also implications for the manufacturer’s ability to use its capacity effectively and accept additional business. If planners do not have sufficient confidence in existing commitments, production capacity can become harder to optimize even while the organization remains reluctant to add more work to the pipeline.

Infosys’ broader order-to-cash proposition reflects those dependencies. Topaz AI Next is designed to support activities including order capture, engineering configuration validation, quote management, fulfillment planning, order changes, receivables, disputes and returns, with specialized agents working across ERP, CRM, PLM, MES and financial systems. Infosys links the opportunity to outcomes including faster quote-to-order cycles, improved order accuracy, stronger OTIF performance and reduced revenue leakage.

Connectivity has not eliminated coordination

Manufacturers have spent years integrating enterprise applications, but the existence of an interface does not mean an end-to-end process has been orchestrated. A system can transfer data successfully into another application without understanding the business context in which that information needs to be used.

Kolukula sees several persistent gaps. Workflows can still break when activity passes between applications, quotation and order intake frequently involves manual intervention, and the information manufacturers receive from customers and suppliers is not always structured in a form conventional automation can readily process.

A customer may submit an engineering drawing, specification or email that requires interpretation before anyone knows which product, configuration or process should follow. Even where EDI and other integrations are established, substantial amounts of communication can remain outside tightly structured system-to-system transactions.

That is where Infosys sees a distinction emerging between conventional integration and Agentic AI. Traditional automation can move information or execute a predetermined workflow effectively when the inputs and next steps are known. The limitation becomes more apparent when a system has to interpret a customer request, understand how information from another application changes the situation or determine what action is appropriate when conditions deviate from plan.

For Kolukula, the potential role of agents is therefore less about replacing every existing automation and more about coordinating the decisions between them. “Agentic AI can interpret the signals coming from different places and effectively act as a coordinator across the process,” he says. “It can understand communications, specifications and dependencies, look at what forecasting or planning systems are telling it, and keep those elements aligned across many orders that may all be at different stages. Traditional automation remains important, but it reaches a limit when information has to be interpreted and the next action depends on understanding what that information means.”

Keeping the customer promise on track

The coordination challenge continues after the quotation becomes an order. Planning assumptions that were valid when the commitment was made can change weeks or months before production begins, creating a need to identify emerging problems early enough to do something about them.

Infosys is applying an agentic approach to monitoring signals from areas such as supply, inventory and capacity. Rather than relying solely on employees to carry out checks at predefined milestones, agents can monitor whether the assumptions supporting an order are beginning to diverge from reality.

If inventory is falling below the expected level or a supplier appears likely to miss a commitment, the system can identify the developing issue and initiate the appropriate response. Depending on the process and level of autonomy permitted, that could mean preparing a sourcing action, raising a change request or bringing the issue to a person before the disruption reaches production.

The technology is not intended to remove human judgment from every decision. Rama says manufacturers remain cautious about delegating activities that can have significant financial, contractual or third-party consequences. Pricing decisions, commercial exceptions and other actions with material consequences continue to require human involvement, even as organizations seek to automate more of the analysis and preparation surrounding them.

That balance points toward a different operating model rather than a fully autonomous factory. Employees can increasingly manage exceptions while technology handles more of the interpretation, monitoring and coordination that previously consumed their time.

For established manufacturers, this also avoids assuming that another wholesale platform replacement is necessary. Infosys positions Agentic AI as an orchestration layer operating above existing enterprise applications, allowing ERP, PLM, MES and other platforms to remain systems of record while agents help connect activities across them.

The next phase of manufacturing digitalization may therefore be less about adding another system and more about making existing systems contribute to the same decision. Across quote-to-cash, the objective is straightforward even if the execution is complex: ensure that the promise made to the customer reflects what engineering, supply and production can deliver, then keep that promise aligned as conditions change.

For manufacturers that have already digitized most of their individual functions, connecting those decisions may now be where the greater opportunity lies.