Blue Yonder and Manhattan Associates Bring AI to Supply Chain Software Deployment, Squeezing Systems Integrators

Supply chain software deployment is becoming a product feature rather than a separate services purchase. At ICON 2026, Blue Yonder Chief Executive Officer (CEO) Duncan Angove said the company is unafraid to disrupt the US$480 billion systems integration industry. Blue Yonder released a manifesto committing to automating the software implementation lifecycle, with Artificial Intelligence (AI) agents and Forward Deployed Engineers (FDEs) handling the bulk of technical migration work.

Manhattan Associates is attacking the same model from a different angle, altering configuration updates to systems for current users. Its new Solution Design Studio is a natural language AI system built for warehouse and transportation operations staff, rather than architects or developers. Manhattan claims reconfigurations that once took months now take minutes. The tool is now out of testing and in use on targeted implementations run by Manhattan's own services teams, and Manhattan expects to begin rolling it out to customers in the coming quarters.

Other vendors are moving in different directions. Kinaxis is shifting deployment work to independent integrators and expects its own services revenue to moderate in 2H 2026. SAP built Joule for Consultants, an AI assistant for the consultants and partners who implement its software. o9 still delivers mainly through partners such as Accenture, though its CEO has pitched agents that let customers change configuration without waiting on o9 or its partners.

Systems integration has been a costly but necessary line item in supply chain Software-as-a-Service budgets. Platform transformations typically run for months or years, Systems Integrators (SIs) bill by the hour, and the work claims between 30% and 50% of the budget for tier-one projects. In-house, AI-powered deployment and configuration threaten that status quo in at least two ways. They move the previously outsourced work inside the vendor, and they compress the timeline far enough to break the economics of hourly billing.

However, the systems integration industry still controls elements that keep it from obsolescence. Integration with systems outside the vendor, change management, and program governance all involve a human dimension AI cannot handle alone. Negotiating with warehouse staff to work differently and holding someone accountable for fixing errors in AI-produced work requires human oversight from a party other than the seller, and especially one who brings expertise from prior experience to the integration process. Vendor FDEs will carry a loyalty bias toward their employer, weakening their ability to advocate for the client. In multi-vendor estates, buyers may retain an independent integrator specifically to hold the vendor accountable.

Growth in AI-powered deployment also depends on how quickly companies like Blue Yonder can hire and train FDEs. Palantir pioneered the FDE model, embedding engineers directly in customer environments. The role demands stakeholder negotiation across operations, Information Technology (IT), and finance, which typical engineers have never trained for. Meanwhile, skilled and trusted consultant teams, expensive as they are, already work with Tier One customers and can offer a lower-stress, traditional deployment. Individual enterprises will weigh their appetite for risk against the cost savings, but broadly, this dynamic will slow AI’s takeover of the systems integrations market.

Even so, the vendor announcements represent a larger trend focused specifically on Software-as-a-Service (SaaS) deployment. In May 2026, OpenAI, along with Bain & Company, Capgemini, and McKinsey, announced its own deployment company, bringing in about 150 FDEs and deployment specialists. Their motivation for doing so is exactly in line with what Blue Yonder and others are thinking. The money does not reside with selling AI tools themselves, but in the services of integrating and deploying AI.

The lack of in-house technical knowledge that has always justified third-party SIs will not disappear, so demand for their services will persist. The real challenge is the disappearance of billable hours, which will force SIs, such as Accenture, Capgemini, EY, PwC, and TCS, to rework their business models.

To defend their market position, traditional SIs should move the value chain from delivery to governance and oversight; that means AI governance (model risk and explainability), data integrity, and lifecycle management. SIs also hold industry-specific expertise in areas such as U.S. Occupational Safety and Health Administration (OSHA) rules, food safety, and customs compliance. Pricing services on outcomes rather than hours will let them keep the margin AI tooling creates, even as typical project hours diminish.

To remain competitive, supply chain software vendors should use Agentic AI to refine deployment and configuration, but they should be cautious about betting that customers will configure systems themselves. End users will not necessarily want to be responsible for building their own AI tools. Non-engineers, including fleet managers, procurement leads, and warehouse floor operators, do not have time in their day-to-day work to vibe-code software solutions. End users can now build, but can we be confident they will?

More likely, clients that outsource deployment and configuration today will continue to do so. Solution providers' biggest win will come from native industry data, not from crowding out the consultants and developers their clients have come to rely on.