IBM introduced Concert as an AI-assisted way to understand and improve the increasingly complicated systems behind modern applications. Ben Ball, a senior product marketing director at IBM, described technology teams as overwhelmed by data from numerous tools, environments and stages of an application’s lifecycle. Concert is intended to reconstruct how an application works and recommend improvements to performance, security and management. The product targets a familiar enterprise problem: organisations possess abundant operational information, but the information is fragmented across systems and difficult to turn into coordinated action.
Application Complexity Creates a Data Problem
A single business application may depend on cloud services, databases, code repositories, monitoring platforms and security tools. Each component produces alerts, logs and configuration data, often with its own language and ownership. Teams can spend substantial time assembling context before they investigate the underlying issue. The problem grows as applications change and as responsibilities are divided among development, operations and security specialists.
Concert seeks to apply AI to this scattered evidence and form a more coherent model of the application. If the system can identify relationships among components, it can help teams understand how a vulnerability or performance problem affects the wider environment. That promise is more useful than another isolated dashboard, but it depends on accurate integrations and current data. A stale dependency map could make an apparently confident recommendation misleading.
AI Recommendations Need Explanations
IBM plans to let users question Concert’s recommendations and explore how the system reached a conclusion. Explainability is especially important in technology operations because a suggested change may affect production availability, security or regulatory obligations. Teams cannot responsibly apply a recommendation merely because an AI system presents it fluently. They need supporting evidence, assumptions and a clear indication of uncertainty.
An interactive explanation can also improve collaboration. A security team may want to understand why a particular vulnerability was prioritised, while an application owner may need to see the operational impact of remediation. If both can inspect the same reasoning, the tool can create a shared basis for decisions. Explanations must nevertheless be grounded in actual evidence; a plausible narrative generated after the fact would not provide genuine accountability.
Organising Unstructured Data Is a Core Opportunity
Ball identified the gap between unstructured operational data and information prepared for AI as a major challenge. Enterprises often expect advanced analysis before establishing consistent naming, ownership and data quality. Concert is expected to evolve with tools that organise information into forms AI engines can use. This places data engineering at the centre of the product rather than treating it as an invisible prerequisite.
The quality of recommendations will reflect the quality of that preparation. Duplicate alerts, missing metadata and contradictory system records can all distort results. Organisations adopting an AI operations layer should therefore begin with defined use cases and a realistic inventory of available data. Cleaning every source is unnecessary, but the inputs that drive high-impact decisions need traceability and dependable access controls.
Deliberate Use Cases Matter More Than AI Adoption
IBM advised organisations to start with a real goal instead of treating the technology itself as the objective. In operations, suitable targets might include vulnerability prioritisation, application dependency analysis or reducing the time required to diagnose incidents. Each use case has measurable outcomes and a known group of users. That makes it possible to compare AI-assisted work with an existing process and determine whether the system improves speed or quality.
Human responsibility remains necessary. Recommendations affecting live systems should pass through approvals that reflect their potential impact. Lower-risk observations may be automated sooner, while changes to production require stronger evidence and rollback plans. Concert’s ability to support questions could help teams calibrate that trust instead of choosing between complete automation and complete rejection.
IBM positions Concert alongside its wider watsonx and governance portfolio, suggesting that it wants to connect operational intelligence with broader enterprise AI controls. The product’s value will depend on how well it integrates with the tools customers already use and whether it reduces, rather than adds to, the number of interfaces teams must monitor. Design partners reportedly found useful insights even in early versions, but production environments will test scale and consistency.
The most compelling idea behind Concert is not that AI replaces technology operations. It is that AI can help humans see a system that has become too interconnected for any one dashboard or specialist to understand alone. If IBM can make recommendations transparent, evidence-based and actionable, Concert could turn operational data into a more manageable decision layer. Success will be visible when teams resolve issues faster and with greater confidence, not simply when another AI product appears in their toolchain.
Adoption should begin in observation mode, allowing teams to compare Concert’s application map and recommendations with known incidents before granting it a larger role. This period can reveal missing integrations, duplicated records and areas where the model lacks context. It also gives operators time to learn which explanations are dependable. Once the evidence is strong, organisations can automate low-risk follow-up while preserving approval for changes with significant consequences. A gradual path fits the reality of technology operations, where trust is earned through repeatable performance under pressure. IBM’s challenge is to make that learning process visible and manageable for customers rather than asking them to accept intelligence as a black box.
