Octave is taking AI-driven innovation seriously. That much was clear from its OctaveLive keynote this summer in Austin, Texas, where the focus remained on amplifying human capability, as well as hiring more people, not fewer, to maintain quality.

Vick Vaishnavi, the head of Octave Reliance, took time to answer questions about the release of the new version of the Octave Reliance cloud-native quality management system, which features key AI capabilities like complaints management and a natural language smart search.

Megan Wallin-Kerth: Customer feedback was said to have shaped this release. What problems were quality teams struggling with most, and how did that influence which capabilities you prioritized?

Vick Vaishnavi: Today’s quality professionals are facing unprecedented regulatory requirements, increasing product complexity, and uncertain supply chains due to geopolitical disruption and tariffs, among other things. What we heard loud and clear from our customers from industries such as automotive, aerospace, electronics, heavy industry, and food/beverage is that they wanted more AI capabilities to support their critical tasks, such as managing complaints, retrieving critical data, or conducting supplier audits. They wanted greater intelligence built into their QMS, which also addresses rising security requirements.   

We’ve heard from customers that they need to do more than simply digitize an existing process; they also want a way to identify which issues deserve attention. More specifically, they want to understand whether a complaint resembles a known problem, or to find relevant information within their quality records more easily.

According to our customers, the objective isn’t automation for its own sake. It’s to help quality professionals spend less time gathering and organizing information and more time evaluating risk, conducting investigations, being able to launch new products faster, and improving existing products and operational processes.

Wallin-Kerth: Could you walk us through a complaint—from intake to investigation—and show where Complaint Intelligence changes the work a quality professional actually does?

Vaishnavi: Consider a customer reporting that a component intermittently fails during use. The initial complaint might be brief, with limited technical information and no clear explanation of the underlying cause.

The complaint first enters the established quality workflow, where the organization captures the information it needs to manage and assess the issue. Complaint Intelligence then helps the quality professional in several ways.

First, it generates a structured summary so the reviewer can understand the reported issue without having to reconstruct the narrative from scratch. Second, it identifies similar historical complaints and related investigations that may provide useful context. Third, it provides AI-assisted assessments of potential product risk and customer urgency to help inform prioritization.

The quality professional can then use that information to assess the complaint, determine the appropriate next steps, and pursue an investigation using the organization’s established procedures. Relevant findings might also inform corrective and preventive action, or CAPA, when appropriate.

Wallin-Kerth: How does the AI assess customer urgency and potential product risk, particularly when a complaint is incomplete, ambiguous, or understates a serious issue?

Vaishnavi: The important distinction is that customer urgency and product risk are related, but they’re not necessarily the same thing. A complaint that sounds calm or contains very little detail can still describe a potentially serious product issue.

Complaint Intelligence is designed to help quality teams assess both dimensions by providing AI-assisted assessments alongside structured complaint summaries and connections to similar historical complaints and investigations. That historical context can help a reviewer recognize a pattern that might not be apparent from an individual complaint.

For example, an incomplete report of intermittent equipment failure might warrant closer attention if similar reports have previously been associated with a more significant quality issue. The value of connecting complaints is that it gives the reviewer additional evidence to consider rather than relying exclusively on the wording of the latest report.

Wallin-Kerth: What safeguards help prevent an AI-generated summary from omitting a detail that could change an investigation’s direction?

Vaishnavi: Our approach should be to treat AI-generated summaries as a starting point for review, not as a replacement for the original complaint. Investigators need access to the underlying record so they can verify the summary against the source and examine details that could affect their assessment.

Quality teams should test whether important facts are retained, whether unsupported conclusions are introduced, and whether the output remains useful for different complaint types.

Wallin-Kerth: What decisions remain with quality professionals, and how can they review, challenge, and document the AI’s recommendations?

Vaishnavi: AI is there to support the quality professional, not to take accountability away from them. The tool can organize complaint information, identify potentially relevant historical records, and offer assessments that help inform prioritization. The quality professional still needs to apply the organization’s standard operations procedures and expertise to determine how the complaint should be handled, whether further investigation is needed, what the evidence establishes, and what corrective action is appropriate.

Wallin-Kerth: What evidence do you have so far that these capabilities improve investigation quality or consistency, beyond simply reducing review time?

Vaishnavi: We see the potential value in three areas: more consistent complaint assessment, better use of historical investigation data, and improved identification of patterns that might otherwise be missed.

Structured summaries can help standardize how reviewers understand complaints. Identifying similar historical complaints can bring relevant context into an investigation earlier. And AI-assisted urgency and risk assessments can help teams direct attention toward issues that warrant closer examination.

Wallin-Kerth: How do you evaluate search accuracy, including across languages and specialized technical terminology? How do you ensure that records weren’t missed, overlooked, or misunderstood?

Vaishnavi: Search quality is about more than returning results that sound relevant. In a quality management environment, the system also needs to help professionals find the records that could materially affect a decision.

Reliance NXG Smart Search combines semantic understanding with traditional search. That combination is important because users might describe the same problem in different ways. One person might use a product’s technical terminology, while another describes the symptom in everyday language. Cross-language search also matters for global organizations where related quality events may be documented in different languages.

Ultimately, any AI-driven feature requires the expertise, oversight, and intelligence of quality professionals. Yet their search capabilities are significantly improved and made easier with Smart Search capabilities.

Wallin-Kerth: Smart Search automatically applies filters based on user intent. Can users see and change those assumptions so their searches don’t inadvertently exclude important records? Do individual algorithms affect this, and to what extent?

Vaishnavi: Smart Search is intended to shift Octave Reliance away from rigid keyword matching to true intent-based discovery. A user asking for recent supplier audit findings, for example, shouldn’t have to manually construct every search condition before getting useful results.

However, convenience must not come at the expense of completeness. A filter inferred from a natural-language query could narrow the results in a way the user didn’t intend. In quality management, that matters if an apparently narrow search excludes a record that would change the investigation.

Wallin-Kerth: For customers in regulated industries, what does validating these AI capabilities involve, and how are subsequent changes to AI behavior managed and documented?

Vaishnavi: Validation begins with the intended use of the capability and the risks associated with relying on its output. Customers need to understand what the AI algorithm is intended to support, where human review remains necessary, and what evidence demonstrates that the capability performs adequately in their particular workflow.

For Complaint Intelligence, that could include testing representative complaints, verifying that critical information is preserved in summaries, assessing the relevance of similar-case recommendations, and evaluating whether urgency and risk assessments provide useful support without creating unacceptable risks. For Smart Search, testing could cover retrieval accuracy, multilingual queries, technical terminology, and the consequences of automatically applied filters.

Wallin-Kerth: As quality systems become more connected and include suppliers and contractors, how do you balance easier collaboration with tight control over sensitive records and access?

Vaishnavi: Supplier qualification, audit findings, certifications, corrective actions, and quality records are often interconnected. When these activities operate in separate workflows, organizations can spend considerable time reconciling information and may have less visibility into supplier-related risk.

The latest Reliance NXG release connects supplier qualification and audit workflows, synchronizes relevant supplier records and audit outcomes, and provides traceability from evaluation through approval. That helps organizations coordinate their work without relying as heavily on manual handoffs.

At the same time, the release strengthens controls for external access through automated identity life cycle management, IP-based access controls, and multifactor authentication for supplier, contractor, and other external user accounts.

Ultimately, connected quality management enables authorized people to access the information they need, when they need it, while access to sensitive records remains governed by organizational policies.