KM and IT Find Collaboration Productive

The role of IT in implementing search solutions occurs primarily when connecting sources and, in the case of search, in the front-end integration, according to Sébastien Paquet, VP of AI strategy at Coveo. “Once the data sources are connected, Coveo does not require developers, just configuration,” he noted. “Knowledge managers can tweak the ranking factors or other performance features.” Coveo provides native connectors for widely used enterprise applications such as SharePoint, along with generic connectors for indexing content from websites, APIs, databases, and other custom sources. “In some cases, custom development is needed to push data to Coveo, typically in ecommerce, where content is scattered,” he added.

Knowledge Relies on Quality Data

As with any application, data quality is very important, and Coveo can help knowledge managers identify where improvements can be made. “Since even large enterprises don’t have the resources to fix every data issue,” Paquet pointed out, “it’s important to know what users are looking for and fix the most necessary content issues so that users get the right answers.” Coveo helps identify problem areas through its use of analytics reports, which can reveal knowledge gaps where users are asking questions that existing content does not adequately answer. That helps knowledge managers focus on the content improvements that will have the greatest impact.

Coveo is delivered as a SaaS solution. “The cloud simplifies deployment and time-to-value and provides automatic scaling,” commented Paquet. “Even more specifically with respect to LLMs, Coveo handles the complexity of running them in production, from prompt engineering and evaluation to maintaining output quality as models evolve, while managing multi-region inference to deliver rapid and reliable experiences at scale, whether the system is being used by a human agent or directly by the customer.” 

Guardrails are embedded, so if a customer asks an agent a question that is outside the scope of the agent’s authority, Coveo will not retrieve the information. Instead, it will provide the agent with an appropriate default response that can be configured by the KM team. The advanced feature of generative answering allows Coveo’s search solution to respond in real time with accurate answers grounded in the enterprise content. “Setting up the guardrails and rules does not require interaction with IT, however,” concluded Paquet. “It is done within Coveo.” Customer Support and More Customer service is a common use for KM platforms. It is a growing market driven both by the desire for human agents to provide prompt and accurate answers to customers, along with a robust self-service option that deflects calls and saves money. According to Mordor Intelligence, the contact center market was $72.86 billion in 2025, and is expected to more than double to $184.24 billion by 2031 (mordorintelligence.com/ industry-reports/contact-center-software-market).

Serviceaide has its roots in IT service management (ITSM), keeping track of system changes, configurations, and trouble tickets from computer users, but eventually broadened its scope to encompass enterprise service management (ESM). “You can’t have ESM without KM,” said William Guinn, co-founder and CTO of Serviceaide. “Knowledge-centered service (KCS) is a core enabler, ensuring that ITSM can manage ESM use cases across an enterprise.”

An AI-based KCS implementation is often an AI entry point for an organization or department looking for increased employee productivity. “An enterprise-wide KCS implementation provides the benefit of a single point of knowledge of all employee needs,” explained Guinn, “and by federation, is straightforward to implement since department information sources can be ingested through connectors available into most common sources such as SharePoint, Confluence, application APIs, intranet pages, and documents in file systems.” Implementations can be rapid, since the security on existing knowledge artifacts is carried forward as it is ingested, and the information can continue to be managed in place.