Architecture

The Coveo™ for ServiceNow integration brings AI-powered search and recommendations directly inside ServiceNow to improve support agent proficiency, enhance customer self-service experience, and allow administrators to make informed decisions based on actual usage analytics data rather than guesswork.

Its search interfaces are hosted in your Coveo organization and are displayed in widgets or used as your built-in Coveo-powered search page. Your custom search interfaces help your ServiceNow instance users, that is, your customer support agents, your employees in general, or your customers, to be more proficient.

Coveo for ServiceNow offers an artificial intelligence powered search experience driven by your data. It not only replaces the built-in search engine in your customer portal, employee intranet, etc., but also introduces new means to search your data, allowing users to find the content that’s the most relevant to their issue. Moreover, the Coveo for ServiceNow search results and content recommendations become more personalized with time, as Coveo learns from user behavior.

Coveo for ServiceNow architecture

As illustrated in the diagram below, the Coveo for ServiceNow application is installed in your ServiceNow instance. It comes with Coveo widgets, a built-in Coveo-powered search page, and the Coveo JavaScript Search Framework. This framework is an open-source library that you can use as an alternative or a complement to the Interface Editor in the Coveo Administration Console when building Coveo-powered search interfaces to display in the widgets implemented in your ServiceNow instance. You can also use it to customize your built-in Coveo-powered search page.

Coveo for ServiceNow connects to your Coveo organization, which hosts your index, the Coveo Search API, the Coveo Machine Learning (Coveo ML), and the custom search interfaces that will be displayed in widgets or used as your built-in Coveo-powered search page.

Coveo also hosts the usage analytics data coming from the search interfaces in your ServiceNow instance. You can access this data in the Coveo Administration Console.

Coveo for ServiceNow architecture diagram

Indexing process

Indexing is the action of retrieving content from a repository, processing this content, and storing the resulting items in a data storage structure called an index.

Typically, you’ll create a ServiceNow source to index the content of your ServiceNow instance. The ServiceNow connector initiates the indexing of ServiceNow data through a ServiceNow account that has access to the content you want to make searchable. This crawling account then sends queries to the ServiceNow REST API and indexes the returned content.

Note

By default, the ServiceNow connector checks for incremental changes in your instance every 15 minutes. You can adjust this frequency in the Coveo Administration Console so that Coveo keeps up with your ServiceNow content updates adequately.

You have valuable knowledge in your ServiceNow instance, but you probably have a lot of helpful content outside of ServiceNow as well. Thanks to a variety of connectors, Coveo can gather all this information in one place, your index, which can then be mined to get this knowledge to your users. In a Coveo for ServiceNow search interface, they can therefore search and find not only ServiceNow content, but also other enterprise content, such as website pages, cloud-hosted files, and emails. The process of indexing this content is similar to the above.

For more information on the indexing process, see Coveo Indexing Pipeline.

Querying process

The query handling process is the following:

  1. A query is made in a Coveo widget or on your built-in Coveo-powered search page and sent to the Coveo Search API. Queries are triggered differently depending on the search interface:

    • With the Main Search and Searchbox, the text entered by the end user in the searchbox becomes the query.

    • With a Case Deflection widget, the end user types in the fields of the case creation form. Coveo extracts the most important keywords from this text and uses them as a query. As the end user continues typing, new queries are fired, further refining the search results.

    • The Recommendations widget doesn’t require any input from the end user to provide content recommendations. When an end user accesses a page that contains a Recommendations widget, their profile and the pages they have already visited are sent as a query. This contextual data will allow Coveo to provide personalized content suggestions.

    • With your built-in Coveo-powered search page, the text entered by the end user in the searchbox becomes the query. When an end user interacts with a facet, queries are also triggered.

  2. The query transits through a query pipeline, which modifies the query according to your query pipeline rules and Coveo ML input, which optimize search result relevance. The modified query is then sent to your index.

  3. Your index orders the matching results from most to less relevant according to the input received from Coveo ML.

  4. The index returns the search results to the Search API, which forwards them to the search interface in your ServiceNow instance. The widget or search page then displays the results to the user via the search interface you built.

  5. The search interface sends user behavior information, that is, which search result user clicks, through the Coveo Usage Analytics (Coveo UA) API.

  6. This usage analytics data is stored in your Coveo organization.

  7. Coveo ML learns from usage analytics data and deducts user behavior patterns. This information helps returning more relevant results every time a ServiceNow user interacts with a Coveo for ServiceNow widget or with your built-in Coveo-powered search page.