Glean’s analytics expansion brings business definitions into the query

Public developments and original analysis · Announcement date: 2026-10-20 · Sources

On October 20, Glean introduced Glean for Data Analysis, bringing data from business applications, BI and warehouses together with context from documents, conversations and tickets. It describes natural-language exploration and analysis that can feed reports and data applications. The announcement lists Snowflake and Power BI as generally available and Tableau and Looker as beta; connections still depend on the organisation’s environment and permissions. A useful query therefore needs information beyond the table itself.

For example, an operations report shows fewer “qualified leads” than last week. Before asking AI to explain deteriorating campaign performance, consider a hypothetical rule change: this week, qualification requires a business email and a purchasing plan; last week, leaving contact details was enough. Attendance may be unchanged while the report is counting a different population.

If the query retrieves only the current filter and treats it as the historical rule, AI can describe a definition change as a business decline. Comparing campaign performance may require recalculating both periods under one rule. Understanding how many leads sales actually received each week calls for retaining the original weekly results and explaining when acceptance criteria changed. Both are valid questions, with different answers.

Retrieve the old definition before choosing a comparison

Glean’s announcement pairs governed structured data with unstructured business context, while its product page describes extending analysis into recurring reports. For this weekly report, relevant context includes the effective date of form rules, the notice explaining the change and the list sales actually accepted. A rule document explains the calculation; an acceptance record helps establish whether the change reached daily work. One cannot stand in for the other.

Mokyun’s published knowledge-retrieval scope includes finding evidence, checking applicable versions and retaining sources. That could contribute to preparation for this analysis: organise successive lead-qualification rules by effective date and attach the relevant definition to each reporting period before an authorised query is run. That preparation supplies definitions and sources; database access, calculations and permissions need separate implementation in the project environment.

Finding a change notice establishes that a rule changed. It does not establish how much that change affected eventual sales when content, channel spending and follow-up may also have changed. A useful report can first identify which figures are comparable, then present explanations still requiring evidence, rather than supply a confident causal story prematurely.

• Mokyun enterprise knowledge and digital employees

Mokyun Editorial · neo@mokyun.com