IFS Copperleaf says its AI assistant covered running costs in four months
The newly published deployment account pairs strong adoption claims with a selective approach to data preparation. A temporary billing pause complicates the cost picture.
Loading page…
The newly published deployment account pairs strong adoption claims with a selective approach to data preparation. A temporary billing pause complicates the cost picture.
Listen to this story
Copperleaf’s deployment illustrates a split approach to enterprise AI data readiness: it fully modeled finance and marketing-to-sales data, while letting IONA query raw data elsewhere and relying on business teams to define its meaning. The company reports 70% activation and four-month coverage of running costs, but that payback period overlaps Databricks’ temporary pause on Genie One user-usage billing; the charges Copperleaf actually incurred are not specified. The results are company-reported, not independently assessed.
IONA launched in July 2026 and connects governed data from nearly 30 systems, including Salesforce, HubSpot and Microsoft 365.
Copperleaf faced tens of thousands of tables and says querying raw data in lower-stakes areas made it useful months sooner.
Business definitions require a named owner and AI Council review before company-wide use; Unity Catalog applies role-based access controls.
IFS Copperleaf says its workplace AI assistant reached a 70% activation rate and covered its running costs within four months. The results, published November 16, describe a deployment built around business teams teaching AI their definitions—not just connecting databases. But the payback period overlaps a temporary pause in some Databricks usage charges.
The assistant, called IONA, launched in July 2026 and runs on Databricks Genie One. Employees can ask plain-English questions across governed data from nearly 30 systems, including Salesforce, HubSpot and Microsoft 365. Copperleaf’s deployment account, published on Databricks’ blog, describes both the reported results and the changes needed to make those answers useful.
Copperleaf’s financial claim is specifically about running costs. Its summary says IONA covered those costs within four months, alongside the reported activation rate since launch. Those are company-supplied results published by its platform provider, rather than an independent assessment of the deployment’s economics.
Databricks’ official release notes add an important qualification: billing for Genie One user usage is paused from July 15, 2026, through January 31, 2027. The promotion excludes service principals—software identities used by applications—whose usage remains billed. The pause therefore overlaps much of IONA’s initial operating period, without establishing which charges Copperleaf incurred.
The implementation account describes a separate tradeoff: how much data preparation must happen before employees get access. With nearly 30 source systems and tens of thousands of tables, Copperleaf initially faced a large modeling queue. The team changed course, retaining fully modeled data for finance and the marketing-to-sales funnel while letting IONA query raw data elsewhere.
Copperleaf says that choice made data useful months sooner. It did not settle what the numbers meant. Different leaders could still produce different active-customer counts because their teams counted differently. Knowing how the business distinguishes a downsell from churn also required knowledge that was not contained in the tables alone.
Users also brought expectations beyond traditional reporting. The team expected comparisons with old reports; employees instead compared IONA with AI tools they already used. They wanted it to remember context, draft follow-ups and work across systems, rather than simply return a chart.
Copperleaf’s answer was to let employees refine business rules through conversation. A user could specify that partner accounts should be excluded, amounts converted to euros or a particular renewal date used. According to the company, IONA also shows its work when an answer looks wrong and directs users to the relevant data owner so the source can be corrected.
That does not mean every conversational correction immediately becomes a company-wide rule. Copperleaf describes two controls around the shared system:
The definitions feed Genie Ontology, a shared layer connecting business meaning to tables, metrics, dashboards and agents. Copperleaf says this lets IONA recognize that a Salesforce renewal, a contract and a finance forecast concern the same customer. One reported use case brings that together: a seller can ask what a customer owns, where expansion opportunities exist and whether the contract permits a move to SaaS.
Loading discussion...
Join the conversation
Explain where you would draw the line.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.