Oracle Adds a Value Dashboard to AI Agent Studio for Fusion Applications
Updated: 6 days ago
Title: Oracle Adds a Value Dashboard to AI Agent Studio for Fusion Applications
Date: 16 April 2026
Type: Blog
Author: SAASiQ (contact@saasiq.ai)
Word count: 1357 words
Reading time: 6 min
Published: 16-04-2026
AI Agent Studio, the tool Oracle Fusion Applications customers use to build and run their own AI agents, gets a Value dashboard in the 26B quarterly update, which reaches the first customers' test environments on 1 May. It reports the time and money saved by each published agent team, and Oracle's readiness note is plain that these are estimates: an administrator sets what one run of an agent saves, and the dashboard multiplies that by the number of runs. Oracle announced the feature on 24 March, calling it the Agent ROI dashboard, alongside an Agentic Applications Builder, and on 9 April it listed the agentic applications it has built for finance, supply chain and HR.
How the Value dashboard works
Oracle added the feature, under the title 'Measure AI Impact with the Value Dashboard', to its 26B What's New for Common Technologies on 27 March. It sits on a Value subtab of the Monitoring and Evaluation tab in AI Agent Studio and shows four headline numbers: Total Time Savings, Total Cost Savings, Total Usage and Total Agents. The figures can be filtered by date range, agent team or business function, and are broken down by Fusion product and by individual agent team.
The savings come from two settings on each agent team. Under a section headed 'Capture estimated Cost and Time savings', the administrator enters 'Time saved each agent run' and 'Cost saved each agent run', then publishes the team again. If nothing is entered, Oracle's default values are applied. The system counts the runs and the customer supplies the value of each one, and Oracle's note describes the metrics as designed to help customers 'estimate and visualize the tangible ROI' of their AI work.
So the dashboard shows how often each agent team is used and what that use is worth on the customer's own assumptions. Oracle's documentation does not describe any link to the general ledger or any check of an agent's output against results in Fusion Financials. SAASiQ's view is that the per-run figures are worth agreeing with finance before an agent team is published, because every total on the dashboard is a multiple of them.
Monitoring that came earlier
The Value subtab joins monitoring Oracle has been adding since last autumn. At AI World in Las Vegas on 15 October 2025, Oracle gave AI Agent Studio a monitoring dashboard showing sessions, latency, error rates and token usage, and said the token figures make costs more predictable by measuring consumption of premium large language models. The same release added a marketplace for partner-built agents, a choice of models from OpenAI, Anthropic, Cohere, Google, Meta and xAI, support for the Model Context Protocol (MCP) and agent-to-agent connections, a credential store, and prompt libraries with version management.
The 26A update, which reached the first group of customers in February, added evaluation of answers drawn from documents. Oracle measures three things: groundedness (how well the answer matches the retrieved content), answer relevance and context relevance. The scoring uses a language model as the judge, and the results appear in the Monitoring and Evaluation tab.
26B also brings Edit in Playground, which lets a builder test one part of an agent team, such as the supervisor, a worker agent or a single language-model step, without running the whole workflow. Changes are kept in a run history.
What Oracle announced on 24 March
Oracle announced the latest AI Agent Studio features at its AI World Tour event in London on 24 March. The Agentic Applications Builder assembles applications from agents in a natural-language environment, without traditional coding. The other additions were workflow orchestration with rules for how work moves between steps and points for human oversight, content intelligence that combines unstructured documents with transaction data, contextual memory across interactions, multimodal inputs and outputs (images, audio and video), and real-time monitoring and debugging with a prompt playground. Oracle said the Agent ROI dashboard would show time saved, cost savings and productivity gains per agent.
Chris Leone, executive vice president of applications development, described the aim as 'the foundation for a more autonomous enterprise'. Oracle also said more than 63,000 partner staff are now certified in AI Agent Studio, against more than 32,000 in October.
The 26B readiness notes show how an agentic application is put together. A builder uses a new Apps tab and combines information displays, Ask Oracle advisors, actions and communications, and only agent teams marked 'Use in apps' can be included. Two new connectors let agents search content held outside Fusion. The SharePoint connector requires Fusion to be registered as an application in Microsoft Entra ID, and both it and the web crawler connector sync through a scheduled process that Oracle says should run no more often than every 30 minutes. Content indexed this way counts towards a 4TB limit.
The agentic applications
Oracle introduced 22 Fusion Agentic Applications on 24 March, each a workspace in which a team of agents works towards one business outcome. At the New York stop of the AI World Tour on 9 April it named 12 for finance and supply chain, among them the Claims Settlement Workspace, the Collectors Workspace, the Cost Accounting Close Workspace and a Sourcing Command Center, and eight for HR.
Oracle describes each one by the outcome it is aimed at. The Collectors Workspace, for example, is meant to help finance teams collect cash faster and lower days sales outstanding, and the Maintenance Operations Workspace to reduce unplanned downtime. Neither announcement gave customer names or measured results. The Value dashboard is the tool Oracle offers for customers to put numbers against those outcomes themselves.
Security and access
When Oracle launched AI Agent Studio in March 2025, it said agents follow Fusion's security configurations, policies and access controls, so users see only the data and functions their roles allow. The 9 April HR announcement repeated that agents inherit the role-based access and data permissions already set up in Fusion. An agent acting for a user gets no more access than that user has.
Two 26A features take agents outside the Fusion screens. Agent teams can now be called through a REST API from other applications, with access decided by the roles assigned to the team and checked by a user assertion token issued through Oracle's identity service. An MCP tool lets agents call external MCP servers without custom REST wrappers, and users can now talk to Oracle agents from Microsoft Teams and Slack, with the Teams integration using OAuth 2.0 to carry the user's identity into Fusion.
What it costs
AI Agent Studio has been available at no additional cost to Fusion Applications customers since its launch, and Oracle repeated that on 24 March. On Oracle's third-quarter earnings call on 10 March, co-chief executive Mike Sicilia said the company had delivered well over 1,000 agents inside its applications and that the embedded agents come at no additional cost. In the same quarter cloud applications revenue was $4.0 billion, up 13 per cent, with Fusion Cloud ERP at $1.1 billion.
The one cost Oracle has pointed to is model usage. Its October release said token usage is measured for premium language models, so a customer choosing one of those for its agents should ask Oracle what the consumption will cost.
The database side
Agents built outside Fusion reach Oracle data through the database. Oracle announced Oracle AI Database 26ai at AI World on 14 October 2025. Oracle has offered an MCP server in its SQLcl command-line tool since July 2025, and since 23 December 2025 each Autonomous AI Database on Serverless infrastructure, on 19c or 26ai, has had its own managed MCP server, which applies the database's existing roles, auditing and Virtual Private Database policies.
On 24 March Oracle also announced Deep Data Security for 26ai, which passes the identity of the end user, or of an agent acting for one, to the database at runtime so that SQL policies decide which rows, columns and cells come back.
26B reaches test environments for the first group of Fusion customers on 1 May and production on 15 May, and the second and third groups take it in June and July.
SAASiQ - Intelligent Solutions for SaaS ©
