September 12

What's New in QORE

We're excited to roll out this release, bringing a refreshed platform structure, a new orchestrating Operator, and expanded governance and observability capabilities. Here's everything that's new.

Terminology changed in this release. Older screenshots, links, and release notes may still use the earlier names.

Was	Now
Workspace	Project
Studio (as a separate space from Console)	Console — Studio and Console merged into one space
Knowledge Hub	Knowledge

Account navigation and projects

Account location moved: Account (previously top-right, showing your name) is now on the left-hand side.

Multi-account support: Users with access to a single account see that view; users with multiple account access get a switcher under the account panel.

Workspaces → Projects: Workspaces have been renamed/transitioned to Projects.

• What lives inside a Project: Apps, Agents, Knowledge Bases, and Design Systems can all be created within a Project.
• Shared assets within a Project: Any asset created within a Project (e.g., a Knowledge Base or Design System) can be reused across other assets in the same Project — e.g., a Knowledge Base built for one agent can be attached to another agent, a Design System can be pulled into a new app — without recreating it.
• Why it matters: Eliminates duplicate records (e.g., multiple copies of the same Knowledge Base) across apps/agents within a Project.

Console and Control Center switcher

A switcher sits on the header, right next to the platform logo, letting you toggle between the Console and Control Center spaces.

Introducing Operator

• What it is: The Operator is a power-packed orchestrator agent with visibility into what's happening across the platform, your account, and your project. At its core, it's a multi-agent orchestrator.
• What it replaces: Previously, users had to go to App Builder, Agent Builder, and Knowledge Builder separately, set each up, and then wire them together. Now, all of that can be done from a single Operator interface — whether you're building new agents or using pre-built ones.

How it works (multi-agent orchestration)

• State your intent

  You tell the Operator what you want to achieve.
• Decompose the intent

  The Operator breaks the request into steps.
• Identify AI agents

  It selects the agents that can deliver each step.
• Orchestrate task assignment

  It assigns work to those agents.
• Receive agent updates

  Completed agents report back to the Operator.
• Present results

  The Operator packages and presents the outcome.

Leveraged in 2 spaces:

• Console: Can web search and web crawl. Given a URL, it can absorb a design system from that URL to build an app, or turn the URL's content into a knowledge base — interpreting intent from the prompt.
• Control Center: Aware of everything happening on the platform. You can prompt it about any module here — e.g., which apps were published in the last few days, which agents got created recently, where the most credit spend/utilization is happening.

Console and Control Center

Console

Your playground for building and consuming. Operator capabilities include:

• Play with data, query and ask questions of data
• Build presentations
• Build apps, knowledge bases, agents, agentic applications
• Web search and web crawl via the Operator
• Capability list is scoped for now but will keep expanding with more skills

Control Center

Where you monitor and stay in control of your assets — visibility into what your agents are doing and how. Built on a few foundational layers:

• AI Gateway: Control which model families and models are used across building and usage in Console. Enables setting model choices per your organization's mandates and guidelines.
• Observe: Surfaces all runs, traces, and actions performed at every AI interaction point across the platform. Logs are captured here — useful for understanding the nature/quality of agent responses and for AI agent auditing.
• Cost Control: Shows current credit allocation, where spend/utilization is concentrated, and a usage-vs-limits picture.
• Evals: Continuous testing/evaluation on production data (online/production evals — full deep-dive coming in a future release).

  • Governance/audit-oriented, not for the developer building the agent — hence it sits in Control Center rather than Console.
  • How it works:

    • Configure evaluators and assign a judge model (a separate LLM) to assess agent answer/action accuracy.
    • The judge model scores responses against the evaluator's parameters.
    • The configurator (governance/auditor) sets a threshold — effectively a confidence bar.
    • If a response's score falls below the threshold, it's flagged for human review (human-in-the-loop).
    • The human reviewer decides the next action: pass learning back to the agent, or flag it to the agent's developer, who can then tweak regression test cases, modify agent instructions, or revisit the agent's knowledge base.
• Govern:

  • Agent Controls: Hawk-eye view of all agents under the account, with ability to enable/disable any agent.

    • Kill switch: stop a runaway/looping agent — kill all agent executions on the account, or kill a specific agent's execution.
    • Feature set will keep expanding in future releases.
  • Publish Review: Governs publishing/promotion of assets (apps, agents, knowledge bases, etc.) once built and ready to go broader — e.g., account-wide or to the store. Adds a publish-control layer with a designated governor/reviewer.
  • Govern Graph: Visualizes FI (financial institution) data, knowledge, products, policies, and regulatory information governing the ecosystem — plotted as a graph showing connections (nodes and spokes) across disparate data records and knowledge systems.

    • Currently a visualization layer only; sample data available on some demo accounts.
    • Future plan: make agents aware of the Govern Graph so it can be built from/leveraged by agents, not just visualized.

Store and upcoming RBAC

• The FinWorks tab / Solutions (previously in the top header) is now part of Console → Store.
• Store is home for apps and agents that have completed the publish process.
• Publish status today:

  • Apps: publish flow available now.
  • Agents: publish currently handled from the backend; in-product publish flow planned for a future release.
• Maker-checker model: the builder (maker) creates the app/agent; the reviewer (checker) reviews it from the Control Center space.
• Access today vs. future: All spaces are currently open to everyone. In the next phase, with updated RBAC roles and permissions, access will be scoped per configuration.

Agent Executor v1.4

• Agent Executor v1.4 has been released alongside these platform capabilities.
• Migration: Agents currently running on lower/older versions can be migrated to v1.4.

  • Before migrating, verify the agent continues to work as expected on v1.4 — check that it does not break or malfunction.

Header changes

Documentation, Help, and Release Notes moved from the top header to the sidebar (bottom left):

• Release Notes → redirects to docs.uptiq.ai (core platform documentation).
• Help and Documentation:

  • User Guide → redirects to platform documentation.
  • Report an Issue → flags issues directly to the support team.

We'll continue to expand on these capabilities in upcoming releases.

Let us know what you think of this upgrade—keep building, and keep the feedback coming!

---

Maintained by Abhishek Paul AI-assisted, human-approved