Qase Product Updates: April 2026

Scaling an engineering organization is a continuous evolution. The tools that work for one team rarely work unchanged for ten, and the ones that work for ten rarely work unchanged at the org level. Each step up changes the shape of the problem.

This month, the work was concentrated at that org-level layer: identity that spans workspaces instead of living inside one, AI billing that matches how enterprises buy software, and a credentials architecture we can put in front of a security team without caveats. Underneath, we made AIDEN's generation pipeline materially faster and shipped a Confluence Cloud integration for teams whose requirements live there.

Here's what shipped in April.

Pillar 1: Identity at the Organization Layer

Multi-Workspace SSO and SCIM.

If your organization runs more than one Qase workspace e.g. one per product line, one for platform, one for a contracted team thenidentity management has been a per-workspace exercise. Provisioning, role mapping, and offboarding all happened separately in each one. That's workable at small scale and tedious at real scale.

We've moved SSO and SCIM up to the organization level.

SAML is configured once for the whole organization. You set the Sign-in URL, IdP issuer, certificate, and the SAML attribute Qase reads for group membership. If you don't specify the attribute, we auto-detect the standard names from Okta, Entra ID, AD FS, JumpCloud, Google Workspace, and OneLogin. Custom attribute names work too but just set it explicitly and we'll use that.

From there, you map IdP groups to workspaces. One group can target multiple workspaces; multiple groups can target the same workspace. At login, Qase reads the user's group membership from the SAML assertion and provisions them into the right workspaces with the right defaults.

The deprovisioning logic is the part worth pausing on, because it's where multi-workspace SSO often goes wrong: when a user is removed from a group, we check whether they still have access through any other group before revoking. No accidental lockouts when memberships overlap. Delete a user at the org level via SCIM and they're deactivated everywhere. It’s one action, comprehensive effect.

AIDEN now runs on Dedicated and Enterprise.

AIDEN credits used to be metered at the workspace layer. That worked for single-workspace customers and got in the way for organizations running multiple workspaces under one contract.

We've moved credit accounting to the organization level. One pool of AIDEN credits, shared across every workspace in the org, with consumption visible in one place. Stripe and metering treat the organization as the customer.

The downstream effect is the headline: AIDEN is now supported on Dedicated clusters. Cloud workflows are unchanged. It’s the same flow, same bills, no migration. Org-level billing also lays the rails for the rest of our paid surface area to follow the same pattern.

Pillar 2: A Faster, Tighter AIDEN Pipeline

We rebuilt AIDEN's generation pipeline this month, and the work falls into two categories: making it faster, and making the credentials story one we'd be comfortable defending in a security review.

Generation is roughly 2.5x faster.

We tore the pipeline apart and rebuilt it leaner. Whole stages were removed or merged where the model was doing redundant work. Action types are now classified up front, so simple steps skip LLM calls entirely. Common selector lookups short-circuit when the answer is obvious. Screenshot, video, and browser-instrumentation work runs in parallel rather than blocking the main loop.

We also moved to newer models with prompt caching. It’s the same quality of reasoning at a fraction of the tokens, and noticeably fewer hallucinations on tricky steps.

Net effect: roughly 2.5x faster generation on average, no quality regression.

Auth Profiles and Functional Environments.

Anyone who's automated against more than one host knows the friction: staging, preview, and ephemeral subdomains are the same application, but you end up duplicating environment configs and login flows for each one. When credentials rotate, you fix it everywhere.

Environments are now functional. A single environment can cover multiple hosts united by a host pattern, so staging, preview branches, and ephemeral subdomains live under one logical environment instead of three copy-pasted entries.

Auth Profiles are a new, reusable entity. AIDEN generates the login scenario once, and you link the profile to any environment that needs it. Rotate credentials in one place; every environment using that profile picks up the change.

Credentials never reach the model. Secrets live encrypted at rest. At runtime, they're injected directly into the browser. Never into an LLM prompt, never persisted alongside the test memory. If it's a secret, the model never sees it.

That's a credentials story you can put in front of a security review without flinching.

Pillar 3: Confluence Integration

For a meaningful slice of teams, requirements live in Confluence. We shipped a full Confluence Cloud integration so those requirements flow into your testing workflow without manual translation.

What's available now:

• Link Confluence requirements directly to test cases for end-to-end traceability.
• Generate test cases from Confluence pages using AI, with output flowing straight into Test Designer.
• A traceability matrix spanning requirements → tests → defects.
• Bi-directional defect linking between Qase and Confluence.
• Confluence-side linking, observability, and an entry on the Atlassian Marketplace.

Pillar 4: Dashboard and QQL Improvements

A couple of small changes that have already shown up across customer dashboards:

Section Title Widgets. Dashboards can now be organized into labeled sections instead of an undifferentiated grid of charts. Find them in the widget list under separators. Particularly useful when one dashboard has to serve multiple audiences. For instance, engineering metrics on top, leadership view below.

Editable chart colors in QQL. QQL charts now support custom colors. One change, and your dashboard starts telling you where to look. You can improve contrast for the screen you actually present from, or highlight the metric that matters this quarter. Small change, but very satisfying.

Pillar 5: Reporters and Ecosystem

The usual round of work on the integration surface:

• The Behave reporter now supports running tests in parallel.
• qasectl is installable via Homebrew (brew install qasectl).
• Cypress and Testcafe reporters support specifying the browser name as a parameter.
• New native reporter for Reqnroll.
• All reporters can now update test case tags during reporting.
• The MCP server supports nested steps when creating test cases.
• 10 bug fixes shipped.

April's work was concentrated where the customer base has been growing fastest: the org level. Identity, AI billing, and credentials all moved up to where they need to be for organizations running Qase at scale. Generation got faster. Confluence requirements now have a real path into the test layer.

Happy testing!