TITAAutomation

Changelog

Product updates and release notes. Newest first.

v0.20.0 · 2026-09-22

The free plan is free, and it does not expire

  • Starter is free forever and you can actually use it. Downloading the Designer used to need an unexpired trial or a paid plan, so the day a 30-day trial ended the account kept a free plan with a monthly run quota and no way to install the app that spends it. Any account can now download the Designer, on any plan.
  • The free plan now runs 100 tests a month, up from 25, with one agent and one project. Enough to evaluate the product properly rather than glance at it.
  • Nothing is taken away when a trial ends. Your projects, workflows and test cases stay exactly as they are. The paid features stop — AI generation, selector healing, vision, SSO, Git sync and CI triggers — and the Starter quotas apply again. No card is ever required to keep using the free plan.
  • Teammates who are not owners or admins can open the Designer. Someone invited as an Author or Operator signed in successfully and was then turned away by a “license required” dialog.
v0.16.12 · 2026-08-26

Security: tenant isolation, shipped defaults and AI spend caps

  • An agent request can no longer choose which workspace it acts on. The tenant now comes from the agent's signed token rather than from the request, so one customer's agent cannot reach another's jobs.
  • The server refuses to start on a shipped default password. Every placeholder credential now blocks startup and names itself, instead of a deployment quietly running on documented defaults.
  • AI calls made during runs consume a daily budget. Vision, selector healing and generation share a cap, so a runaway loop cannot spend without limit.
  • launchApp no longer starts shells and script hosts (cmd, PowerShell, .bat, .ps1, .vbs and friends). A workflow step that can start a shell can run anything on the machine. An agent that genuinely needs it can opt in, and can also be restricted to a named list of applications.
  • Rate limiting identifies the real caller and now covers more than sign-in.
v0.16.11 · 2026-08-26

Fix: Workflows and Test Cases load again in the portal

  • Both pages showed “Application error” instead of your data. They are back, and the underlying fault — a server-side render that failed on an unexpected response — no longer takes the whole page down.
v0.16.10 · 2026-08-26

Selector healing tells the truth about what it did

  • A healing hint no longer promotes the wrong element. A hint asking for a data-testid matched elements that merely had some other attribute, so healing could confidently pick the wrong control.
  • The run trace no longer claims a hint worked when it did not. “hint-applied” was decided before the result was known.
  • Workflow lists are strictly per project. The query returned every workflow regardless of project, so a second project showed the first one's work.
v0.16.9 · 2026-08-26

Delete workflows and test cases from the portal

  • Both lists have a delete button on each row, behind a confirmation that names what you are about to remove. Until now, tidying up a list meant opening the Designer.
  • The button appears only for roles the server actually lets delete — Owner, Admin and Author — rather than showing everyone a button that fails.
  • Deleting a test case also deletes its workflow, unless another test case still uses it. The confirmation says so, because a workflow vanishing from a list you were not looking at is otherwise indistinguishable from a bug.
v0.16.8 · 2026-08-25

Fix: saving an older workflow no longer fails

  • “Workflow not found” when saving a workflow created before projects existed. Those rows carry no project id; the Designer lists them deliberately, but saving one was refused. Listing, renaming, saving and deleting now share one rule, so they cannot drift apart again.
v0.16.7 · 2026-08-25

Deleting a test case cleans up after itself

  • A deleted test case no longer leaves its workflow behind in the Workflows list, where the test looked like it had moved rather than been deleted. A workflow still used by another test case is always kept.
  • Leftover workflows can be deleted at all. Anything created without a project id refused every attempt to remove it.
  • “Must use desktop source” no longer reads like maximize is off-limits for browsers. The message now says what each source does.
v0.16.6 · 2026-08-25

Tell the AI how you want a step fixed

  • “Suggest fix (AI)” takes your instructions. It used to guess on its own; now you can say what you want changed and it works from that.
  • Steps can carry a healing hint for when they break mid-run. Describe the element in your own words once, and healing has something to work from instead of guessing from a broken selector.
v0.19.0 · 2026-09-22

TITA now speaks 12 languages

  • Japanese, Korean, Vietnamese, Thai and Russian join the seven already there. The whole product is translated — every page, dialog and menu in both the portal and the Designer: English, Español, Français, Deutsch, Português, Nederlands, 中文, 日本語, 한국어, Tiếng Việt, ไทย and Русский. Switch from the language menu in the header, or in the Designer under Settings → Language; it changes instantly.
  • Three long-standing wording faults went with them. The warning about no agents being connected stopped mid-sentence and lost the advice that followed it; a schedule with more than one retry read “3 ies”; and several dialogs showed escape sequences as text instead of line breaks and quotes.
  • The portal is no longer rendered in Times New Roman. Its font was referenced everywhere but never defined, so every page fell back to the browser's default serif. It now uses the system font — which also gives Japanese, Korean and Thai their proper typefaces.
v0.18.0 · 2026-09-22

Work on a project together without overwriting each other

  • The Designer's Project Explorer keeps itself up to date. Workflows, folders and test cases that teammates add, rename, run or delete appear within seconds, so nobody writes a test that already exists. If someone saves the workflow you have open and you have nothing unsaved, their version loads.
  • Two people saving the same workflow no longer erase each other. A save that started from an older version is refused, and the Designer asks whose version to keep — with yours written to a backup file if you choose theirs.
  • Clicking in the Project Explorer does what it looks like. Folders now have a folder icon and say “(empty)” when they hold no tests, workflows open on a single click, and a test whose workflow has been deleted says so instead of opening a blank canvas.
  • Every request checks who is asking. The server confirms you are a member of the project on each request, not only at sign-in, so removing someone takes effect immediately. Update the Designer for this one: versions older than 0.18.0 cannot sign in.
v0.17.4 · 2026-09-03

Deleting a project really deletes it

  • Projects can be deleted from the Projects page (owners and admins), and everything inside goes with them: workflows, test cases, folders, environments, schedules and publish history. Until now those survived their project — invisible, still counted against your storage, and in the case of schedules still starting jobs for a project nobody could open. Deleting asks you to type the project name, because it cannot be undone. Run history is kept as the execution record.
  • An empty repository works as a Git remote. Creating an empty repo and pushing a project into it is how most people start, and it failed with “No default branch found”. Clone Repo now attaches it and reports nothing to pull; Commit, then Push, publishes the project into it.
  • Re-pointing a project at a different repository works. References to the previous remote survived the switch, so the branch lookup resolved a branch belonging to a repository no longer in use.
v0.17.3 · 2026-09-02

CI/CD, documented end to end

  • The Guide gains a CI/CD section. It covers the whole path — mint a token in the portal, copy a test case or folder id, then trigger a run, poll it and fetch JUnit from your pipeline — including the failures that mislead: a 401 means the value is not a TITA CI token (a Gitea or GitHub token gives the same answer), and “No matching test cases” means the ids belong to another project, because the token decides the project. CI triggers and Git sync are Enterprise.
  • Choose Generic Git for Azure DevOps, GitLab and other hosts. The Designer's provider list offered names that changed nothing at all; it is now Generic Git, Gitea and GitHub, with guidance on which to pick. Azure Pipelines cannot run a Jenkinsfile, so a repository holding one still needs an azure-pipelines.yml making the same calls. (The portal's provider field is a different thing: it decides which API opens pull requests.)
  • A project folder can be re-pointed at a different remote. Linking a folder that had been linked before failed with “a branch named 'main' already exists”.
v0.17.2 · 2026-09-02

A failed repository link no longer leaves your token behind

  • A failed source-control link no longer leaves your access token in .git/config. The Designer puts credentials into the remote's URL while cloning and scrubbed them only when it finished, so a failed attempt left the token in plain text — precisely when someone retries, opens the file, or sends the folder to a colleague for help. Every failure path now scrubs first.
  • A failed link says which of two things went wrong. “Failed to authenticate user” is the same answer a server gives for a bad token and for a repository it will not confirm, which sent people to check their password when the URL was what was wrong.
  • Test case and folder IDs can be copied from the portal. The CI API takes them and nothing displayed them, so wiring up a pipeline meant reading them out of browser DevTools. Every row now has a copy control for its own id and for its folder's.
v0.17.1 · 2026-09-01

Minimize and maximize accept a browser window

  • minimizeWindow and maximizeWindow take Source “browser”. The obvious choice when the window you want minimized is the browser used to fail validation and tell you to pick desktop instead.
  • A local run minimizes the browser, not whatever is in front of it. In the Designer these steps acted on the foreground window, so the same workflow behaved one way locally and another on an agent.
v0.17.0 · 2026-08-27

19 new actions, and steps that cannot run now say so

  • 19 actions, in new palette categories. Files: copy, move, delete, exists, wait for a file, list a folder, unzip. Data: extract with a regular expression, format a date, calculate, parse JSON, read a CSV row, assert a variable. Clipboard and input: set and get the clipboard, send keys. Integration: send an email. AI / Vision: assert what is on screen and extract a field, both without a selector. Each behaves the same in a local run and on an agent.
  • Loops and conditions actually run. Both were in the palette but neither runner acted on them: a loop's body ran once whatever the count said, and on an agent the run failed outright. A loop now repeats its steps a set number of times or while a condition holds, and an if runs its steps only when its condition holds — with a cap, so a condition that never goes false cannot hang an unattended run.
  • A local run no longer reports success for steps it cannot perform. Anything outside the local runner's repertoire logged “Simulated action” and passed, so a desktop test could finish green having done and verified nothing. It now fails the step and says whether to run it on an agent instead.
  • A failed step says why. The reason reaches the failure badge and the step tooltip, instead of every failure reading “step returned false”. And sendEmail takes its server and login from environment variables set per environment, so no password is ever written into a workflow.
v0.16.5 · 2026-08-24

The Designer’s test tree behaves itself

  • A blank workspace stays blank. Opening Designer with a blank workspace correctly showed an empty Project Explorer — and then quietly filled it with an earlier session’s folders and test cases the moment you did anything else, like recording a step or cancelling a capture. A blank workspace now stays a clean scratchpad for the whole session.
  • Clicking a test case opens it. Clicking a test case in the Project Explorer did nothing at all — drag-and-drop support was treating every ordinary click as the start of a drag and skipping the open. Clicks open, drags still drag.
v0.16.4 · 2026-08-24

Record Replacement shows you it started

  • Re-recording a single step no longer feels like a dead button. In Designer, Record Replacement did arm replace mode and switch to the Recorder — but its confirmation appeared at the top of the step properties panel while the button itself sits at the bottom, past every selector field. If you had scrolled down to fix a selector, which is the usual reason to reach for it, nothing visible happened at all. The confirmation now sits directly above the button, and the button changes to “Replacing step N — record now” while it waits for you to record.
v0.16.3 · 2026-08-24

Unattended web tests run headless

  • Runs triggered from CI go headless. A pipeline run no longer tries to open a browser window on a build agent that has no desktop to show it on — runs that stalled or failed for that reason now simply run. If your runner does have a desktop and you want to watch, pass {"executionMode":"visible"} in the trigger body.
  • Schedules can run headless too. A new Headless (web) checkbox on Designer’s Schedules tab stops a nightly web test from opening a real browser window on whichever machine picks it up. It’s off by default, so existing schedules behave exactly as before.
  • Downloads survive a credential rotation. Rotating the storage credentials behind the installer could leave the Download button serving an error page instead of the installer. The download now confirms the file is reachable before redirecting, and serves it directly if anything is wrong — slower, but working.
v0.16.2 · 2026-08-22

Dead runs stop pretending to be alive

  • Runs that died with their agent no longer sit at “Running” forever. An agent shut down, disconnected, or restarted under a new identity mid-run left its run marked Running indefinitely — some for a month — with nothing in the product able to clear it. Those are now closed out automatically. They are deliberately not retried: an agent that merely lost its network may still be driving a real desktop, and a second attempt would replay those clicks on top of the first.
  • A dead-lettered run reads as “Failed”. It previously showed a raw queue term in a grey pill — alarming, and not what cloud sync displayed for the same run.
  • An SSO save refused for plan reasons says so. A Team admin who ticked “Enable single sign-on” was told only workspace admins can change SSO settings — which was false, and sent them hunting for a permissions bug instead of showing that SSO needs Enterprise.
  • Reset a locked-out user’s two-factor authentication. Losing both the authenticator app and the backup codes used to require manual database surgery. Platform owners can now reset a member’s 2FA from the member list. It’s audited, and it also clears a half-finished enrollment that would otherwise block the user from starting over.
v0.16.1 · 2026-08-04

License Plate Reader: a gate that keeps working when the internet doesn’t

  • Decisions happen on site. An on-site receiver enforces your allow list locally, so a barrier still opens when the link to the cloud is down. Decisions taken during an outage queue up and arrive once the connection returns — verified against a 22-second cloud outage, with allow and deny both decided correctly on site.
  • Pair a receiver from the portal. Sites show whether a receiver is paired and when it last called home, with buttons to issue, re-issue and revoke its token. Activity gains a second table for on-site decisions showing when each was made and when it arrived — the gap is the length of the outage. Translated in all seven languages.
v0.16.0 · 2026-07-27

Plans now do exactly what the pricing page says

  • Your plan's limits are real. Test runs per month, agents, projects and organizations are now enforced to the numbers on the pricing page — Starter 25 test runs / 1 agent / 1 project, Pro 300 / 2 / 1, Team 1,200 / 8 / 10, Enterprise unlimited. Previously these were advertised but never applied.
  • See where you stand. Billing & plans now shows a usage meter for each limit, turning amber as you approach it — so you'll never be surprised by a blocked run.
  • Your demo is still unlimited. While your 30-day demo is running there are no caps at all; your plan's limits apply once it ends.
  • Nothing you already built is taken away. If you're over a limit after a trial ends or a downgrade, you keep everything — you just can't add more until you upgrade or free something up.
v0.15.9 · 2026-07-24

New Pro plan — go solo on the cloud, scheduling & AI for $29/mo

  • Pro sits between Starter and Team. A single-seat plan at $29/mo (or $290/yr) that unlocks the cloud portal, up to 2 agents, schedules & headless runs, AI test generation and self-healing selectors — everything the free Starter plan doesn't include — without stepping up to Team's per-user pricing. Vision automation, team roles and SSO stay on Team and Enterprise.
  • Clearer pricing page. The pricing section now shows all four plans side by side with a Pro column in the full comparison table, so it's obvious what you get at each step.
v0.15.8 · 2026-07-14

AI fixes your flaky steps — heal a selector or switch to Vision

  • Suggest fix (AI). Select any step in the Designer and click Suggest fix (AI). The AI looks at the step and your screen and tells you, in plain English, how to make it reliable — either a sturdier selector or switching the step to Vision (clicking by what it looks like) when there's nothing stable to grab, like a video or live camera feed. You see exactly what will change before you apply it.
  • Automatic help when a run fails. If a step fails during a local run, the Designer now offers the fix automatically as soon as the run finishes — no hunting for the broken step.
v0.15.6 · 2026-07-13

Get one email with your whole suite's results — screenshots and all

  • Suite summary emails. When you use Run all on your Test Cases, you can now get a single email once the whole batch finishes — a pass/fail count up top, then every test case with its result. For each failure it shows the exact step that failed and why, with the screenshot embedded right in the email so you can see what went wrong without leaving your inbox.
  • Turn it on in one click. Enable it under Settings → Notifications ("Email a summary after each suite run"). It uses the same recipient list as failure alerts.
v0.15.5 · 2026-07-13

Environments: run the same test against dev, staging or prod

  • Define named variable sets. Under the new Environments page, an admin can set up Staging, Production and any others — each holding key = value pairs like a base URL, credentials or API key. Values can be marked secret so they're masked in the app and never returned by the API.
  • Pick one at run time. When you run a test case, choose an environment from the run toolbar and its variables flow into your ${variable} references — so one test suite covers every environment without editing steps.
v0.15.4 · 2026-07-13

CI/CD: JUnit test reports

  • Run your tests in your pipeline and get native reports. A CI run's results can now be downloaded as JUnit XML, the format Jenkins, GitHub Actions, GitLab and Azure DevOps understand out of the box. Create a CI token on the Git sync page, trigger a run, then fetch /api/ci/runs/<runId>/junit.xml — your pipeline shows pass/fail per test case like any other test suite.
v0.15.1 · 2026-07-13

Get emailed when a test fails

  • Run failure notifications. Turn on email alerts so you hear about failures the moment they happen — no need to watch the Runs page. Enable it under Settings → Notifications and add the people who should be notified (or leave it to the workspace owner). Each email shows what failed, the last output, and a link straight to the run.
v0.15.0 · 2026-07-10

New: API testing

  • Test your APIs, not just your UI. Add an HTTP Request step to send a request (GET/POST/PUT/PATCH/DELETE) with headers, body and authentication, then assert on the response — status code, response time, headers, or any value in the JSON (via JSONPath).
  • Chain requests. Capture a value from one response (e.g. a login token) into a variable and reuse it as ${token} in later steps — API or UI.
  • Runs anywhere. API tests need no browser or screen, so they run fast on any agent (even headless) and locally in the Designer. Filter your Test Cases by the new API type.
v0.14.23 · 2026-07-09

Fix: recording cursor ring stays on the pointer at any screen scaling

  • The blue “recording” ring no longer drifts away from your mouse. On displays set above 100% scaling (e.g. 125% or 150%), the ring could appear offset from the real cursor — worse the further you moved from the top-left — which made recording feel unstable. The ring now tracks the pointer exactly at any display scaling.
v0.14.21 · 2026-07-08

Install the portal as an app on your phone

  • Add TITA to your home screen. The portal is now an installable app — open it on your phone and choose “Add to Home Screen” (or “Install app” in desktop Chrome/Edge) to launch it full-screen with its own icon, no browser bar. Run and monitor your automations like a native app.
v0.14.20 · 2026-07-08

The portal now works great on a phone

  • Menus, lists and dialogs are now mobile-friendly. A tap-to-open navigation drawer replaces the sidebar on small screens, the Runs, Test Cases and Schedules lists show as easy-to-read stacked cards instead of scrolling sideways, and pop-up forms (create project, invite a teammate) fit the screen with full-width buttons. Manage your automations from anywhere.
v0.14.9 · 2026-06-29

Fix: recorder no longer captures read-only fields

  • The web recorder was recording read-only/disabled fields you never typed into. It now only captures fields you can actually edit, so recordings match what you did.
v0.14.8 · 2026-06-29

Fix: Schedules page shows the real next-run time

  • The Next Run column said “just now” instead of when the run is actually due. It now shows the time correctly (e.g. “in 58 minutes”).
v0.14.7 · 2026-06-29

Fix: cloud agents now connect after sign-in

  • Fixed agents showing Offline after a fresh install. A sign-in bug pointed the agent at a local address instead of the cloud, so it never appeared online. Update, sign in, and your agent connects to the cloud automatically.
v0.14.6 · 2026-06-29

Fix: stuck "Queued" runs and stale agent status

  • Failed runs no longer get stuck on "Queued." A run that failed and was retried could sit Queued forever; retries now run and resolve correctly.
  • Agent status is honest now. An agent that stops is marked Offline instead of showing Online forever, so the Agents page reflects reality.
v0.14.5 · 2026-06-28

Fix: portal runs now start + Stop button per test case

  • Fixed “Any available agent” runs that never started. Runs queued from the portal to any agent could sit waiting forever; they now route correctly to an online agent.
  • Stop a single run. After you click Run on a test case, the button becomes “Stop” so you can cancel that run without leaving the page.
v0.14.4 · 2026-06-28

Heads-up before running desktop tests

  • The Test Cases page now warns you when desktop tests are in view. Desktop tests take over the agent’s screen, mouse and keyboard — so run them on a dedicated agent, not the PC you’re using. (“Headless” only applies to web tests; desktop automation always runs visibly.)
v0.14.3 · 2026-06-28

Filter test cases by Web or Desktop

  • The Test Cases page now has a Type filter (All / Web / Desktop). Show — and “Run all” — just your web tests or just your desktop tests. Each test case’s type is detected automatically from its workflow.
v0.14.2 · 2026-06-28

Fix: desktop recorder stops when you close the recorded app

  • Closing the app you’re recording now ends the recording and brings the Designer back up. Previously, closing the target app could leave the recorder running and the Designer minimized. The recorder now tracks the app you’re working in and finishes cleanly when it closes.
v0.14.1 · 2026-06-27

Fix: Designer now shows the translated text

  • Resolved a display bug where the desktop Designer showed internal labels (like “Login_SignInButton”) instead of the actual wording. All 2,052 strings across English and the six other languages now render correctly. Update to get the fix.
v0.14.0 · 2026-06-26

TITA now speaks 7 languages

  • The whole platform is now available in English, Spanish, French, German, Portuguese, Dutch and Chinese (简体中文). Pick your language and the interface switches instantly.
  • Cloud portal: every page, table, dialog, menu and the in-app Guide are translated — switch language from the language menu in the header (or on the login screen).
  • Desktop Designer: the full Windows app — menus, ribbon, panels, dialogs and messages — is localized. Change it live under Welcome → Settings → Language, no restart needed.
  • This website is translated too — use the language picker in the top nav.
v0.13.13 · 2026-06-10

No more phantom "update available" loop

  • The desktop app only prompts to update when there's actually a newer installer. Previously it could nag about a version the Download page didn't have yet, sending you in circles. The update check now follows the downloadable installer version.
v0.13.12 · 2026-06-10

The update popup's download button works again

  • "Download now" downloads. The update reminder's button wasn't triggering the download; it now links straight to the installer, so one click gets the new build.
v0.13.11 · 2026-06-10

Web recording captures everything, and Studio returns to front

  • Closing the recording browser reliably ends the recording. Before, if you had other Chrome windows open, closing the recording window didn't stop the recorder — so later steps (password, login, a camera click) were lost and Studio stayed minimized. Recording now uses a dedicated, monitored window, so closing it stops cleanly, saves every step, and brings Studio back to the front.
v0.13.10 · 2026-06-10

Web recordings drop stray coordinate clicks

  • No more phantom clicks that hit the wrong spot on replay. A stray coordinate click could sneak into a browser recording and, on a differently-sized window, land on the Log In button before the username was typed. Browser recording now relies entirely on the extension's precise selectors, so ordinary clicks are recorded as stable css/xpath — not coordinates.
v0.13.9 · 2026-06-09

Cleaner web recordings — no duplicate clicks

  • Browser recordings no longer double up each click. With the extension installed, Studio sometimes recorded both a coordinate click and the real selector click for the same press, which looked jumbled and out of order. Each click now produces a single, clean selector step.
v0.13.8 · 2026-06-09

Recordings stay in their own test case

  • A new recording no longer merges into whatever test case you had open. Studio showed recorded steps in the open workflow as you captured them, which meant they appended onto an unrelated test case. Recording now starts a clean draft when a workflow is already open, so your capture stays separate and saves to the folder you choose.
v0.13.7 · 2026-06-09

A smarter "update available" reminder

  • You're reminded to update only when you actually are behind. When you sign in, if a machine in your workspace is running an older app build than the latest download, the portal shows a one-time popup with a download link. It's driven by the version your agent reports, so workspaces already on the latest build see nothing, and the reminder clears itself once you update — no more banner nagging people who are already current.
  • Studio run bar fits the window again. A long "no agents online" hint used to stretch the run-controls row and push the Capabilities / Retries / Input fields off-screen; that hint now wraps on its own line.
  • Recordings no longer break on dynamic element IDs. The recorder used to pin a click to an element's id even when that id was a throwaway per-session value (a GUID or hash) — so the step worked once and failed on the next login. It now recognizes auto-generated ids and chooses a stable selector instead. Re-record affected steps to pick up the change.
  • Web-recording extension now installs with the app. The browser extension that captures your clicks used to be developer-only; the installer now ships it, and Studio has a one-click "Browser Extension" button that opens the folder and your browser's extensions page with load instructions.
v0.13.6 · 2026-06-09

Your running agent shows up on the Agents page

  • Agents that are actually running now appear. The Agents page was reading an old registry that a cloud-connected agent never writes to — so it said "No agents registered" even while your agent was happily passing runs. The Agents page, the dashboard count, and the run-target picker now read the registry your agent really reports to.
  • Host name and version, at a glance. Each agent now reports its machine name and build on every heartbeat, so the fleet view shows which machine is online and which version it's running.
v0.13.5 · 2026-06-09

Opening a workflow shows your whole project

  • Focus mode is off by default. Opening a workflow from a recent item used to narrow the Test Cases tree to just that workflow's folder (with a "Focused on…" banner and a "Show all" button). That hid your other folders and was why new folders seemed to disappear. Now opening a workflow shows the entire project — no banner, no extra click.
v0.13.4 · 2026-06-09

New test-case folders stay put

  • Folders you create no longer vanish. When the Studio is opened from a recent workflow it focuses the Test Cases tree on that workflow's folder. Creating a new folder briefly showed it, then a background refresh filtered the empty folder back out — so it flashed in and disappeared. Folders you create while focused now stay visible.
v0.13.2 · 2026-06-08

Why runs get stuck "Queued"

  • The Studio now explains why there are "no agents." A test run needs an online agent to execute it — without one, runs sit in "Queued" forever. Previously the Studio just showed "No agents loaded yet" with no reason. It now tells you exactly what went wrong (the agent runtime is missing and you should reinstall, no agent token was issued for your plan, the agent couldn't start, …) and what to do next.
  • Run capability now matches your workflow automatically. A desktop workflow used to default to requesting a "web" agent, which could route it to the wrong place. The required capability is now derived from the workflow itself — desktop, web, or both.
v0.13.1 · 2026-06-08

Desktop click reliability + owner access fixes

  • More reliable desktop clicks (new installer). Coordinate-based clicks now adapt to the window size, so a recording made at one screen resolution no longer misses when run on a different machine — and a click that can't actually be performed now fails honestly instead of reporting success. Reinstall to get it (first desktop-app update since 0.5.0).
  • Fixed a permissions bug where the owner of a workspace was wrongly blocked from Agents, Workflows, Test Cases and Schedules (and from upgrading their plan). Owners now have full access to their workspace.
v0.13.0 · 2026-06-07

AI plans & efficiency

  • Built-in AI is now part of the Pro plan. The Studio AI assistant uses our hosted AI on Pro and above; on lower plans you can still use AI by adding your own Anthropic API key in the Assistant tab.
  • Faster, cheaper AI for the in-app help assistant and selector healing, with no meaningful quality drop — plus sensible usage caps so costs stay predictable.
v0.12.1 · 2026-06-07

Help assistant — daily limit

  • Added a sensible per-user daily message cap to the in-app help assistant to keep usage costs predictable.
v0.12.0 · 2026-06-07

In-app help assistant

  • Ask for help right inside the portal — a new chat assistant (the bubble in the bottom-right of every page) answers questions about using TITA: projects, agents, Git sync, two-factor auth, billing, and where to find things. It tailors answers to your plan and points you to human support for anything account-specific.
v0.11.1 · 2026-06-07

Git sync credential fixes

  • Stopped the browser from autofilling your portal login into the Git sync username and access token fields, and clarified that the token is a Git personal access token — not your portal password.
  • The Git username is now correctly required (it was mislabelled optional).
v0.11.0 · 2026-06-07

Change your password

  • Self-service password change — the Settings → Security page now lets you change your sign-in password directly from the portal: enter your current password and a new one. (It was previously a placeholder.)
v0.10.0 · 2026-06-07

CI triggers — data-driven runs & pipeline templates

  • Data-driven CI runs — when a CI pipeline triggers your tests, each data-driven test case now fans out into one real run per data set (just like in the Studio), each with its own pass/fail — so a single bad data set fails the gate.
  • Copy-paste pipeline templates — ready-made Jenkins, Gitea Actions, and GitHub Actions examples (plus a portable shell script) make wiring the “open PR → run tests → gate the merge” loop a copy-paste job.
v0.9.0 · 2026-06-07

Git integration — CI test triggers

  • Trigger tests from your CI pipeline — generate a per-project CI token (under Git sync → CI test triggers) and let Jenkins, Gitea Actions, or GitHub Actions kick off a headless test run and gate a pull request on the result. The pipeline triggers a run, polls until it finishes, and gets a PASS/FAIL. Runs execute for real on your connected agent, so it’s a genuine test gate — not a dry run.
v0.8.0 · 2026-06-07

Git integration — pull requests

  • Open a pull request instead of pushing directly — a new option under Git sync. When enabled, Push sends your test changes to a review branch and opens (or updates) a pull request against your target branch, so they go through your normal code review and CI before landing. Works with GitHub and self-hosted Gitea; the portal shows a direct link to the PR.
v0.7.0 · 2026-06-07

Git integration — two-way sync

  • Pull from Git — the Git integration is now two-way: Push sends a project's tests to your repo, Pull brings the repo's version back. A status banner shows which side is ahead, and pulling asks before discarding any unpushed local changes.
  • Tests are stored in a clean, human-readable format (named by test, not internal ids), so a push/pull round-trip keeps your Git history meaningful.
v0.6.0 · 2026-06-06

Git integration (tests-as-code) & onboarding

  • Git integration (Enterprise) — connect a project to your own Git repo (GitHub or self-hosted Gitea) and push its workflows & test cases as files on demand. Configure it under Git sync (provider, repo URL, branch, access token); your tests get versioned alongside your code.
  • New accounts create their own first project (with a friendly first-run welcome) instead of a pre-made one.
  • The marketing site is now fully mobile-friendly (hamburger menu, pricing that stacks on phones).
v0.5.0 · 2026-06-05

Security hardening & sign-in improvements

  • Two-factor authentication (2FA) — opt-in for portal sign-in, with QR enrollment and one-time backup codes (Settings › Security). Recommended for admins/owners.
  • Inactivity auto-logout — signed out after 60 minutes idle (with a warning), so unattended sessions don't stay open.
  • Show/hide password toggle on the portal and Studio sign-in.
  • SSO users now start at the read-only Viewer role; admins promote members from Settings › Team.
  • A clearer privacy page covering operator access and subprocessors.
  • Fix: Studio Data Manager “Update Selected” button label is no longer cut off.
v0.4.0 · 2026-06-04

Enterprise single sign-on (SSO)

  • Sign in with your company identity provider over OpenID Connect — Okta, Azure AD / Entra, Google Workspace, and other OIDC providers.
  • Configured per organization in Settings › Single sign-on (issuer, client ID/secret, email domain); the page shows the exact redirect URI to register at your IdP.
  • Users whose email matches your domain are routed to your IdP, with optional auto-provisioning and a default role on first sign-in.
  • Additive and safe — password sign-in is unchanged until an organization turns SSO on.
v0.3.0 · 2026-06-04

Project backup & restore

  • Back up a whole project — workflows, test cases & folders, and data sets — to a single .titabackup file from the Studio's Manage ribbon.
  • Restore a backup into a brand-new project (auto-created) or merge it into the current one.
v0.2.0 · 2026-06-04

Cloud-connected platform, per-user & annual pricing, team workflows

  • Cloud-connected app — Studio & the agent sign in to the cloud; the agent enrolls automatically on sign-in (no local engine to run).
  • Slim ~134 MB installer — Studio + agent only.
  • Auto-onboarding — registering creates your organization & first project; no license key needed.
  • Per-user pricing: Team $80/user/mo, plus a new annual option ($800/user/yr) with a Monthly/Annual toggle.
  • Projects page — create & manage multiple projects in the portal.
  • Team management — invite teammates by email with roles; one-click invite redemption.
  • Demo: the 30-day trial now unlocks every feature.
  • Fixes: Studio sign-in, billing plans loading, and installer startup.
v0.1.0 · Initial release

Record, run, and verify across a fleet

  • Studio with the App/Web Recorder for desktop and browser flows.
  • Workflows, test cases, folders, and input data sets per project.
  • Agent fleet execution — run on any agent or pin to a specific one.
  • Headless or visible runs (web tests headless for speed).
  • Cloud portal: dashboard, workflows, test cases, agents, schedules, and live runs.
  • Run detail with a step timeline, output logs, and captured screenshots.
  • One-click Windows installer bundling Studio, agent, and engine.

← Back to home