Set savings as one dollar value per resolution
The Estimated savings dialog now asks for a single number: the dollars a resolved conversation is worth. On Analytics, choose Define savings (or Edit) on the Resolution card and enter it in Value per resolution.
- The agent’s Objective settings and Atlas home ask for the same value.
- The Time saved per resolution field is removed from the dialog.
- The Resolution rate and Deflection rate evaluations no longer show Time saved (hours) and Dollars saved with Add estimate in the evaluation editor.
Attach files to Atlas messages
You can now attach documents, spreadsheets and screenshots to Atlas messages, in the Atlas panel or on Copilot Home, and ask Atlas to build or answer from them.
- Attach up to five files per message with the attach button in the composer, by dropping them in, or by pasting.
- PDF, Word, Excel, CSV, text, Markdown, PNG, JPEG, GIF and WebP files are supported, with documents up to 10 MB and images up to 5 MB.
- Each file shows a chip that reads Reading, then ready, or failed with a reason.
- Atlas reads each file as text, and reads further into a long file when it needs to.
- Once you approve, Atlas can add a file to your knowledge base.
See what each Atlas write changed, step by step
A finished Atlas write now has View changes beside its row, opening a modal that shows each change in the reply as Before and After cards, so you can check the result without leaving the chat.
- A collapsed run shows View N changes in place of the single button.
- The modal steps through every change, such as Change 3 of 7, with Previous and Next or the arrow keys.
- A change made in another environment or agent has Open in that environment, which switches to it and opens the item.
Changes too large to keep say so instead of showing cards.
Keep Zendesk structure when importing a help center
A Zendesk import into your knowledge base can now keep the help center’s structure, filing articles in folders by language, category and section, the way Zendesk shows them.
- In Knowledge base, choose Add data source, then Zendesk.
- Under the folder picker, tick Preserve Zendesk structure.
Articles are filed inside the folder you selected, under a folder per language such as English (en-us), then its category, then its section and any nested sections. Importing again reuses the same folders.
Widget preview stays visible beside Atlas
With Atlas open, the Widget page now shows the widget preview beside Atlas instead of closing Atlas or leaving a blank column, so you can watch Atlas edit it.
Faster processing is now a per-model toggle in model settings
You can now switch Faster processing on or off for individual models: the agent’s default model, a playbook or a prompt.
- For the default model, open the model pill in the Agent page header and use the Faster processing toggle.
- For a playbook, open its Model settings and use the toggle below the other model settings.
- For a prompt, open its model settings and use the toggle below the other settings.
A playbook or prompt left unset inherits the default model’s value. A project that had Faster processing on in Behaviour keeps it until you turn it off.
The toggle appears only for models with a priority tier, and hovering its title shows the cost multiplier against standard. Evaluations and Tests simulations always run at the standard tier.
Knowledge base answers cite Zendesk help center pages
Your agent can now cite the public help center page of an article from a Zendesk data source as a source.
- With Show source URL(s) on in the knowledge base tool settings of a chat agent, a reply based on a Zendesk article lists that article’s help center page as a source.
- In the knowledge base, opening a Zendesk article or a Shopify product shows a read-only Source URL that opens the public page in a new tab.
Articles you imported before this change get a Source URL too.
Keys need new permissions for protected environments
You can now give a personal key or service key permission to merge into, publish and delete protected environments, or keep it from doing so. A key created with chosen permissions is now refused on those actions unless it carries the matching permission.
- In Workspace Settings > Personal keys (or Service keys) > New key, choose Custom under With these permissions to find the Protected environments group with Merge, Publish and Delete.
- Read & write now includes Merge and Publish.
Keys created before this change with chosen permissions don’t carry the new ones and must be recreated. Keys with no permission limits, and AI clients signed in to the MCP server with full workspace access, are unaffected.
Zendesk imports now apply your LLM chunking strategy
The LLM chunking strategy you pick when you import a data source from Zendesk is now applied to the imported articles, including on each scheduled refresh.
Import knowledge base documents as a single chunk
You can now keep a step-by-step guide, or any other document, whole in the knowledge base, so your agent retrieves it in one piece instead of in fragments.
- In Knowledge base > Add data sources, open LLM chunking strategy in the File, Plain text, Web page(s), Sitemap or Zendesk import and choose Keep each document in one chunk.
- Each document is stored as a single chunk; one longer than about 30,000 characters is split as usual.
- The option makes no LLM call.
The document create endpoints, such as Create text document and Create URL document, accept it as singleChunk.
Choose Alper, a Turkish ElevenLabs voice, for voice output
Your voice agent can now speak with Alper - High Quality Broadcast Voice, a male Turkish ElevenLabs voice in a broadcast style.
In Settings > Behaviour > Voice output, choose ElevenLabs as the provider, then pick Alper - High Quality Broadcast Voice from the Turkish voices. These options live with your agent’s Behaviour settings.
Atlas conversations keep running after you close the tab
Atlas replies now keep running on Voiceflow’s servers, so you can close or reload the tab in the Atlas drawer or on the Work with Atlas page and come back to the finished reply.
- Open a conversation that is still working in any tab, and the reply catches up and then streams live.
- In History, a conversation shows Running while it works and Needs you when it finished while you were away.
- The History button counts the conversations that need you.
- Stop, or Escape in the composer, ends the reply in every tab.
- A reply cut off by a restart ends with “Atlas was interrupted” and a Continue button.
Tests page shows only the current environment’s tests
The Tests page now shows only the tests and Past runs of the environment you are in, including when you switch environments quickly.
Manage workspace phone numbers through the API
You can now list, rename, assign, unassign and delete phone numbers in your workspace from the API, the MCP server or Atlas.
- Listing takes a
workspaceIDor aprojectIDand shows the agent, environment and version eachvoiceandsmstarget answers as. - Assign sets one target at a time,
voicefor calls orsmsfor texts, optionally pinned to an environment’s published or draft version. - Unassign clears one target or both and keeps the number in the workspace, while delete gives it up.
- Create buys a US or Canadian number by area code or connects a Twilio, Vonage or Telnyx number, from the API only.
Use a personal key or service key with the new Phone numbers permissions, set under Settings > Personal keys > New key > Permissions > Phone numbers.
Import sheets with empty trailing cells into the knowledge base
Table imports into your knowledge base now accept .xlsx and .xls sheets whose rows end in blank cells and read those columns as blank, but data in a column with no header is still refused.
Search prompts with Cmd or Ctrl+F on the Agent page
You can now find any word in a long prompt or a playbook’s instructions with Cmd or Ctrl+F, and reach Behaviour settings in a tab right beside the prompt.
- In the prompt card on the Global prompt or Instructions tab, or in a playbook’s instructions, press Cmd or Ctrl+F to highlight every match, with a count and next and previous buttons.
- Open the new Behaviour tab with Shift+3, where the search filters setting rows.
The Behaviour tab appears in the prompt card for agentic-framework projects. Conversation-flow projects keep Behaviour in the rail.
Atlas keeps working when closed and has a history column
Atlas keeps working when you close it mid-reply, and its launcher shows the result in a pill. Past conversations are easier to find, rename and come back to.
- Press Shift+A to toggle Atlas, or drag the launcher to redock it.
- On the Work with Atlas page, History opens as a resizable column. Right-click a conversation to Rename or Delete it, and unread conversations show a dot.
- Atlas questions step through with next, previous and Skip.
- When you run a test Atlas suggested, a feedback card lets you rate it and leave notes for Atlas.
See tool calls before each reply in transcripts
Transcripts now show which tools your agent called before each reply.
- A line above each agent reply names the tool it called, or what a system tool did, such as Searched the knowledge base, and shows in red when the call failed.
- Click a tool line to see its log, then choose Copy, Ask Atlas or Open in logs.
- Logs open as a resizable column on the left of the conversation.
- The Metadata and Evaluations panel can be hidden and resized.
- View options toggle navigation, tool calls, latency, input type and events.
- Previous conversations appear as a History row in Metadata.
Environment dialogs stop you acting on unseen changes
The environment dialogs for publishing, discarding, saving a version, pulling from Main and merging now keep their change preview as it loaded. They refuse to act on changes you have not reviewed when a teammate edits the same environment.
- In Publish, Discard changes, Save version and Pending changes, a Draft changed banner appears if someone edits the draft while the dialog is open. It names who made the last change and how long ago, and Publish and Discard changes stay disabled until you press Refresh.
- Pull from Main watches both drafts the same way.
- Merge shows a banner when either environment publishes, and stays disabled until you press Reload.
Atlas conversations start in Auto mode
New Atlas conversations open in Auto, so Atlas applies its edits without waiting for your approval.
- The mode button is at the left of the strip above the input, in the Atlas drawer or the Atlas page.
- Click it to switch to Manual, where every change waits for your approval.
- A reopened conversation keeps the mode it was saved with.
- The effort button on the right of the strip is separate and cycles Low, Medium and High.
Search tools, playbooks and workflows by name after /
Type / followed by part of a name to find a tool, playbook or workflow and insert a reference to it directly. This works in the global prompt and in playbook instructions.
- The menu shows up to five matching items under each category, with name matches first.
- Integration tools also match their provider’s name.
- MCP tools also match the name of their server.
Widget chat preview accepts photos and PDFs over 5 MB
The chat preview on the Widget page now accepts large phone photos and PDFs instead of failing with a file size error, when you attach them in a chat widget with file attachments allowed.
Edit chunking strategy for knowledge base URLs and sitemaps
You can now change the LLM chunking strategy of a URL data source, a sitemap or a sitemap page in the knowledge base after importing it, with no need to delete and re-import. The new strategy is used on the next sync.
- For a URL document or a sitemap page, open it in the knowledge base editor and use the LLM chunking strategy section between Metadata and the chunks.
- For a sitemap, the same dropdown comes first in its Import settings and sets the strategy for the pages it adds later.
Atlas opens the specific Settings page you ask about
When you ask Atlas where a setting lives, it now opens the exact page, such as Environments, Secrets or Settings > API keys, instead of stopping at Settings.