What's new at Atomicwork?

Available in sandboxes now. Coming soon to all accounts.

What it is — Workflows now have their own Automation section in the main navigation, instead of sitting inside Settings.

Every time a workflow runs, Atomicwork records what happened: which version ran, what triggered it, each action in order, and its result.

Why it matters — When an automation doesn’t do what you expected, you shouldn’t have to guess why. You can now open the run, see the exact step that failed and what it received, and retry it, all without leaving Atomicwork.

What’s new

• Easier to find. Workflows are grouped by how they start (event-based or scheduled) and by scope (workspace or organization). Workflows for custom objects appear here too.
• Run history for every workflow. See how many runs succeeded, failed or are still in progress, and click any of these counts to filter the list. A chart shows your success rate over time.
• A detailed view of each run. Every action is listed in order, with timestamps and the data that went in and came out. Retries are listed separately, and you can retry a failed action manually.
• Linked to the ticket. From a ticket, click View logs to open the run that acted on it. From the run, you can jump back to the ticket or the workflow.

How it works — Go to Automation → Workflows, choose a scope, open a workflow and select Run logs.

NOTE: Run history starts being saved the day this feature is enabled for your account. Runs from before will not be available.

Live since Oct 9, 2026

What it is — AI Assist is a built-in assistant for agents and admins. It opens in a panel next to the page you’re on and already knows what’s on that page. Ask in plain language and it looks things up, explains what you’re seeing, and creates or changes setup for you. It replaces Agent Assist.

Why it matters — Most Atomicwork tasks start with “where is that setting?” or “what’s going on with this ticket?”. AI Assist removes that step, so agents get to the fix faster and admins can set things up without learning every menu.

For agents

• On a request list: “What should I pick up first?” AI Assist ranks your queue, flags requests that are at risk or overdue, and spots duplicates.
• On a request: a summary of what the requester needs, what’s been done and what to do next. It also finds similar requests resolved before, with links.
• Triage: “What’s blocking this ticket?” points to the exact automation step that failed and shows its error. It can also suggest a priority and an assignee.
• Drafting: replies to requesters and updates for stakeholders on major incidents. AI Assist never sends anything for you.
• Incidents and problems: how many people an incident affects, workarounds from past incidents, a likely root cause, and a draft known-error article.

For admins

• Code Library: write, explain and fix automation scripts, and publish new versions.
• Custom objects: design object types and fields, find and update records, including bulk updates after you confirm.
• AI Workforce: create and configure AI coworkers, attach skills and tools, and answer questions like “which coworkers are near their budget?”
• Workflows: build workflows written in code from a plain-language description.

Ask anything about the platform. If AI Assist can do the task with your permissions, it does it. If not, it gives you the steps from the Atomicwork documentation, with the articles it used.

Built to be safe

• It acts as you. It can only see and change what your role allows, and changes appear in audit logs under your name.
• It asks before anything that’s hard to undo or that other people will notice, such as publishing, deleting, bulk updates or running a script.
• It saves drafts first, so you can review before anything goes live.
• It can’t be talked around your permissions. Instructions hidden in tickets, and claims like “my manager approved this”, are ignored.

How it works — Click AI assist in the page header. Pick a suggested prompt or type your own. Click the context tag above the chat box if you don’t want it to use the current page.

Read the full guide here

Available in sandboxes now. Coming soon to all accounts.

What it is — We’ve rebuilt the eight default reports: Requests overview, Service requests, Incidents, Changes, Request aging, Agent performance, Group performance and Satisfaction survey. Every metric now has one definition, and every number can be clicked to see the requests behind it.

Why it matters — A report is only useful if you trust the number and can act on it. When the reopen rate says 28%, you should be able to see those reopened requests, not export a spreadsheet and go looking.

What’s new

• One definition per metric. Time to resolution, SLA compliance and the other metrics are calculated the same way in every report, so numbers no longer differ between reports.
• View breakdown on every tile and chart. Click any number to see the exact requests behind it. You can add columns, filter further and export the list to CSV.
• More accurate resolution times. Time when the SLA clock was paused is no longer counted. If a request is reopened and resolved again, its total time is used.
• The same filters in every report. These include Follower, Tags and Created on, with options like is any of, between and in the last 7 days.
• Saved views. Save a filter combination and come back to the same analysis later.
• Fresher data. New requests appear in reports within an hour.

What you can answer now

• Is our backlog growing? Requests overview shows created against completed each day. If the created line stays above the completed line, the backlog is growing.
• Where are requests getting stuck? Request aging shows how long requests spend in each status. It also groups open requests by age, so you can see whether the delay is at intake or at resolution.
• How is each agent performing? Each request is credited to the agent who responded first, or to the agent assigned when it was resolved. Agents with five or fewer requests are hidden, so the comparison stays fair.
• What keeps breaking? The Incidents report lists the most and least reported incident forms and categories, alongside response times and how long incidents have been open.
• Are employees happy with the outcome? Satisfaction is now based on each workspace’s own survey scale (2, 3 or 5 points), and only the latest response to each request counts.

Good to know

• Some numbers will look different from the old reports. That’s expected: metrics are now calculated consistently, and paused SLA time no longer counts toward resolution time.
• Metrics count different dates depending on what they measure. New-request metrics count requests created in the period, and resolution metrics count requests resolved in it. Each report’s help article lists the date each metric uses.

How it works — Go to Analytics > Workspaces > [Workspace name] and open any report. Click a tile or chart to see the requests behind it.

Read more in the Requests overview, Request aging, Agent performance and Incidents report guides.

The AI Workforce is generally available

Live since Oct 9, 2026

AI coworkers are now available to every Atomicwork customer.

Since launch we’ve made the coworkers more robust and practically useful. We've added memory, scheduled tasks, actions labeled “on behalf of” the requesting user, a human reporting manager, your own MCP connections, a library of reusable scripts, a guided setup flow, AI usage shown on each ticket, and checks before each coworker goes live.

Why it matters — The controls you’d expect for any new hire (oversight, accountability, limits) are in place, so you can move AI coworkers from trial to production with confidence.
​

How it works — Go to AI Workforce on your nav pane to get started. Certified coworkers are already in your account as drafts, ready to review and publish.

A new People page you can set up your way

Live since Oct 14, 2026

What it is — The People page in Directory now works like the Requests and Changes lists. You can save views, filter and sort by any attribute, choose your columns and search from the toolbar.

Why it matters — In large organizations, finding a specific group of users meant scrolling through a long, fixed list. Now you can save a view once, such as all contractors in one country, and come back to it, share it with a teammate or export it in a click.

How it works — Go to Directory → People, add filters and save them as a view. Use the view’s menu to share, duplicate or export it to CSV. You can also invite several users at once from Add, and deactivate a user directly from the list.

Replying and commenting, tidied up

A few quality-of-life updates for replying and commenting on requests:

• Switching between reply and private note keeps your text and attachments.
• A private note no longer carries over the reply’s cc and bcc.
• Comments keep their formatting, including numbered lists.
• Each detail panel now clearly shows when it’s loading, has failed or is empty.
• Empty fields show a clear prompt, such as “Assign agent” or “Set due date”.
• Assistant answers use headings and lists, so they’re quicker to scan.
• Custom domain setup is now one row on the settings page.

Why it matters — Fewer lost drafts and fewer “did that load?” moments in the screens you use most.
​

How it works — Default setting. No setup needed.

Problem assessment, navigation and editor polish

More QoL updates:

• Problem assessment answers can be edited on the problem itself and save automatically.
• Knowledge, workflows, apps, skills and the MCP Store now share the same left navigation, and you can open folders from the keyboard.
• AI coworker cards show spend as usage, and run times are written out in full.
• List views always show ID and Subject.
• The portal logo now shows in dark mode.

Why it matters — Each fix is small, but together they make everyday work faster and more predictable.
​

How it works — No setup needed.

A new layout for request detail pages

Available in sandbox. Coming soon to production environments.

What it is — Service requests, incidents, problems and changes now share one layout. The subject, description and key fields come first. Approvals, linked assets and AI usage sit in collapsible side panels.
​

Why it matters — When an incident leads to a problem and then a change, every page is laid out the same way, so nobody has to relearn where things are halfway through.

​How it works — No setup needed. Collapse or expand a side panel once and it stays that way on the next ticket you open.

---

Category, type and item classification for every request type

Live since Sep 29, 2026

What it is — Subcategory is now called Type, and there’s a new Item level under it, for example Applications → Crashes → Outlook. This is the standard Category-Type-Item (CTI) model used in IT service management, and it now covers service requests, general requests and changes as well as incidents and problems.

Why it matters — More precise classification makes routing more accurate and reports more meaningful.

How it works — Admins choose which request types use it, and can change that at any time. Existing reports and filters that use Subcategory will now show Type.

---

A comment thread on every step of a change

Live since Sep 29, 2026

​What it is — Each task, approval and form in a change now has its own comment thread, with public replies, private notes, attachments and @mentions.
​

Why it matters — Discussion stays on the step it’s about. You can also require a comment before a step can be completed, which gives auditors a clear record of what was done.

​How it works — Turn comments on per action group or per action in the change template builder, and set each one to Optional or Required. Existing templates don’t change until you do.

---

Bulk close tickets that need fields filled in before closing

Live since Sep 29, 2026

What it is — Bulk close no longer skips tickets that need fields filled in before closing. You fill in the fields once and the whole selection closes.

Why it matters — No more closing those tickets one by one. Your values only fill empty fields and never overwrite existing ones.

How it works — Select the tickets, choose Change status → Closed, and fill in the required fields when asked.

---

Approval workflows can now branch on “cancelled”

Live since Sep 24, 2026

​What it is — Approval steps in a workflow now have three outcomes: approved, declined and cancelled.
​

Why it matters — If a request is withdrawn while an approval is pending, the workflow takes its own path instead of waiting forever or going down the wrong one.

​How it works — In the workflow builder, add actions under the new Cancelled branch of any Send for approval step.

---

Reply and change status in one click

Live since Sep 18, 2026

​What it is — The Send button now has a “Send and set as” option. One click posts your reply or private note and moves the ticket to its new status.
​

Why it matters — Tickets no longer sit in the wrong status because someone forgot to update it after replying, so SLA and reporting numbers are more accurate.

​How it works — Click the arrow next to Send and pick a status. If that status needs resolution fields, you’re asked for them first.

---

Atom creates tickets from the conversation, no form needed

Live since Sep 15, 2026

​What it is — Atom can now create a ticket from your conversation with it. Nobody has to open a form, fill it in and submit it.

Why it matters — One less step between asking for help and getting a ticket created.
​

How it works — Contact your account team to turn it on. A setting to turn it on yourself is coming.

---

Easier-to-follow reasoning from Atom

Live since Sep 15, 2026

What it is — When Atom works through a task as an AI coworker, its step-by-step reasoning is now cleaner and easier to follow in Slack, Teams and the portal.

Why it matters — You can see what Atom is doing at each step without the noise.

How it works — No setup needed.

---

Restart a stage when a change approval is declined

Live since Sep 9, 2026

What it is — When an approval in a change is declined, you can add actions, restart the current stage or go back to the previous one.

Why it matters — A declined approval no longer leaves the change stuck. The requester and assignee are notified, and every restart is recorded in Activities.

How it works — With Restart a Stage, you now have more control over what happens when an approval action is declined. You can:

• Add additional actions
• Restart the current stage
• Restart the previous stage

Existing change templates need to be republished once after editing before the feature works on them.

---

Mention and follow agent groups

Live since Sep 9, 2026

​What it is — You can now @mention an agent group, or add it as a follower on a ticket, alongside users and AI coworkers.
​

Why it matters — Bring a whole team into a ticket in one step. Members still get only one notification each.

​How it works — Type @ followed by the group name in a reply, or add the group from the Follow menu on the ticket.

---

Connect MCP servers with less setup, and use Atomicwork MCP from Claude Code and Codex

Live since Sep 8, 2026

​What it is — Bring Your Own MCP now supports CIMD (Client ID Metadata Documents). If an MCP server supports CIMD, you can connect to it without registering a client or entering a client ID and secret.

Atomicwork MCP also supports OAuth now, so you can connect from Claude Code or Codex by signing in with your Atomicwork account. You don’t need an API token.
​

Why it matters — Connecting a new MCP server used to mean registering a client and copying credentials around. Now it takes a URL and a sign-in.

​How it works —
​To connect a CIMD-compatible server to Atomicwork:

1. Go to AI Workforce → MCP store → Add MCP server and paste the server URL.
2. Atomicwork reads the server’s configuration, selects OAuth 2.0 (Client ID Metadata Document) and fills in the OAuth endpoints.
3. Add a name and category, click Connect, and sign in to the server.
4. Once connected, the server’s tools appear in the list. They’re off by default. Turn on the ones you want and publish a policy to make them available to AI Coworkers.

To use Atomicwork MCP from Claude Code or Codex:

1. Add the server using your tenant URL, https://<your-tenant>.atomicwork.com/mcp, with no auth header:

• Claude Code: claude mcp add --transport http Atomicwork https://<your-tenant>.atomicwork.com/mcp, then run /mcp
• Codex: codex mcp add Atomicwork --url https://<your-tenant>.atomicwork.com/mcp

1. A browser tab opens. Sign in to Atomicwork and click Allow Access.
2. Go back to Claude Code or Codex. Atomicwork’s tools are ready to use.

---

Find past Atom conversations in Slack

Live since Sep 3, 2026

What it is — Your earlier conversations with Atom are now under Agents & Tools in Slack’s left sidebar.

Why it matters — No more scrolling through DMs to find an earlier answer from Atom.

How it works — If you don’t see Agents & Tools, click on More in the sidebar to find the option. Only conversations started after September 3, 2026 will appear under Agents & Tools.

---

Start a workflow from a user event or on a schedule

Live since Sep 2, 2026

​What it is — Workflows can now start from user events (a user is created, updated or removed) or on a schedule that runs against a filtered set of users.
​

Why it matters — Onboarding checklists can start the moment a new hire’s record is created, and you can send announcements to one part of the org or repeat reminders until someone acts, with no manual follow-up.

​How it works — Same workflow builder and actions. Choose User as the trigger source.

---

A round of smaller improvements:

• Request creation opens in a central window with one searchable catalog picker.
• Canned responses preview instantly.
• The editor adds underline, strikethrough and blockquote.
• Fields show type icons, and hovering a ticket ID shows a preview.
• Security, Request, Incident, Problem and Change settings are each one page.
• Long dropdown lists are now searchable, and menus work with screen readers.
• Convert, move, clone and promote open in the same kind of window as create.
• On mobile, the reply editor’s toolbar and Send button stay visible when the keyboard is open.

Each change is small, but together they take friction out of the things you do all day.

Managing users often means piecing together access, ownership, and history from several different screens. We've put it all in one place.

What's new:

Workspace-level visibility — Workspace agents and admins can now view user profiles within their own workspaces, with information and tabs scoped to what's relevant to them.

Configurations — Workspace admins and org admins can see a user's platform access, the approval policies they're part of, and the applications they own, directly from their profile.

User Activities — Every change to a user's profile is now available to admins in one place: role updates, group changes, and other edits, with a clear history of what happened and when.

No more switching browser tabs to answer "what does this person have access to" or "what changed on this account." It's all on the profile.

Import custom object records via CSV

Live since Aug 20, 2026

What it is — You can now bulk-import records into a custom object using a CSV file, instead of creating them one at a time.

Why it matters — Custom objects are only useful once the data's actually in them. CSV import means you're not stuck manually re-entering existing data to get started.

How it works — From any custom object, use Import → CSV.

---

Agents now see their own tickets the way requesters do

Live since Aug 19, 2026

What it is — When an agent files a ticket for themselves — a laptop request, an access ask — they now view and interact with it through a requester-only lens, separate from the agent tools they use on tickets they're assigned to work.

Why it matters — Agents are IT staff, but when they're the one asking for something, they shouldn't have to sift through agent-only tooling meant for tickets they're resolving, not filing.

How it works — Applies automatically as of Aug 19.

---

ADP joins the list of supported user sync sources

Live since Aug 19, 2026

What it is — Atomicwork now supports ADP as a source for syncing user identity data, alongside Azure AD, Okta, JumpCloud, Workday, Rippling, and Slack.

Why it matters — If ADP is your system of record for employee data, you can keep Atomicwork user records current without a manual import or a workaround through another connected tool.

How it works — Connect ADP under Settings → Integrations → User Sync.

---

MCP Store: connected-instance view, CrowdStrike server, and easier browsing

Live since Aug 19, 2026

What it is — The MCP Store now shows which MCP servers are already connected to your instance, adds a new CrowdStrike MCP server, displays provenance badges on each listing, and supports sorting by category.

Why it matters — As the number of available MCP servers grows, finding the right one and knowing what's already connected matters as much as the servers themselves.

How it works — Browse the MCP Store under Settings → Integrations → MCP Store.

---

Sync and manage JumpCloud devices as assets

Live since Aug 19, 2026

What it is — JumpCloud is now supported as an asset source — device records sync automatically, and you can trigger remote actions on JumpCloud-managed devices directly from Atomicwork.

Why it matters — If JumpCloud manages your device fleet, this closes the gap between what JumpCloud knows about a device and what your service desk can act on.

How it works — Connect JumpCloud under Settings → Integrations → Assets.

---

Asset details, reorganized

Live since Aug 19, 2026

What it is — The asset details page moved to a full-page layout, with a new Linked Requests tab showing every request tied to that asset, and actions consolidated into one place instead of scattered across the page.

Why it matters — If you're troubleshooting a device, seeing every request linked to it in one tab beats piecing it together from memory or search.

How it works — Live as of Aug 19. No setup needed.

---

Views, filters, sort, and columns now available on every list

Live since Aug 18, 2026

What it is — Assigned to me, Requested by me, Mentioned, and Followed lists now support the same Views, Filters, Sort, and Columns controls available elsewhere in the product.

Why it matters — These lists get checked constantly, and they were the last ones missing the controls you already rely on everywhere else.

How it works — No setup required — the controls appear automatically.

---

Clone more than just requests

Live since Aug 18, 2026

What it is — Cloning now extends to change templates, catalogs, custom forms, SLAs, business hours, approval policies, global approval policies, and verified answers.

Why it matters — Building configuration from scratch every time you need something close to an existing setup wastes time. Now you can start from a copy instead.

How it works — Look for the Clone action on any of the object types listed above.

---

A Kanban board view for requests

Live since Aug 18, 2026

What it is — Requests can now be viewed and managed as a Kanban board, alongside the existing list view.

Why it matters — Some teams think in stages, not rows. This gives them a view that matches how they actually work a queue.

How it works — Switch to Kanban from the view selector on any request list.

---

Your AI coworkers just got more capable — and more manageable

Live since Aug 17, 2026

What it is — A set of upgrades to Atomicwork's AI workforce: coworkers can be scheduled to wake up and run tasks, actions they take are labeled "on behalf of" the requesting user, each coworker gets a reporting manager for oversight, they support bring-your-own MCP connections, retain memory across conversations, support code-based tool controls, and admins can choose which model powers a coworker — including Gemini and Fusion.

Why it matters — As AI coworkers take on more real work, you need the same basics you'd expect from any hire: oversight, accountability, and control over how they operate. This gives admins the governance layer that makes scaling AI coworkers safer, without slowing down what they can do.

How it works — Configure scheduled wake-ups, reporting managers, and model selection from AI Coworker settings. Memory and "on behalf of" labeling apply automatically once enabled.

---

Atom's Slack conversations now live where you already are

Live since Aug 11, 2026

What it is — Atom conversations that used to branch off into a separate thread now continue in your existing DM thread with Atom, the same way a conversation with a colleague would.

Why it matters — Jumping between threads to keep a conversation going added friction that had nothing to do with getting your issue resolved. Keeping it in one thread means less hunting for context.

How it works — Applies automatically to all Slack-connected workspaces. No setup needed.

---

Cleaner sync log, manual sync, and a persistent footer on source pages

Live since Aug 11, 2026

What it is — Source sync pages — for user, asset, and knowledge sources — got a cleaner sync log, an on-demand manual sync option, and a persistent footer so key actions stay visible while you scroll.

Why it matters — Small friction in a page you check often adds up. This makes checking sync status and triggering a resync faster.

How it works — No setup required.

---

Jump between custom object config and records faster

Live since Aug 4, 2026

What it is — Custom objects now have a quicker path between the configuration screen and the records themselves, instead of navigating through separate menus.

Why it matters — If you're actively building out a custom object, bouncing between "define the fields" and "see the data" used to cost more clicks than it should.

How it works — No setup required.

---

Three new self-hosted MCP servers: Azure Cloud Ops, Jamf, and Okta

Live since Jul 31, 2026

What it is — Self-hosted MCP servers for Azure Cloud Ops, Jamf, and Okta, so you can connect these systems to any MCP-compliant AI client, including Atom.

Why it matters — MCP is becoming the standard way AI tools connect to external systems. These three servers let your AI assistant act on cloud infrastructure, device management, and identity directly, without switching tools.

How it works — Hosting and connection steps vary by server.

---

New workspaces now sync users from multiple sources by default

Live since Jul 30, 2026

What it is — Every new Atomicwork tenant is now provisioned with multi-source user sync enabled by default, so a single user profile can pull from more than one identity source out of the box.

Why it matters — Most IT teams don't run identity out of a single system. Defaulting to multi-source sync means new workspaces don't need extra configuration to reflect that.

How it works — No setup needed for new tenants. Existing tenants can enable this manually — see "Sync one user profile from multiple sources" below.

---

Bring your Guru knowledge base into Atomicwork

Live since Jul 28, 2026

What it is — A new integration that connects Guru as a knowledge source, surfacing Guru cards alongside your other knowledge content.

Why it matters — If your team already maintains documentation in Guru, you don't need to duplicate it in Atomicwork to make it available to agents and Atom.

How it works — Connect Guru under Settings → Knowledge → Sources.

---

Scope custom objects to specific workspaces

Live since Jul 21, 2026

What it is — Custom objects can now be scoped to a specific workspace instead of applying globally across your entire instance.

Why it matters — Not every custom object belongs everywhere. Scoping keeps a workspace's data model relevant to that workspace, instead of exposing objects meant for one team to everyone.

How it works — Set the workspace scope when creating or editing a custom object.

---

Control how Atom responds, channel by channel

Live since Jul 16, 2026

What it is — Admins can now set one of three response modes for Atom on a per-Slack-channel basis, instead of one setting for the whole workspace.

Why it matters — How much you want Atom jumping into a conversation in your IT help channel is different from how much you want it in a general team channel. This gives you that control.

How it works — Set the response mode under the channel's Atom settings in Slack.

---

Atom's reasoning in Slack is easier to follow

Live since Jul 16, 2026

What it is — When Atom is working through a response in Slack, its reasoning now streams in a more consolidated way instead of appearing as scattered fragments.

Why it matters — Watching an answer come together shouldn't feel choppier than just waiting for the final response.

How it works — No setup required.

---

Approve and reply to Atom, right inside Teams

Live since Jul 14, 2026

What it is — Approvals and replies in Microsoft Teams no longer require a trip to the Atomicwork portal — you can now handle them inline, in the same message thread.

Why it matters — Every approval that requires switching apps is a chance for it to sit unread. Keeping approvals inline means faster turnaround on access requests, changes, and other approval-gated work.

How it works — Applies automatically to Teams-connected workspaces. No setup needed.

---

Sync one user profile from multiple sources

Live since Jul 10, 2026

What it is — A single user profile can now pull data from more than one sync source — for example, Azure AD for identity and Workday for org data — instead of being limited to one source per user.

Why it matters — Most IT teams don't keep all employee data in one system. This lets Atomicwork reflect that instead of forcing a single source of truth that doesn't exist in reality.

How it works — Configure additional sync sources under Settings → User Sync.

---

Custom objects are now live on all accounts

Live since Jul 2, 2026 (rolled out to sandboxes Jun 17)

What it is — Custom objects — a way to model data structures specific to your business, beyond the built-in request, asset, and user types — is now live on all production accounts.

Why it matters — Not every process fits neatly into a ticket or an asset. Custom objects let you define your own — software licenses, vendor contracts, whatever your workflows need — and build views, workflows, and automations around them the same way you would for any built-in object.

How it works — Create your first custom object under Settings → Custom Objects.

---

Your user profile pages, rebuilt

Live since Jun 23, 2026 (enhanced Jul 8)

What it is — The user profile page had a full redesign in late June, followed by a round of enhancements on Jul 8 — reorganized information, updated navigation, and clearer access to related requests and assets.

Why it matters — The old layout buried information agents and admins reference constantly — this reorganizes it around what you actually look up.

How it works — Nothing to configure. Give it a look next time you're checking a user's history.

---

The App Store is now organized by category

Live since Jun 17, 2026

What it is — The App Store now groups available integrations and apps by category instead of a single flat list.

Why it matters — Browsing for what you can connect is easier when things are grouped by what they do.

How it works — No setup required.

---

A new home for our developer docs

Live since Jun 16, 2026

What it is — An upgraded developer documentation site at developers.atomicwork.com, covering the API, webhooks, and now the MCP Server.

Why it matters — If you're building against Atomicwork's APIs or MCP Server, the docs are easier to navigate and more current than the previous version.

How it works — Visit developers.atomicwork.com.

---

Google Workspace and JumpCloud now supported as identity providers for IGA

Live since Jun 12, 2026

What it is — Atomicwork's Identity Governance & Administration (IGA) module now supports Google Workspace and JumpCloud as identity providers, alongside existing options.

Why it matters — If your access governance runs through Google Workspace or JumpCloud, you can bring that into IGA policies and entitlement reviews without a workaround.

How it works — Add Google Workspace or JumpCloud as an IDP under Settings → IGA → Identity Providers.

---

Notion is now a supported knowledge data source

Live since Jun 8, 2026

What it is — Notion pages and databases can now be connected as a knowledge source, surfacing alongside your other documentation.

Why it matters — If Notion is where your team actually keeps documentation, this saves you from re-hosting it elsewhere just to make it searchable in Atomicwork.

How it works — Connect Notion under Settings → Knowledge → Sources.

---

See how conversations turn into requests — and how many get deflected

Live since May 28, 2026

What it is — A new analytics view showing the flow from conversation to request, alongside usage and deflection metrics — how many issues Atom resolves without a ticket ever being created.

Why it matters — Deflection is one of the clearest signals that AI is actually reducing load on your team, not just adding another interface. This gives you the number, not just the feeling.

How it works — Available under Analytics → Conversations & Requests Flow.

---

Jamf, Kandji, and Intune assets now sync with their full attribute set

Live since May 27, 2026

What it is — Assets synced from Jamf, Kandji, or Intune now bring over their complete set of attributes, not a limited subset. If you use any of these integrations, you may see new fields populated on existing assets.

Why it matters — Partial data meant gaps in reporting and audits. Full ingestion closes that gap — but if you built filters or reports around the old limited field set, it's worth a quick check that they still reflect what you want.

How it works — Review any saved asset filters or reports if you rely on Jamf/Kandji/Intune data — new fields are now available to filter and report on.

---

User exports now include workspace-level roles

Live since May 26, 2026

What it is — Exporting your People/User list now includes each user's workspace-level role, not just their global role.

Why it matters — If you manage access across multiple workspaces, this was the detail missing from exports before — you had to check each workspace individually.

How it works — Export as usual from the People page — the new column appears automatically.

---

A few quality-of-life improvements to IGA

Live since May 20, 2026

What it is — You can now delete an application in IGA, jump straight from the Grants view to the linked request, and see a user's last-logged-in date.

Why it matters — These are small, but they're the kind of thing you reach for constantly during access reviews.

How it works — No setup required.

---

A more consistent layout across Analytics

Live since May 18, 2026

What it is — Analytics pages now use full-width layouts and a unified filter bar, matching the design used elsewhere in the product.

Why it matters — Switching between reports shouldn't mean relearning where the filters are each time.

How it works — No setup required.

---

Audit logs: time-based filtering, CSV export, and login/logout tracking

Live since May 12, 2026

What it is — Audit logs now support filtering by time range, exporting to CSV, and tracking login/logout events.

Why it matters — Audits and investigations need a specific window of activity, not the entire history — and being able to export it makes that usable outside the product too.

How it works — Apply a date range or export from the Audit Logs page.

---

Jump straight to the latest comment or activity

Live since May 7, 2026 (extended to change and problem records May 13)

What it is — A "Jump to Latest" control that takes you straight to the newest comment or activity on a request — now extended to change and problem records as well.

Why it matters — Scrolling through a long history to find the most recent update wastes time on tickets that have been open a while.

How it works — No setup required.

A UI update across Service Requests, Incidents, Changes, Problems, General Requests, and Major Incidents:

One status switcher — replaces the old "Mark Resolved" button. Agents and employees now view and switch between any status from the same place.

Cleaner reply box — your cursor starts above the signature, not buried below it.

Wider layout on large screens for easier reading.

Tidier details panel and tabs with icons for faster scanning.

​What it is — We're rolling out identifier-based deduplication for hardware assets. When more than one tool reports the same physical device — say Kandji and an Atomicwork Discovery scan both pick up the same MacBook — Atomicwork now merges them into a single asset record, with every contributing source listed on it.

Why it matters — Without this, every integration creates its own record for each device, so it shows up as two or three separate assets. That inflates your asset count, splits lifecycle history across records, and makes audits and license tracking harder than they should be.

Now you get one accurate record to assign, ticket, and report on, pulled together from Intune, Kandji, Jamf, Azure AD, Atomicwork Discovery, and manual/API/CSV additions.

How it works — Matching runs automatically, using shared hardware identifiers to confirm two records belong to the same device.

Note: If you're importing assets via CSV, import before connecting your integrations to avoid duplicate rows.

​Learn more
​

1. Bring your own MCP client
​
​What it is — The Atomicwork MCP Server lets you connect Atomicwork to any MCP-compliant client and build on our APIs.

Why it matters — MCP is becoming the standard way AI tools connect to external platforms. This means you're not limited to the Atomicwork interface to get things done.

You can ask your AI assistant, like Claude for instance, to pull up your open tickets, summarize what's pending, or even mark a completed request as resolved, all without switching tools.

How it works — Head over to the MCP documentation here to get started.

​​2. Sign in with JumpCloud SSO
​
​What it is — Atomicwork now supports SAML 2.0 single sign-on with JumpCloud. Your team signs in with their existing JumpCloud credentials – no separate Atomicwork password – and access is managed through JumpCloud groups.
​
​Why it matters — One less password to manage, one less place credentials can leak, and access control that lives where your identity policies already do.
​
​How it works — Admins can set it up in Settings → Security → JumpCloud SSO. Note that only one SSO provider can be active per tenant at a time.

3. A cleaner way to compose

What it is — A revamped composer across Request, Service Request, Incident, Change, and Problem detail pages – cleaner, denser, and faster to compose.

Why it matters — The old composer's fixed toolbar and stacked attachments ate up space and slowed you down mid-reply. This redesign clears that clutter without losing any functionality.

What's changed —

• Reply/Note is now a segmented control with icons instead of tabs.
• Formatting tools live in a floating bubble toolbar that appears when you select text.
• Attach, emoji, URL, request summary, and canned responses are consolidated into one '+' icon dropdown next to Ask Atom.
• Send is now an explicit labeled button.
• Broadcast sits next to Ask Atom in the bottom toolbar.
• Attachments scroll horizontally with an arrow-based navigation instead of stacking vertically.