← What's new
June 2026
Improved
- Stala only schedules interviews in the right stage Stala now checks the candidate's current stage before booking. If that stage can't host interviews, she explains why and offers to move the candidate forward first — with your confirmation — before scheduling.
- Atlas can publish and unpublish jobs You can now ask Atlas to publish or unpublish a role to your careers site directly from the chat. Atlas confirms before making the change, and the job status updates on the page instantly — no refresh needed.
- Job status updates in real time When a job is published, unpublished, or its description is edited — by you, a teammate, or Atlas — the page updates automatically. If someone edits the job description while you have it open, a banner warns you before you can overwrite their work.
- Smarter suggested actions after each AI turn The action chips shown at the end of an AI conversation now reflect what actually happened and the real state of the job — so you won't see suggestions that don't apply. Chips now also appear on the job description, intake form, interview plan, and pitch kit editors.
- Connect LinkedIn from the Command Centre A new Connect your LinkedIn card in the Command Centre's Getting Started section lets you link your LinkedIn account in one click — without navigating to Settings.
Fixed
- Scorecard reminder email links to the right place The Submit Scorecard button in interviewer reminder emails now takes you directly to the candidate's interviews tab at the correct stage, instead of landing on the job page.
New
- Bulk shortlist, reject, or reset sourced candidates Select multiple sourced candidates and update their review status in one step — shortlist, reject, or move back to new. A bulk-action bar appears as soon as you have a selection, and if any updates fail you get a clear summary with the option to retry just those.
Improved
- Stala only schedules interviews in the right stage Stala now books interviews in the candidate's current stage only. If that stage can't host interviews, she explains why and offers to move the candidate forward to the next schedulable stage — with your confirmation — before booking.
- Credit top-up charges show a clear description AI credit purchases now appear in your billing dashboard with a readable label — including the credit amount and your organization's name — instead of a raw payment reference.
- Interviewer reminder emails show the correct time The scheduled interview time in feedback reminder emails now appears in your organization's date format and timezone — not as a raw UTC timestamp.
Fixed
- Scorecard reminder button links to the right place The Submit Scorecard button in interviewer reminder emails now links directly to the candidate's interviews tab at the correct stage, instead of the job page.
New
- Buy AI credits whenever you need them A new Buy Credits page lets you pick a quantity, see the live per-credit price (which steps down with volume), and pay securely. Credits appear in your balance automatically and never expire — they roll over month to month on top of your plan allowance.
Improved
- AI Settings split into Context and Credits tabs The AI settings page is now split into two tabs — AI Context for your organization's hiring context, and AI Credits for your balance, usage history, and credit purchases. Each is easier to find and navigate.
Fixed
- Horizontal overflow and mobile scrolling Long messages, links, and tables in chat no longer push the page sideways. On mobile, pages now scroll with natural momentum, the address bar collapses as you scroll, and nothing hides behind the home indicator at the bottom of the screen.
Improved
- Actionable hard-to-source banner with filter suggestions When a role is hard to source, the banner now tells you exactly which filters are limiting results and proposes specific changes — move these skills to nice-to-have, broaden the title — so you can widen the candidate pool in one click instead of guessing.
- Job description quality score on the job page A new quality chip on the job page scores your job description and flags findability gaps — so you can spot and fix issues before they affect how many candidates you see in sourcing.
Improved
- Publish nudge when your job is ready to go live When your interview plan is complete, a Ready to publish? prompt now appears on both the Job Description and Interview Plan tabs — so you can publish in one click from wherever you are. Jobs with a low-quality description also show a warning before you publish.
- Confirmation after adding a candidate After you add a candidate to a role, Atlast now tells you that an email has been sent to confirm their application and share a link to their portal — so you know they've been notified without having to check.
- Candidate signal legend is easier to read on mobile The candidate signal legend now opens as a full-width panel on phones instead of a cramped popover, making it much easier to read on smaller screens.
Fixed
- Job description quality score no longer gets stuck The quality score on a job description was stuck in a permanent "Checking quality…" state on every load. Scores now appear reliably, and any job affected by this issue resolves automatically the next time it's opened — no action needed.
- Dashboard scrolling is smooth again on mobile On phones, the dashboard was fighting the browser's address bar while scrolling, causing jank and cutting off content at the bottom of the screen. Scrolling is now smooth and nothing hides under the browser chrome.
Improved
- Publish nudge when your job is ready to go live When your interview plan is complete, a Ready to publish? prompt now appears on both the Job Description and Interview Plan tabs — so you can publish in one click from wherever you are, without navigating away.
- Warning before publishing a low-quality job description If your job description scores below the quality threshold, you'll now see a warning before publishing — giving you a chance to strengthen the post before candidates see it.
- Confirmation after adding a candidate After you add a candidate to a role, Atlast now tells you that an email has been sent to confirm their application and share a link to their portal — so you know they've been notified.
- Candidate signal legend is easier to read on mobile The candidate signal legend now opens as a full-width panel on smaller screens instead of a cramped popover, making it much easier to read on a phone.
Fixed
- Job description quality score no longer gets stuck Quality scores were stuck in a permanent "Checking quality…" state on every load, even when a score had already been computed. Scores now appear reliably, and any affected job resolves automatically the next time it's opened — no action needed.
New
- Job description quality score Every job description now shows a quality score (0–100) with a Strong / Good / Needs attention band. Click it to see a full breakdown — clarity, completeness, consistency, and more — plus specific suggestions to fix. The score updates live as you edit.
- Sourcing tells you when it broadened your search When your role's criteria are too narrow to match anyone, candidate search now automatically relaxes them and tells you exactly what it changed — so you understand why results look different and can decide whether to tighten things back up.
Improved
- AI-drafted job descriptions are more accurate AI drafts no longer pull in unrelated requirements for non-technical roles, flag contradictions like mismatched seniority and years of experience, and now reliably save the salary range you provide.
- Your message ratings persist across sessions Thumbs up and thumbs down ratings you give to AI agent replies now stay highlighted when you reopen a conversation — so you always know what feedback you've already left.
Improved
- Your message ratings stay visible across sessions Thumbs up or down ratings you give to AI agent replies now stay highlighted when you reopen a conversation — so you can always see what feedback you've already left without rating the same message twice.
New
- Rate AI responses with a thumbs up or down You can now give any Atlas agent reply a thumbs up or down — and add an optional note when something misses the mark. Your ratings help the Atlast team improve the agents over time.
- Send feedback any time from the Help menu A new Send Feedback option in the Help menu — and a floating button on every page — lets you share a bug report, idea, or general comment whenever something comes to mind. No need to leave the app.
Improved
- Candidate list and quick view work on mobile On phones, candidate cards now stack cleanly with labeled action buttons. The quick-view panel opens full-screen with the Reject and Advance buttons always in reach — nothing hidden behind the browser bar. Desktop layouts are unchanged.
Improved
- AI agent questions survive when you leave a session If an AI agent asks you to choose between options and you navigate away, the question is now waiting for you when you come back — so you can answer right where you left off instead of starting over.
- Sidebar status dot for agent activity A small dot next to Agent Sessions in the sidebar now shows you at a glance what's happening — amber and pulsing when an agent needs your input, blue while it's working, and red if something went wrong.
Improved
- Agent sessions stay scoped to the job you're editing When you open an AI agent from a job's description, intake form, interview plan, or pitch kit, it now reconnects to the conversation already working on that job — or starts a fresh one — instead of picking up an unrelated chat.
- "Needs you" status survives a page refresh The indicator showing which agent conversations are waiting on your input now persists across page refreshes. A status pill in the top bar and a toast notification surface any session that needs your attention, so nothing gets missed.
- Resume right where you left off Reopening an agent session now offers to take you straight back to the artifact it was working on — no need to navigate there manually before picking up the conversation.
- Pitch Kit section headings are properly formatted Pitch Kit sections now display correct, localised headings. Long artifact titles in a narrow panel no longer push their action buttons off-screen.
Improved
- Resume your last job or candidate when reopening an agent When you reopen a session with Atlas, Stala, or Talia, a banner now shows where you left off — click it to jump back to the job, candidate, or list you were working on without losing your current page. The context chip also shows the actual item name instead of a generic label.
New
- Run multiple AI agent sessions in parallel You can now run Atlas, Stala, and Talia sessions at the same time and switch between them freely. Starting a new session no longer interrupts one that's already working — each session keeps running in the background and saves on its own.
Improved
- Agent conversations are saved automatically Your conversations with Atlas, Stala, and Talia are now saved as each reply finishes — so nothing is lost if you navigate away, switch sessions, or your connection drops mid-response. Picking up where you left off now resumes the same session reliably.
Improved
- Candidates list is easier to scan A new legend explains what the match indicators mean, and each candidate row is cleaner at a glance. Hover over a candidate's match details to see the breakdown without an extra click, and use the new View Pipeline shortcut from the quick-view panel.
- Job picker shows status and ID When you select a job from a dropdown, each option now shows whether the role is published and its job ID — so you can tell similar-sounding roles apart without guessing.
New
- Filter prospects by industry and current company When sourcing, you can now narrow results by the industry of a prospect's current employer — require people working in fintech, exclude banking, and so on. You can also target or exclude specific companies by name, with logo-assisted autocomplete to tell similar names apart. Both filters save with your search and apply every time you re-run it.
Improved
- AI agents now know who you are When you ask an AI assistant "who am I?" or "check my availability", it now uses your actual account — not a guess based on a similar name. Stala also suggests real teammates as interviewers and can tell you when someone is free.
- Talia now calls sourced people "prospects" Talia consistently uses the term prospects for people you find through sourcing — not "candidates", which is reserved for people who've applied. Her action chips, introductions, and LinkedIn outreach all use the right term now.
Improved
- Include and exclude filters in Refine Search Every filter in the Refine Search form — role, sub-roles, skills, job-title keywords, and seniority — is now editable and supports both include and exclude. Say "must have React" and "never WordPress" in the same search, without two separate passes.
Fixed
- Invalid saved searches now prompt you to fix them If a saved sourcing search contains filters that can no longer run, Atlast now shows a clear message with a one-click option to reopen and correct it — instead of silently returning no results.
Fixed
- Sourcing scores are now consistent and accurate Talia was sometimes missing key skills — like Flutter or Dart — when scoring a candidate, causing strong matches to score low or vary wildly across re-runs. Relevant skills are now always included in the review, and scores are consistent every time.
- Forward-nudge banners hide once you've moved on The "Continue to next stage" banners on the Intake Form, Job Description, and Interview Plan tabs were staying visible even after you'd already completed the next step. They now disappear once the downstream stage is done.
Improved
- Required skills are now enforced in prospect searches Prospects who don't meet a role's must-have skills are filtered out before you see them. Match scores also reflect skill gaps directly — missing a key skill drops a prospect out of the top tier — and the score explanation names the specific skills that were missing.
- Clearer feedback after every prospect search A toast confirms how many prospects were found when a search completes. If nothing matches, a dialog explains why and nudges you to broaden your must-have skills, location, or seniority — instead of leaving you with a silent empty state.
- Sourcing page buttons are easier to scan The three action buttons on the sourcing page now have a clear hierarchy: Find more prospects is the primary action, Send to HM is secondary, and Refine search sits as an outline button — so the most important action is obvious at a glance.
Improved
- Ask your AI assistant about pipeline gaps Your AI assistant can now tell you which candidates have a completed interview but no scorecard yet, and which ones still don't have an interview scheduled — with a clear per-candidate breakdown and a suggested next step for each.
Fixed
- "New chat" now starts a fresh conversation Clicking New chat in the AI assistant menu was appearing to do nothing instead of clearing the previous conversation. It now starts fresh immediately. Switching to an assistant you're already talking to re-opens your existing conversation instead of wiping it.
Improved
- Stala explains blocked stage moves and suggests next steps When a candidate can't move forward because interview feedback is still missing, Stala now tells you exactly why and offers the right action — schedule the interview, wait for it to happen, or chase the specific interviewers who haven't submitted a scorecard yet.
- Automatic reminders for overdue interview scorecards Interviewers who haven't submitted feedback after an interview now receive automatic email and in-app reminders. Reminders start a day after the interview and stop once feedback is in. Admins can turn this off from organization settings.
- LinkedIn connection status stays accurate automatically If your LinkedIn connection drops or is removed, Atlast now detects it straight away and shows the account as disconnected — so you're prompted to reconnect instead of outreach silently failing against an account that's no longer active.
- Removing a team member disconnects their integrations When you remove a user from your organization, their connected Google Calendar and LinkedIn accounts are automatically disconnected — so a departed teammate's integrations can no longer be used from the platform.
Fixed
- Agent chat no longer covers menus and dropdowns Menus, dropdowns, and the help panel were being painted over by the agent chat when it was docked or maximized. They now always appear on top, as expected.
- Agent panel layout is remembered after a refresh If you docked or floated the agent chat and then refreshed the page, it reopened maximized instead of restoring your chosen layout. Your last panel mode is now saved and restored correctly.
- Typing an email in Add User no longer gets wiped If you typed an email address while the group list was still loading in the Add User dialog, the field could be cleared — leaving the Create button stuck disabled. Your input is now preserved regardless of how fast the groups load.
New
- Terms & Conditions consent gate on job applications You can now require candidates to accept your Terms & Conditions and Privacy Policy before submitting an application. Turn it on in Careers Page settings, add your URLs, and a consent checkbox appears on the application form automatically.
New
- Stala moves candidates through your pipeline Ask Stala to move a candidate to a new stage — or shift everyone in a stage at once. She confirms bulk moves before applying them and tells you exactly who moved and who couldn't, with a reason for each. All existing pipeline rules still apply.
Improved
- Pipeline board updates in real time When a candidate moves to a new stage — by Stala, by you, or by a teammate in another session — their card slides to the right column instantly, no refresh needed. Hired candidates now also show a Hired badge on their card.
Improved
- Atlas understands what you're looking at When you ask Atlas for help, it now knows which job, candidate, or list is on your screen — including any rows you've selected. Requests like "shortlist these" or "write a JD for this role" work without you having to explain the context first. You can also detach the context chip to keep a conversation independent of the current page.
Fixed
- Answering Atlas's clarifying questions now resumes correctly In some cases, responding to a clarifying question from Atlas would fail to continue the conversation. The exchange now picks up reliably after you answer.
New
- Next-action chips on every AI reply After each reply from Atlas, Talia, or Stala, you'll see a row of suggested next steps as clickable chips. Tap one to fill the composer with a ready-made prompt — edit it if you like, then send.
Fixed
- AI single-select questions now advance correctly Clicking an option in a single-select AI question — such as a hiring manager picker — was highlighting your choice but not submitting it, leaving the question stuck open. It now advances as expected the moment you select an option.
New
- AI agents remember your preferences Atlas, Stala, and Talia now build a memory of your working style from past conversations. You can view, edit, or delete what they've learned from the Agents Memory tab in your profile — and trigger a fresh re-learn any time.
Improved
- Layouts adapt better in split-panel views Pages across the command center, candidate profiles, jobs, and settings now respond to the width of their own content area — not just the browser window. Panels and grids stay correctly proportioned when a sidebar or side panel is open.
Fixed
- Sourcing dialog no longer goes blank on the last prospect When you rejected or shortlisted the last remaining prospect on the New tab, the preview dialog was staying open but showing nothing. It now closes automatically once there are no more prospects to review.
Fixed
- AI agent sessions restore correctly after reload Reopening an AI agent session now restores the full conversation — including any answers you gave to in-chat questions — exactly as you left it. Previously, reloading the page after answering a question would lose your reply and sometimes show duplicate question prompts.
Improved
- Integrations settings page is easier to understand The Settings › Integrations page now opens with a short intro explaining what each section does — API keys for reading jobs, webhooks for receiving change notifications, and signing secrets for verifying those calls. Saving a webhook or generating a secret now confirms with a success message.
New
- Public Jobs API for your own website You can now pull your published roles directly into your own website using a new public Jobs API. Generate a named API key under Settings → Integrations to fetch a paginated job list or full job detail. Candidates still apply through your Atlast careers page.
- Webhooks for real-time job updates Register a webhook endpoint to receive a call from Atlast whenever a job is published, unpublished, updated, or deleted — no polling needed. Every delivery is signed so you can verify it came from Atlast. Manage the endpoint and its signing secret from Settings → Integrations.
- Self-serve API keys and webhook management Admins can create named API keys, revoke existing ones, and configure the webhook — including which events to receive and rotating the signing secret — all from the new Integrations tab in Settings. A built-in developer reference explains how to authenticate and verify signatures.
Improved
- Tighter layout on job and candidate pages Job header spacing and candidate list layouts have been tightened up so content flows more cleanly and is easier to scan at a glance.
Improved
- Atlas opens beside what you're editing When you click Edit on a job description, intake form, interview plan, or pitch kit, Atlas now opens automatically — docked on the right, already focused on that item. Start refining it in conversation straight away, no need to summon Atlas or tell it what you're looking at.
Fixed
- Stray chat strip no longer appears at the bottom A thin chat panel strip was appearing at the bottom of the screen while Atlas was closed. It now stays hidden until you open Atlas.
Improved
- AI agents respect your role permissions Atlas, Stala, and Talia now check your permissions before acting. If you ask an agent to do something you don't have access to — scheduling an interview, messaging a candidate, sourcing prospects, or editing a job — it tells you clearly and stops, instead of failing with an unclear error.
- Atlas stays with you as you navigate Atlas now follows you as a docked side panel when you move between sections, instead of staying full-screen. It closes automatically on pages it can't help with, like settings or activity logs.
- Atlas edits appear on the page instantly When Atlas updates a job description, intake form, interview plan, or pitch kit, the change now appears on the page right away and is saved — no need to refresh. Atlas also no longer appears behind other page elements when opened.
- Targeted job description edits stay focused Ask Atlas to change a single field — like minimum years of experience — and it makes just that change and confirms. It no longer interrupts small edits with unrelated questions or multiple-choice cards.
New
- Smart Dock: summon Atlas anywhere you're working Press ⌘K or click the agent icon to open Atlas as a side panel, a floating window, or full-screen — without leaving the page you're on. Atlas sees what you're viewing, your conversation follows you as you navigate, and drafts appear in a collapsible artifacts panel.
- Switch agents and revisit past sessions from the agent menu The agent menu in the top bar lets you switch between Atlas, Stala, and Talia, start a new chat, or jump back into a recent conversation — with a link to all your past sessions.
Improved
- Sourced prospects explain how they were found Prospect invite pages now include a clear explanation of how the candidate was sourced, the basis for their assessment, and a link to the Candidate Privacy Notice — available in English and Spanish.
Fixed
- False "re-run sourcing" banner no longer appears After sourcing prospects, a "Job description was updated — re-run sourcing" banner could appear even when nothing had changed. The banner now only shows when the job description has genuinely been edited since sourcing last ran.
Fixed
- Sourcing staleness banner no longer appears by mistake After sourcing prospects, the "Job description was updated — re-run sourcing" banner could appear even when nothing had changed. It now only shows when the job description has genuinely been edited since the last sourcing run.
Improved
- Talia surfaces better-matched candidates Talia now respects the experience level a role actually needs — so a job asking for two years of experience no longer surfaces candidates with fifteen-plus. Searches for commercial roles like Account Manager and Customer Success return relevant results again, and individual-contributor titles are no longer mistaken for senior management roles.
New
- AI conversations now persist across sessions Your chats with Atlas, Stala, and Talia are now saved automatically. Resume any conversation from where you left off — even after closing the tab, refreshing the page, or switching devices. Each session is shareable via URL.
- Session list on the Command Center A new session list on the Command Center shows all your AI conversations with a preview of the last message and when it was active. Filter by agent, rename sessions to keep things organized, or archive ones you're done with.
- Artifact preview panel alongside your conversation AI-generated artifacts — job descriptions, intake forms, interview plans, and pitch kits — now appear in a side-by-side preview panel and are saved with the session, so you can pick up exactly where you left off.
New
- Add follow-ups after an outreach is already sent You can now add a new follow-up step to an outreach sequence even after the initial message has gone out. Open the Follow-ups schedule in the conversation panel, click Add follow-up, and write or generate the message — the new step slots in alongside the existing ones.
- Apply link included in follow-up messages Automated follow-up messages now include a personalized apply link for the role, framed as a natural reminder. The link resolves when each message sends, so it always points to the right place and makes it easy for a prospect to act on a nudge.
Improved
- Outreach composer is cleaner and easier to use The Connection Note and Full Message cards are now collapsible, the four composer sections share consistent numbering and spacing, and Regenerate draft sits inside the section it regenerates — so the rest of your form stays put while a new draft loads. The daily-send counter moved next to the Send button where it's relevant.
- Follow-ups panel is tidier in the conversation view The Follow-ups schedule is now collapsed by default, showing a compact summary — messages sent, scheduled, and when the next one goes out — so it doesn't dominate the conversation until you need it. Edit and delete actions on individual steps are easier to find.
Fixed
- Follow-up timing labels no longer show blank values Scheduled follow-up steps were displaying "Wait · Scheduled for" with missing values. Timings and the next-scheduled date now show correctly every time.
New
- Fair Use Policy now available in the Help menu A new Fair Use Policy page is now accessible from the Help menu, available in English and Spanish — so you can review the policy without leaving the app.
New
- Automated follow-up sequences for LinkedIn outreach When you message a prospect on LinkedIn, you can now attach a sequence of follow-ups that send automatically if they don't reply. Choose how many messages to send and how many days apart — Atlast drafts each one for you, and they stop the moment the prospect replies.
- Stay in control of in-flight sequences After sending, you can edit, reorder, or delete scheduled follow-ups, or cancel the whole sequence. Follow-ups pause automatically if you unpublish a job and resume when you republish — so a quick edit won't lose your in-flight messages.
Improved
- Get notified when permission templates are updated When Atlast publishes an update to a built-in permission template, any linked groups show an Update available badge and your inbox gets a notification that takes you straight to the affected group to review and apply the change.
New
- Create permission groups from preset templates When creating a new permission group, you can now start from a built-in template — Admin, Recruiter, Coordinator, Hiring Manager, Interviewer, or Read-only — instead of building from scratch. Pick a template, give the group a name, and you're done.
- Reset a group back to its template baseline If you've customised a permission group and want to start over, a new Reset to template option restores it to the preset baseline. A preview shows exactly which permissions will be added or removed before you commit — nothing changes silently.
- Get notified when a template is updated When Atlast improves a built-in template, any groups linked to it show an Update available badge and you'll receive an inbox notification that takes you straight to the affected group so you can review and apply the change.
Fixed
- Hitting the group limit no longer signs you out Trying to create a permission group when you've reached your plan's limit was incorrectly signing you out. It now shows a clear inline message prompting you to contact billing instead.
Improved
- Talia handles the full sourcing-to-outreach flow In one conversation, you can ask Talia to find prospects, review a candidate's profile, shortlist or reject them, route the shortlist to a hiring manager, and send a LinkedIn message — all without leaving the chat.
- Sourcing list and messages update in real time The New, Shortlisted, and Rejected tabs — and their counts — now refresh automatically whenever a prospect is moved, whether by you, a teammate, or Talia. Messages Talia sends also appear live in the conversation thread.
- Talia works from your existing pipeline Talia now reads the prospects already in a job's pipeline, so asking her to "shortlist the two in New" acts on the candidates you can already see — no fresh search needed. When no job is in context, she shows a picker so you don't have to retype the role.
