Recently active topics

I have a few plugins that collect information to fill out a ticket. The different fields to fill out are slots in the conversational process. The slots have a resolver strategy to pull the possible options from a api. I have the slots set to Infer slot value if possible. However inferring never actually works when a resolver strategy is applied.For a access management plugin, I can say I want access to <exact application name in system> and the plugin will show the first x options. It normally ends with the correct option and says things like, if you mean <exact application name in system> type “yes” or the number in the list. It however always makes the user confirm the option instead of inferring.I am getting around this by throwing a generate_structured_value_action at the beginning after asking for a description of the ask, but it is possible the user won’t provide enough information to determine the correct option and it feels like a hack when slots should be the answe

Hello, Nevermind

ScenarioWe’re seeing a small but recurring UX gap in our Moveworks → UKG plugin flow and wanted to check if others have solved this or if there’s a recommended pattern.Current behavior User asks Moveworks to perform an action (example: approve timecard). Plugin detects expired/invalid OAuth token for UKG. Moveworks prompts the user to re-authorize via an auth link. After successful authorization, the user must explicitly type something like “retry approve timecard” to continue. Only then does the plugin re-execute the original action. From the user’s point of view, it feels broken or incomplete — they believe the task should continue automatically once access is granted.Expected / desired behaviorOnce the user completes authorization successfully: Moveworks should automatically resume the original plugin execution The user should not be required to type a manual “retry” command The experience should feel like a single continuous flow, not two separate conversations Question

Hey everyone 👋We are excited to announce User Consent Auth in Agent Studio, which lets you securely deploy Agent Studio use cases that respect user permissions and access levels, is now in Preview!What are the capabilities of this launch?Permission Consistency. Ensure users only access data they are permitted to see. Any user interacting with an app through Moveworks does not gain privileged access through a generic service account. User Control. Individual users can grant access to their own data. Unlike JWT where a developer can impersonate any user, this allows users to control exactly what types of data and actions the connector is allowed to interact with. Tokenized Security. Short-lived access tokens are generated and continuously rotated with longer lived refresh tokens to reduce the risk of damage from a privileged access token being intercepted. Revocability. Each user authorization is associated with a token that can be easily audited and revoked at any time in case of secur

I wanted to get clarification on lifecycle management for custom views created. I created a pilot HR Custom View and shared it only with myself for testing, but I’m not seeing any self‑service way to edit, delete, or re-share that view after it has been created/shared. Based on what I’m seeing, it appears I would need Moveworks support to update/remove/reassign views—which could result in multiple duplicate views as we expand sharing to Little Joe domain admins.   Could you please confirm the following? Is this expected behavior? Once a Custom View is created (and/or shared), is it intentionally not editable/deletable/re-shareable by admins? Is there an admin role or feature we’re missing? For example, does Bot Analytics Admin (or another role) have the ability to manage view lifecycle (edit/delete/change sharing) that may not be enabled for us today? If lifecycle management requires vendor support, can you advise the correct path to update/delete the pilot HR view I created so we don’

We’re working on a use case where users request specific software, and our organization maintains a list of available software on a SharePoint site. Our solution uses a compound action: it fetches the software list via HTTP and script actions, then filters and displays results based on what the user asks. We’ve added both positive and negative utterances as well.The issue is that when someone asks for a particular software (for instance, “I need Teams”), our workflow works sometimes. But if the user says something like “I need to install Teams,” the system often triggers an internal plugin—such as installation help forms or knowledge base articles—instead of using our custom plugin. Even though our internal team has disabled some pre-built plugins, we still get unwanted results from forms. This means the intended custom plugin either fails or returns unnecessary or irrelevant information.Sometimes it works (“I need Teams” gives the right result), while other times, like with “I need to

Background:In June, before Webhooks were available, setting up instant approval notifications through the Concur API was difficult because Concur doesn’t have native automation tools. At that time, we were told we’d need to regularly poll Concur records to check for updates, since Webhooks weren’t an option yet.Current Situation and Questions: Webhooks Availability: Now that Webhooks are available, can we use them to automatically trigger approval notifications in SAP Concur? In other words: Instead of our bot checking for updates all the time, can Webhooks now handle this automatically? Impact on Our Bot Performance: Given that we would like to use real-time Webhooks (and not scheduled ones), what does this mean for our bot’s performance? Specifically: Will this drastically reduce the number of API calls our bot makes? Will it make our bot more responsive, especially considering hundreds of expense approvals happen daily? Scheduled Webhooks: I understand that Scheduled Webhoo

Hi all, we have numerous (200+) PDF documents containing tables, and some consist entirely of a single table relating to rates etc. Editing these manually into a readable format for Moveworks would be very time-consuming for the various departments and would slow adoption. Is there a automation tool that can be used to edit and modify the tables for improved readability? How do other companies edit their existing and legacy documents? Thanks

ProblemA failure in ingestion from a single application currently stops the entire ingestion flow, preventing updates even when other sources are healthy.Request:Introduce an active–passive failover switch for each application connector within the ingestion pipeline. So if there is an issue with one application the whole ingestion process still completes and only attributes used from the failed application is not updatedBenefitsPrevents full ingestion stoppages Ensures maximum data freshness from available sources

Form update conditions don't support $LENGTH() for string validation. Is this intentional or a limitation? Any workarounds?this is in reference to Rich Ticket Filing. I was trying to do some validation, because ServiceNow cuts off the short description after 160 characters. And we have folks that use that field for the description. Basically, my intention was to show the helper label that explained the limit if the user entered more than a certain number of characters in that field. I tried a few different conditions like:$LENGTH(field_short_description) > 143field_short_description.$LENGTH() > 143field_short_description.$LENGTH > 143and also the one you see in the screenshot, all of them seemed to result in the error “Invalid DSL expression” which makes me think that length isnt available to use in the condtions.  I’m aware this isnt an agent studio issue but just hoping someone with some knowledge could shed some light on this for me. Thank you

Post updated with new information on Dec 12, 2025The GPT-5 series of large language models stands out for delivering state-of-the-art reasoning. Better reasoning (i.e. thinking and decision making) represents a major leap for agentic tasks, offering improved instruction following, more effective plugin (tool) use. OpenAI released GPT-5.1 on November 13. We’re excited to share we’ve conducted a full set of evaluations to assess its performance relative to GPT-5 and GPT-4.1.The GPT-5 series of large language models stands out for delivering state-of-the-art reasoning. Better reasoning (i.e. thinking and decision making) represents a major leap for agentic tasks, offering improved instruction following, more effective plugin (tool) use.Live today, in the Moveworks AI AssistantGiven the GPT 5.1's strong performance, we have upgraded to this new model in the Moveworks Reasoning Engine. No setup work or downtime is required to see this change, as this has already been made in your AI Assista

Hey builders!Office Hours #2 is in the books and we went FULL dev mode today 🔥The questions were chef’s kiss, we went deep on the exact pain points you’re hitting in production right now.Here’s everything we crushed:Using a single webhook listener to trigger multiple ambient agent plugins → Use event filters to route one webhook to the right plugin (PTO vs Approval demo’d live) → No more “one webhook per use case”— one gateway listener to handle all plugin triggers “My API returns a giant payload and it has no filtering capabilities” survival guide → Chunking + parallel blocks + LLM actions = 🏃‍♂️ → Pagination patterns in compound actions that work “My connector has way too many permissions” → Always. Provide. Access. Rules Default slot values without prompt engineering? → Optional slots with default fallbacks are not here yet!  → Prompting slot descriptions is still the most reliable workaround File Slots round 2 – now with constraints! → You can validate file extensions in your slo

We are excited to inform you about a major upgrade to Moveworks search capabilities, coming February 2026 for our US and EU customers, followed by GovCloud, AU, and CA customers later in the year.Read on for when these changes are coming and what you can do to get ready! A brand new Enterprise Search plugin for the AI AssistantWe are rolling out a new Enterprise Search plugin in AI Assistant for all US and EU customers, which uses an upgraded and scale-ready search and relevance ML system. This new plugin represents our new and improved search that is capable of finding the most relevant content across millions of documents.In early customer deployments, we see an increase of ~10% in success rate for answer seeking queries in Assistant with this plugin!The Enterprise Search plugin will replace the current Knowledge Search plugin. All customers in US and EU regions will be migrated to the Enterprise Search plugin by March 2, 2026 as a platform-level change. All of your organization’s in

We're excited to launch a public archive of The Agent Builder newsletter, your resource for technical breakdowns, implementation patterns, and real-world builds from the Moveworks developer community. Click here to find the archive. What You'll Find HereThis archive features past issues of our newsletter, covering:Technical deep-dives into agent architecture and design patterns Real-world builds showcasing what developers are creating with the Moveworks platform Implementation guides with code examples and best practices Platform updates and new capabilities as they roll outHow It WorksNew issues are published here. You can browse the archive anytime to catch up on past content or reference specific technical patterns.Want It Delivered to Your Inbox?Get each issue delivered directly to you as soon as it's published.👉 Subscribe now: aka.mw/newsletterWhether you're exploring the archive or subscribing for inbox delivery, we're glad you're here. Let's build something great together.

Welcome to the first issue of The Agent Builder — a lightweight update for the developers building agents. Editions will cover technical breakdowns, common patterns, actionable updates, and real builds. What are File Slots?Slots describe the inputs that your agent collects from a user to proceed with a task. File slots make it possible for your AI agents to collect files like expense receipts or invoice PDFs.File Slots help AI agents escape JSON to become Multipart.The core tension you may have come across: agent frameworks are JSON-first. LangChain, Semantic Kernel, even our old setup, everything flows through structured payloads. Files don't fit that model, so you rely on hacks:Base64-in-JSON: Bloats ~33%, longer transfers, memory spike on decode "Paste a link" strings: The backend fetches, there's auth headaches and timeout risks Custom multipart builders: They're a boundary creep and third-party libs can leak resourcesMultipart/form-data is the right way for binaries, they're effic

Here's the situation we face when a user submits time through the Workday Portal: our bot can successfully approve or deny the time off. However, the problem happens when a user submits time directly through the bot. In these instances, the approval/denial process doesn't seem to work as expected. The error we get back is “This event step is not awaiting your action”. When I review the data through the RaaS report, it appears that working time-off requests have two true approvers, whereas those that are not working only have one. The API call we are using to request time off is /api/absenceManagement/v1/{{tenant}}/workers/{{worker_id}}/requestTimeOff.Has anyone experienced this issue before?

Hello,I would like to know if it is possible to add a clickable link for Live Agent Chat inside the Smart Handoff text.Right now, users first choose a team (for example, IT or Finance). Then they choose an option in the second drop-down (for example, “Finance Applications supported by EAS”). After they choose this option, a Smart Handoff text box appears with more details.Today, the Live Agent Chat is shown as a separate option in the drop-down list, and the link is hard coded under each domain. Instead, I would like the Smart Handoff text itself to include a clickable link that will start a live chat with an agent.In other words, when the Smart Handoff text appears (for example, for “Finance Applications supported by EAS”), I would like the phrase “Live Agent Chat” in that text to be a clickable link that opens the live chat.Is this possible to configure in Smart Handoff (for example, by adding a URL or action to the description text field). If yes, can you please share how to set thi

I have Plugin which is triggered from a webhook. Plugin sends the message to slack with 2 button.When a button is clicked the conversation process is to be kicked to start the conversation to the user. When Add Funds is clicked I need ask couple of questions to get the in put and send to an External API.this is the error msg i get when i validate it  Any pointers how this can be achieved will be really helpful

Hello EveryoneTrying ingest Unily , created connector as per https://help.moveworks.com/docs/content-integration-unily.But in internal knowledge ingestion , no wizard is coming up(not like we have during service now setup), and not able to ingest, can anyone guide on this the right way 

Hi All,I am trying to send a slack message from a plugin.I have a formatted HTML message coming as a part of json payload.as shown below:{"message": "<br>The below <b>Purchase Order</b> is set to expire in 90 days or less.<br> Please review the details, and then indicate if you would like to initiate a renewal or not.<br> • <b>PO Number</b>: <a href="https://www.google.com">PO1100004112</a><br> - <b>Requisition BU</b>: TEst, BU.<br> - <b>Supplier</b>: Test Supplier <br> - <b>Requester</b>: test.user@test.com<br> - <b>PO Description</b>: Test Support Fee<br> - <b>PO Total Amount</b>: 50,000.00<br> - <b>PO Remaining Amount</b>: 18,500.00<br> - <b>PO Expiration Date</b>: 31-Dec-2025","recipients": ["dhotre.virupaksha@airbnb.com"],"context": {"slots": {"messageId": "62cd4a70cf8011f0b478a201a1fa3

Hi everyone! I’m new here to sharing my introduction. Here’s a bit more about me:What I’m working on: I’m leading efforts to scale AI-driven support across multiple business units, focusing on smarter ticket routing, better self-service adoption, and faster resolution times with Moveworks.What I want to learn or discuss: Excited to hear how others are driving AI adoption in enterprise teams, especially non-technical ones, and measuring real business impact. Tips for balancing automation with a human touch are always welcome.Fun fact: When I’m not at work, I’m rock climbing, or training my dog to sit through virtual meetings, he’s now the best (not so) silent observer on the team.

At present, there is no support for analytics related to ambient agents. Although logs of successful notifications are available in the logs section, corresponding analytics data for ambient agent use cases is not accessible. This limitation is impacting the integrations we have implemented to monitor success metrics.Could you please let us know when will you introduce analytics support for ambient agents?

If you use a slot resolver for a conversational slot and set the Slot Inference Policy to Infer slot value if available, it will pull from the User Profile.. for example, their location from ingested user data.  This can be an issue for some plugins where similar data exist but has nothing to do with what the plugin is asking for.  It would also be beneficial to be able to define what each ingested element from the User Profile is, allowing context for Moveworks to better understand when to use those slots.Example:We’ve built a Conversation Flow where one input is a state, and we provide a list of states. If a state matches the user’s profile, the Infer slot value will automatically select that state instead of presenting the choice to the user—even if we add notes in the plugin description, conversation process, and input variable advising not to use the user profile location. This matters because the state they select is intended for a location they need information on, not necessari

Hello Moveworks Admins,We're excited to announce that User Geocodes will be available in the Imported Users screen on 7th January 2026. This new feature provides visibility into user location data, making it easier to troubleshoot location-based answers not getting served and validate identity configuration. 🌏 What are Geocodes?Geocodes are structured location coordinates extracted from location-related fields (city, state, country, zip code) in your identity source during user ingestion. This allows Moveworks to know “where this user is”, which is then used in search and personalization—for example, boosting KB documents that match the user’s region when the user asks a generic question like “what are my benefits?”Read more about Geocodes in User Identity. 🌟 Key BenefitsFaster Troubleshooting: When users report unexpected answers, admins can quickly check if the user geocode is missing or incorrect to determine whether the fix is required in identity configuration (location fields,