Product Forums

Hello Team,I am working on a CMDB search skill, when there is no data i would like to show few default messages, somehow its now working.. but with data case is working well… below is my default  instructionany tips and tricks would be appreciated.  display_instructions: '"You must always respond with exactly one complete sentence in plain English.  Do not use bullet points, line breaks, or labels under any circumstances.  Do not explain the search process or mention that you looked for keywords.  If there is application data available, combine the application name, owner,  description, tower, and platform executive into a single user-friendly summary.  If there are no results or data is missing, you must respond only with this text,  \"No application information found based on exact keyword search.\"  In all cases, after the sentence, you must append the following text exactly and without modification

Hi communityI have noticed now we have live suggestions/data entries:In the conversational process there is a slowness because this is always saving and removing me from typing.Furthermore while creating a script in the script actions I have the same situation I cannot code there anymore - since this is always suggesting entries and stops my typing and everything breaks and is slow.Is anyone having the same issue?I understand this is a new feature but it would also a possibility to have the option to disable it? or solve it by increase the speed of the instance. This might also be a issue with the browser - dont know - I am using EdgeThank you

Hello Moveworks Admins,We’re excited to announce that enhanced On-Premise Agent screen will be available in Moveworks Setup on 11th November 2025. These new features provide comprehensive visibility and control over your deployed agents, making agent management easier and more effective.🌟 Key BenefitsMonitor agent health and status: Track whether your agents are successfully connecting and serving requests from the Moveworks platform Streamlined agent lifecycle management: Create, upgrade, and stop agents directly from the Moveworks Setup UI Troubleshoot connectivity issues: Identify and resolve agent health problems with detailed status monitoring Centralized operations logging: Track all create, upgrade, and stop operations with comprehensive audit trails Simplified deployment: Generate secure org

We are about to introduce a new role in Analytics called the Interactions Viewer. This role ensures secure and streamlined access to interaction-related data across all web-based data products.🔍 About the Interactions Viewer RoleUsers with the Interactions Viewer role will have access to the following products:Interactions Table Feedback Table Channel Resolver InteractionsAdditionally, they shall be able to drill down into widgets , ie, click to view interactions in Plugin insights.🔐 Access ManagementAll users who currently have access to any of the products listed above have automatically been assigned the Interactions Viewer role.Super Admins can add or revoke this role for users as needed. We recomme

I have two plugins built in Agent Studio—View Addresses and Update Address—that are meant to work back-to-back: users typically view their current address, then say “update address” to start the edit flow. When the update is invoked immediately after a view, the handoff does not preserve context: the update flow behaves as if state is lost, and it doesn’t re-display the primary address as part of the update confirmation steps (we must display the primary address to show the user what they are updating). Notably, if the user starts with “update my address” directly (without viewing first), the flow works correctly, which suggests the issue is specific to inter-plugin state/intent handoff or session variable carryover. 

we already have movework chat-bot application integrated within Teams.and Now, we have created our own AI-Agent application. what is the best possible way to integrate it with movework chat-bot. 

I am using a compound action with a notify step (just to use buttons like Approve, Reject). The bot is giving false positives (like time off has been approved), and it is giving me more information than what I have asked it to display via the compound action. I am using this compound action in a conversational process, and have seen this duplication before when used like this.

In the Hackathon, the roadmap of Moveworks was presented to us. We were told that voice integration is in the pipeline. Do we have the timeline? Where can I find more updates on the progress? 

As part of our ongoing efforts to enhance our customer support and improve our knowledge base, I am exploring efficient ways to analyze raw interaction data. Specifically, I am focused on identifying knowledge gaps by reviewing instances where our chatbot did not provide a direct answer and directed users to "Get Help" options.Currently, the volume of these interactions is substantial, making manual analysis time-consuming—often taking several hours to sift through thousands of rows. Despite filtering the raw interactions and utilizing my companies internal GPT model, I have yet to find a scalable and effective method to streamline this process.It’s important to note that we are not setup for Knowledge Writer or or using similar automated tools for this task; everything is done manually. I would greatly appreciate hearing from anyone who has developed efficient, automated, or innovative approaches for:Quickly identifying knowledge gaps Effec

I have a api call to get a list of categories. From there I have another to get a list of subcategories. I see slot resolver strategies do not have access to the data element from the conversational process. Is there a way I can pass this information with input variables in a reliable way without asking the user for the category twice and letting them select a improper value?

I have tried both removing the progress updates as well as customizing the progress updates and I have had no luck with either. Is this a Moveworks bug or is there some other way to disable these progress updates apart from not including them in a compound action? Here is an example of the behavior I want to avoid.  

🎉🚀 Agent Architect Visualization Upgrades!!! 🚀🎉We just came off a month of intense hacking with all of you at our San Francisco, London, & Bangalore Hackathons. First of all, THANK YOU so much for coming out – it was very inspiring seeing everything you built & continue to create. We heard from you at our hackathons it’s frustrating when Agent Architect generates Compound Actions or Data Mappers in JSON instead of YAML, NO MORE!Agent Architect code outputs can now be toggled between YAML & JSON with a click. BONUS – Agent Architect diagrams (Mermaid) can now be visualized, expanded, & downloaded 👉 Tell me what you wanna see next! Happy hacking!- Ajay & The Agen

I’m trying to get clarity around the relevance of newly ingested articles for the monthly recommendations. If a knowledge article is added to the knowledge base AFTER a ticket has been created, will this article be included in the “VIEW ARTICLES TO UPDATE” section, or is it considered an outlier since it was not part of the knowledge base when the ticket was created?

We’ve been receiving feedback about requests and ticket becoming harder to accomplish via our Moveworks assistant. Below is one example, and I have 3 main questions:Why is the agent ignoring the Ticket route so many times? Even at the end, it is not presenting TICKET WITH IT as an option! If domain has been identified, is there a way to prioritize a handoff option? Is there a solution to manage catalog items for the assistant better? The lack of context and added logic to catalog items often hurts the experience as the user can be served with up to 4 different request forms at once Moveworks agent ignores ticket request, finds a match with catalog item. This is OK since we do want catalog items fulfilled when needed, but the lack of context hurts accuracy

We’re excited to announce that new Ingested Knowledge and Files Screens will be available in Moveworks Setup on November 12, 2025. These screens provide comprehensive visibility into content imported from external systems, making content troubleshooting easier and more effective.🌟 Key BenefitsCheck health of connected systems: Monitor whether Moveworks can successfully crawl and ingest files from your connected systems Troubleshoot record-level issues: Identify and resolve issues such as lower than expected record counts, missing records, a record not serving to users, and more Periodic checks on key metrics: Keep periodic checks on ingested records, serving records, and non-serving records Perform UAT of newly configured system: Validate that content is successfully ingested and indexed. Additionally, verify ingested and serving record

We have the listener functionality and are leveraging it for a larger use case. We need a way to show how much people are actually using it for adoption reporting. However this information is not included in the analytics ui page or in it’s associated api. The most I can see on usage is in the logs, but that is not exportable so not really usable. Is there any other way to get reporting on how many times the listener is called as well as how many times people actually proceed with using it?

Does Moveworks track confidence level/scores of the bot’s responses to a user? (even if this is tracked only on the back end)What are some ways to determine the accuracy of responses besides asking users to report feedback?

Hi all,My team is currently working with Channel Resolver in one of our Slack channels. We had someone submit a sensitive ticket via this channel which should have been submit privately.After this, we deleted the message within Slack, but our Moveworks bot was linked to the ticket still. When the end user or any of our IT employees would reply, channel resolver would create a new message in our resolver channel, rather than in the deleted thread.Has anyone experienced this before, and if so, do you have any guidance on how to prevent sensitive information from being shared in the channel? I believe the fix may be to disable channel resolver, but wanted to reach out to this community and see if anyone has had an approach to rectify this.

Great to connect with you! I recently joined the Moveworks community after attending a Hackathon event in San Francisco. I have a pipeline of tasks that I am eager to automate and am highly motivated to make progress but am interested to find the best way to apply my efforts.What I’ve done so far is to complete two of the Quickstart projects and about half of the course videos of “Agent Studio Developer: Level 1 Certification.” I plan to work through the remainder of the videos and a couple more of the Quickstarts. For those of you a bit further along in your automation or AI usage journey, where do you recommend that I focus my next effort?For context, I have a “commercial/analyst” type role. I regularly code in Python collaborating for my job on Github, but I’m not a professional developer. I experimented with Power Automate about two years ago, setting up a few automations that I continue using until now, but got stuck back then in implementing some

i have an agent studio plugin that uses the User object as a slot to get a user context. what i’m finding is the bot is responding with a default set of properties even though i’m not specifying which values from the user object to display. i’m using the built-in mw.generate_text_action and defining a slot in the conversation process as a User object type, which is grabbing the user context correctly. but the bot response message is showing the same default properties (full name, email, role, manager, dept, location) and i don’t want to display those, at least not like that. i tried using display_instructions property in the output mapping, but that doesn’t work. the default properties are still displayed.where is the plugin getting this format from? and how do i override it or define my own template?

 At Moveworks, we’re committed to continuously improving the reliability, transparency, and security of our platform. To that end, we’ll soon be upgrading our SIEM log processing pipeline.This change unifies all customers under a single, modernized system — delivering more structured logs, clearer schemas, and more frequent updates, all while reducing operational complexity behind the scenes. 🧩 What’s Changing Updated File StructureYour SIEM logs will now be stored in a versioned subdirectory (v1) within your SFTP bucket. This helps clearly distinguish between versions of the logging pipeline.Old File Path New File Path logs/2024-08-20/2024-08-20_audit_log.json logs/v1/2024-08-20/2024-08-20_audit_log.json

We have observed that our bot is experiencing intermittent response timing issues this morning in both Microsoft Teams and the web platform. At times, it may take several minutes to reply; at other moments, it responds within a minute. Occasionally, it may timeout, displaying a message to try again shortly.We have checked the Moveworks issues dashboard and did not find any reported problems. We would like to understand if others are experiencing similar issues, especially considering ongoing concerns with AWS services.If you are facing related AWS issues that affected your bots, please share how you addressed them. Your insights could help us find a solution more quickly.Thank you!

Any Idea why my user_email variable would not be passed through here or be blank as you can see in the canonical_endpoint? I have also tried replacing “user.email_addr” with “mw.user.email_addr”. Why are neither of these working? Is my syntax incorrect? Or could I have incorrect bot configurations? I wouldn’t need a slot here would I? 

disregard… i’ve simplified the plugin and took out the compound action and just referenced the built-in action in the conversational process directly. it’s sending one message now, but it’s not the right format. that’s a separate issue though, so i’ll create a separate post for that. this one can be closed.i have a plugin with a conversational process that runs a compound action. the compound action runs a built-in action that returns a single value.the bot is responding with three messages, which is what i don’t want. where are those messages coming from? my understanding was the bot will send a message AFTER the last action completes. if there is only one action, why is the bot responding with multiple messages?the last message sent is more or less what i’d want displayed to the user, not the others. 

We have on prem agents connected to our ITSM (ME/SDP).  Any way to use them via bot to reach into an end users’ machine to add them as a local admin to conduct elevated tasks - but them remove it after a time limit?  Thanks!