AI Assistant

Note that we have not determined the specific date for when in January this update will be automatically applied. You will learn more details about the timeline in early January 2026. We’re excited to share that next year (Late January 2026), both Assistant on Web and the Embedded AI Assistant will be updated to reflect Moveworks’ new visual identity. This update brings a cleaner, more modern interface and improves accessibility and consistency across all web experiences.Assistant on Web Embedded AI Assistant As part of th

We are excited to share an upcoming improvement in search quality: Moveworks AI Assistant will now be able to better answer questions from long KB articles with the help of an improved chunking algorithm that handles long articles better. This change will be rolled out gradually to all customers and expected to available for all customers in all regions by Thursday, December 18. What problem are we addressing?We are improving search performance on documents that contain very long chunks, which are Moveworks-extracted sub-parts of the article.“Chunking” is an important part of processing KB articles or web format documents (it is also sometimes referred to as “snippetization” in Moveworks terminology), which breaks down every article into smaller parts when it is ingested.Every chunk is converted into an embedding, which is a semantic representation of the text that captu

We’re using Live Agent Message Brokering and I’m trying to understand how often users are using the translation feature.I know users can set their preferred language before starting a live agent chat, and I also know I can filter to see how many users are starting live chats overall. What I’m looking for is something more direct and easy to consume:A way to quickly see how many users are using Live Agent Message Brokering with a language set to something other than English, without having to do a lot of manual filtering, coding, or custom calculations.Ideally, I’d like a dashboard or out‑of‑the‑box report where I can see something like:“10 users had their language set to Italian when using Live Agent Message Brokering” “8 users had their language set to German when using Live Agent Message Brokering” “5 users had their language set to Spanish when using Live Agent Message Brokering”In other w

Hey Team,We want to know if there’s a way to disable the ability for users to block messages that come from the Employee Communications module? Those are critical messages and while we can understand people may want to block the general ticketing messages but we need to make sure those are sent to people.Anyone else experience this, or find a way to get through, or should we raise this as a feature request?

I’m facing an issue with the search functionality in the Moveworks setup when trying to check ingested files.Currently, when I type a file name in the search field, the cursor automatically jumps out of the search box. Because of this, I have to click back into the search box repeatedly to continue typing. This makes it very difficult to type the full file name, and the experience becomes quite frustrating.At the moment, the only workaround I’ve found is to copy and paste the file name directly into the search box to avoid the cursor jumping issue. Can you please help look into this or suggest if there’s a known fix.

Our organization is redesigning a support portal and we would like to pass user queries from the page’s search to AI assistant for web.We’re fine with redirecting to the AI assistant web page, but we want to link directly into a conversation with the user’s query as the first prompt of a new chat. I can see that chats create a new URL with the following pattern: /assistant/t/<chat-id>, and it stands to reason that the landing page at /assistant is passing the text I enter to the assistant and generating the URL with the chat id.What is the URL where I can pass text directly to the AI assistant to open a page with the assistant already working on the query? Would it be something like /assistant/q?text=’How do I access the VPN’

For Answers/knowledge search, sometimes the bot returns a response with links/URLs directly in the response, but sometimes it provides just text and user has to dig into the references to find the links. Is there a way to force the response to include the URLs so that user can navigate to information more easily?Also, is there any way to update the GenAI prompt that’s fed to the bot so that we can have the bot respond more appropriately according to our organization’s needs?

We have many articles in our platform that contain a list of dates. These could be holiday dates for a specific year(s), pay dates for specific year(s), etc.  When users ask for the upcoming date(s) in the specific list, the bot often serves incorrect dates compared to the prompt and current date.  Example:Prompt: When is the next holiday for US Employees? Bot response: The upcoming holidays for US employees are New Years Day January 1, 2025.  You’ll see the year is incorrect. We also have other instances where employees were served previous months, not upcoming months.  Any ideas for optimizing the content for better serving? Also interested if this could be looked into as a potential improvement for the Moveworks platform to further optimize this type of interaction. 

I read somewhere that Moveworks supports OBO workflows for Confluence but can’t seem to find the docs for it. Can someone point me to the documentation for this? 

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

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

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?

When a user asks Moveworks Copilot a question, that prompt gets stored as raw interactions in the analytics in Moveworks admin portal. How long is the raw interaction data available for? If we want to shorten the period, where do we configure this? Thank you.

When a ticket is intercepted by the bot, the user receives a notification saying “Please check this information, it might help with your ticket.”The response includes a link, but that link is not served or accessible within the ticket context.In the classic version ticket interception, the same link appeared and worked directly from the ticket, which helped agents quickly see what information was shared with the user.It would be great if this functionality could be restored in the Copilot version, so that the link is available within the ticket context as well.

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.

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!

Hi there,We have a highly manual processes within our company we are trying to automate. We have very large excel spreadsheets that contain vendor details. E.g. their location, services they provide (e.g. diggers, trucks etc), and their different rates for weekday, night, Saturday, Sunday and public holidays.Our people on site engage with this spreadsheet daily to see what contractor is best for the job, but it is incredibly complex, and difficult to navigate especially on a phone / tablet which is often used on site. We want to improve this by ingesting all of this information into a document that Synap can surface to the people on site. This means they will interact with the bot and say "I am in the Dunedin region and I need to hire a truck with driver, that offers a 6 wheel tripper on a Monday" - Give me the rates of people who provide this, and then provide me the cheapest contractor who provides this service"Our CSM has kindly suggested that

Hi Moveworks,Back in January, the Microsoft partnership with Copilot was announced. Where can I find more information on this integration? We just launched Microsoft Copilot for our org and we would like users to have access to our Moveworks Assistant in this new environment. 

Hello,I have a question about the AI Assistant.Right now, we use our bot to send messages in it's own chat.I want to know if our bot is able to send messages that are directly replies to posts in other parts of Microsoft Teams, like groups, channels, or send anything to distribution lists.Can the AI Assistant reply directly to a message in those Teams places? Or can it only send new messages in its own channel?Thanks for your help!

We currently have a deep link for our web bot that opens directly in Slack when a user clicks on it. We’re looking to implement similar functionality for our web bot on the ServiceNow Support Portal.Use Case:When a user clicks the "Get Support" button on the ESC portal, we want the Bot to automatically pop up, providing a seamless support experience. 

Hi,We have one reported issue with switching to Afrikaans in Teams.When I say ‘switch to Afrikaans’ I get the following message:I'm sorry I don't have information on how to address this issue. Please use the form below to submit a ticket for assignment to our support teamWhen I ask the Moveworks bot in Teams to switch to Dutch it works fine.Your documentation below shows support for Afrikaanshttps://help.moveworks.com/docs/multilingual-support-mls#what-languages-are-supported-by-moveworksCan you provide guidance on this?