Token Limit Exceeded? What's Actually Going On and What to Do About It ?
Hi All, Based on some recent experience across the organisation with token limit issues, I wanted to put my thoughts down and actually dig into what's happening under the hood, rather than just chalking it up to "we need a bigger plan." If you work anywhere near the Microsoft ecosystem these days, you're probably touching more AI tools than you realize. Copilot in Word and Excel, GitHub Copilot while you code, Copilot Studio if you're building agents, maybe Security Copilot or Copilot for Sales depending on your role, and increasingly Azure AI Foundry if your team is building anything custom. I work across a good chunk of this stack day to day, and at some point, almost everyone runs into the same wall: "Token limit exceeded." "You've reached your usage limit." "Upgrade to continue." The first instinct is usually to assume you did something wrong wrote too much, uploaded too big a file, or just need a fatter subscription. Sometimes that's the actual story. But honestly, often, that error message is standing in for three completely different problems that all happen to look identical from the outside. One is about how much text a model can physically process at once. One is about your license or credits running dry. And one has nothing to do with size at all it's just about how fast you're sending requests. Once you know which of these three, you're dealing with, the fix becomes obvious. Until then, "upgrade your plan" feels like the only lever you've got even when it isn't. This post walks through what a token is, why Microsoft's various Copilots each handle this differently, and what habits genuinely cut down on these interruptions instead of just throwing money at the problem. Part 1: So What Is a Token, Really? A token isn't a word, and it isn't a character it's somewhere in between. It's the small chunk of text a model's tokenizer breaks your input into before it can do anything with it. Take a word like "unbelievable." A tokenizer might split it into three pieces something like "un," "believ," and "able." Short, everyday words usually come out as a single token. But code, technical jargon, acronyms, and non-English text tend to fragment into a lot more tokens than you'd guess just by looking at the word count. This is why every AI tool has a ceiling on how much it can handle in one go, and that ceiling isn't measured in words or characters it's measured in tokens. Your prompt, any documents or emails it pulls in as context, the back-and-forth history of your conversation, and the response itself all draw from the same pool. Once that pool runs dry, something has to give: the tool truncates, rejects the request outright, or quietly summarizes older context to make room. The part that trips people up: token count doesn't map cleanly to word count. A short, dense paragraph full of code or acronyms can eat up more tokens than a much longer plain-English message. Part 2: Three Different Limits, One Confusing Error Message This isn't always obvious upfront, even to a lot of admins managing these tools: "token limit exceeded" is really a stand-in phrase for three separate limits, and they don't behave the same way. This isn't unique to Microsoft either every major AI platform bundles these same three things behind similarly vague error messages. Microsoft's stack just makes a good case study because so many of us touch multiple pieces of it in the same week. The context window is the ceiling on how much text a specific model can process in a single request everything from your prompt to retrieved documents to chat history. This is tied to the model itself, not your subscription. Swap from one model to another inside the same tool, and this ceiling can move without you doing anything differently. Your license, credits, or feature allowance is a completely separate thing. This is what Microsoft 365 Copilot plans track through AI credits and feature limits, and it's what Copilot Studio measures through Copilot credits at the environment level. A single action summarizing an inbox, generating an agent response, running an analysis deducts from this pool regardless of how small your actual prompt felt. Run out, and you get blocked, even if you're nowhere near any context window limit. The rate limit is about speed, not size. Copilot Studio, for instance, enforces quotas measured in requests per minute or per hour to keep the system stable under load. Send messages too quickly, which happens easily with automations, flows, or bots, and you can get throttled even with a tiny prompt and plenty of credits left. The reason this matters: a plan upgrade only ever fixes the second one. If you're actually running into the model's context window or getting rate-limited, paying for a bigger license won't change anything, and that mismatch is exactly where most of the frustration comes from. Part 3: How This Plays Out Across the Microsoft AI Stack The Microsoft ecosystem isn't one AI tool wearing different outfits it's genuinely several different systems, each handling tokens and limits in its own way. Here's a tour of the ones people run into most. Microsoft 365 Copilot (the one living inside Word, Excel, Outlook, Teams) doesn't work off a single published token number the way a developer tool would. Instead, it dynamically pulls together your prompt, recent chat history, and relevant snippets retrieved from Microsoft Graph your files, emails, and messages and quietly summarizes or drops older material to stay within bounds. Where this usually breaks isn't the context window at all; it's the AI credit and feature-limit system running out, often without much warning until you're mid-task. GitHub Copilot Chat is more like a traditional developer tool. It has a fixed, published token window tied to whichever model you've selected, and that limit applies consistently whether you're in the browser, VS Code, or the CLI. The failure mode here is usually a long conversation or a big multi-file context quietly creeping past that ceiling. Copilot Studio, where a lot of custom agent-building happens, runs on Copilot credits per interaction, plus its own requests-per-minute and requests-per-hour quotas at the environment level. If you're grounding an agent in SharePoint content, there's also a separate file-size ceiling to watch content over a certain size can get silently excluded from generative answers depending on your tenant's licensing. Azure AI Foundry (recently renamed to Microsoft Foundry, in case you've seen both names floating around) is where this gets more directly in your control. If your team is building custom applications on top of Azure OpenAI or other models in the Foundry catalog, which now includes everything from GPT to Phi to Claude to Llama, you're working with explicit, published context windows per model, and you're billed per token rather than per credit. It's a different mental model entirely: less "you hit a wall," more "you're paying by the word, so design accordingly." Security Copilot, if your org uses it for threat analysis and incident response, runs on its own capacity model pooled compute units at the tenant level rather than a simple per-user cap. It's easy to assume this behaves like M365 Copilot license limits; it doesn't. Copilot for Sales, embedded in Outlook and Teams for CRM-connected work, and Copilot in Power BI, which now goes beyond generating summaries to actually helping build and refine semantic models, both draw from their own feature-specific allowances layered on top of whatever base Microsoft 365 or Power Platform license you're on. And then there's the multi-model wrinkle that trips up teams the most: because tools like Copilot Studio and GitHub Copilot let you choose between GPT-based models, Claude, and others, the exact same prompt can have a different effective context window and a different token cost purely based on which model handled it that day. This is a big, underrated reason behind the "it worked fine yesterday, why not now" complaint. Part 4: What Actually Helps ? Some of this is genuinely outside your control, but a fair amount isn't. If you're just using these tools day to day, the single biggest habit shift is not letting conversations run forever. Long threads in Copilot Chat or Copilot Studio keep accumulating history, and that history eats into the same budget as whatever you're asking right now. Starting fresh periodically costs you nothing and buys back a lot of headroom. Large documents are worth splitting up before you feed them in, especially for SharePoint-grounded agents, where oversized files can get quietly excluded rather than cleanly rejected you won't necessarily know it happened unless you're looking for it. And it's worth resisting the urge to default to the heaviest, most capable model for every single task. Lighter models are usually faster, cheaper, and often sit under a more generous limit than the flagship ones, and most everyday tasks genuinely don't need the biggest model available. Before you go asking IT for a license upgrade, it's worth a quick sanity check on which limit you actually hit. If it's a rate limit, waiting a minute and retrying usually solves it outright. If it's a context window problem, trimming your prompt or starting a new session fixes it. An upgrade only helps if you've genuinely run out of credits or feature allowance, and that's worth confirming before you file the request. If you're on the building side Copilot Studio agents, Foundry applications, anything with RAG-style grounding a couple of things pay off quickly. Keep an eye on credit or token consumption proactively rather than discovering it's gone when the agent goes down mid-conversation. Be deliberate about what goes into system prompts and orchestration instructions, since those draw from the same budget as the end user's actual message, often invisibly to whoever's chatting with the agent. And spend real time getting chunk size right for knowledge sources too large and you're burning budget on irrelevant context, too small and the agent loses the thread. Part 5: Quick Checklist Before You Escalate Is this actually a context window problem -prompt, history, and attachments too big for the model in use? Have you genuinely run out of credits or feature allowance on your plan? Could this be a rate limit -too many requests too fast, especially from a flow or automation? Did the underlying model change since last time, quietly shifting the effective window? For Studio or Foundry work, is this a tenant or environment-level limit rather than something tied to you personally? Closing Thoughts Tokenization is one of those things that stays completely invisible right up until it isn't. Across a stack as sprawling as Microsoft's M365 Copilot, GitHub Copilot, Copilot Studio, Foundry, Security Copilot, and everything layered on top "token limit exceeded" almost never means one single thing. It means you've hit one of three very different walls, and each one needs a different response. If your team builds or maintains any of these tools, this is genuinely worth putting in front of people early. Most of the "why did this break" tickets in this space aren't about tokens at all. They're about nobody knowing which limit actually got hit, or where in this increasingly large ecosystem it happened. I'm curious how this shows up for others has your team standardized on one model across these tools, or are you juggling several depending on the task? I'd love to hear what patterns you've run into. Cheers, and happy reading. - By Surya Vennapusa, MCT
Issue Sideloading Word Extension into Word Online via GoDaddy M365 Account
Hi everyone, We’re currently facing an issue while trying to sideload a Word extension into Word Online. From what we understand, the standard process is to upload the add-in through the Microsoft 365 Admin Center, after which users within the organisation can access it through the “Add-ins” section in Word Online. However, our organisation’s Microsoft 365 account was purchased through GoDaddy instead of directly from Microsoft. Because of this, attempts to access certain Microsoft 365 Admin Center features appear to get redirected to the GoDaddy Admin Center instead. We suspect this may be preventing us from accessing the required deployment/app management options needed for sideloading the extension. Has anyone faced a similar issue with GoDaddy-managed Microsoft 365 accounts? Any guidance or workaround suggestions would be greatly appreciated. Thanks in advance.Study and Learn Agent: your study coach, built for learning
It's 11 pm. A student is at the kitchen table with a chemistry problem they can't crack, an essay due tomorrow, and a quiz in the morning. They open their laptop, open an AI chatbot, and in thirty seconds, they have an answer, an essay, and a study guide. The thinking didn't happen. The grade might still come. That moment is why so many educators and IT leaders feel a knot in their stomach about AI in the classroom. The concerns are real, and we built with them firmly in view. Now picture the same student, same kitchen table, same 11 pm. This time the Study and Learn Agent is beside them. Patient. Tireless. Knows the material because the student is studying with their own notes. Asks the right question at the right moment. Pushes them to try first, then helps them see what they missed. Quizzes them. Introduces flashcards, fill-in-the-blanks and matching activities. Helps them build and check their understanding. The student does the thinking. The Study and Learn Agent coaches the thinking. Most students have not had access to that kind of support. Potential is equally distributed, but opportunity is not. Today, with the general availability of the Study and Learn Agent in Microsoft 365 Copilot, every student K12 and Higher Education with a Microsoft Education license can get personalized coaching when they need it, where they need it, at no extra cost. Study and Learn Agent is in the left navigation and works across any subject the learner is studying. It explains concepts, supports writing without doing the writing for the students, gives step-by-step coaching on problems, generates flashcards, runs quizzes, and creates activities to build and check understanding. The agent is designed to lead with a question so the learner stays in the driver's seat. It is available in English (US) and coming to additional languages in the coming weeks. ⚙️For IT admins — read this first Study and Learn Agent runs inside Microsoft 365 Copilot Chat. For your K-12 students aged 13–17, Copilot Chat is OFF by default. You have to turn it on. Until you do, students in your tenant cannot access Study and Learn. The good news: Study and Learn is available to all licenses A1 / A3 / A5, and the agent is available from the left navigation bar of M365 Copilot app and in the Chat dropdown at https://aka.ms/studyandlearn. Action: Enable Copilot Chat for your 13–17 student group in the Microsoft 365 admin center using these resources: Full step-by-step video tutorial: https://aka.ms/enablecopilotchatvideo IT documentation: Education Tenant Identifier Student Age Groups 🎉What is available at GA Conversations that coach: Scaffolded conversations on any subject or topic K12 through HED for: Understanding concepts Working through step-by-step problems Getting writing support The agent recognizes the kind of help a learner needs and adapts the conversation accordingly. Soon, images will show inline in explanations to make abstract concepts concrete — especially for visual subjects like biology, geography, and chemistry. Practice activities that stick: Flashcards to learn terms and definitions, vocabulary, and recall facts Quizzes with multiple choice and open-ended questions with per-answer explanations Fill-in-the-blank for understanding how things work, or the sequence of connected events or facts in a process Matching activities that connect terms to definitions, causes to effects, concepts to examples — building the mental web that makes knowledge usable All activities are created with the learner’s own materials, or with just a topic name. Quizzes can also be created using a web-linked resource Students can chat with the agent during and after completing an activity to get immediate and remediation help with the agent still not giving away answers Responsible AI and data privacy: Study and Learn is built into the Microsoft 365 Education environment that schools already manage, giving IT administrators familiar controls and enterprise-grade data and privacy protection rooted in Microsoft's responsible AI principles. Students get a structured, accountable AI experience, and schools get a credible, learning-first option they can deploy with confidence. An AI literacy resource teaches responsible AI use from the very first interaction. Coming soon! Inline images: GA ships with links to images, images embedded in the flow of the conversation are coming soon Multiple languages: Study and Learn has been optimized for and Generally Available in English (US). Additional languages will become available in the coming weeks. Explanation and engagement callouts: Visual call outs for questions, tips and moments of persistence 💡Built on learning science The pedagogy is the product. The Study and Learn Agent is grounded in four research-based principles about what makes learning stick. They aren't a layer on top of the experience — they shape what the agent does in every interaction. Adaptive scaffolding: meeting students where they are by activating what they already know, then providing enough support to stretch them into what's next. In practice, this is why the agent opens a chemistry problem by asking what the learner already understands about molar mass — not by launching into a worked solution. It then tunes its support — worked examples, hints, or step-by-step guidance — to match. The result is a learner who stays productively engaged instead of overwhelmed or under-challenged. Productive struggle: asking before telling, so students retrieve, attempt, and reason their way toward answers. This is why the agent invites a first attempt before offering help, and why it surfaces a misconception as a question rather than a correction. Mistakes become data, not failure — the moments where actual learning happens. Active learning: practice that sticks, with retrieval-based activities including flashcards, fill-in-the-blanks, quizzes, and matching. The agent generates activities from the learner's own materials and lets them re-attempt the items they missed. Learners can pause during an activity and chat with the agent about a card they don't understand — building clarity in the moment, without the agent giving the answer away. Pulling knowledge out of memory, with feedback in the loop, is what builds durable understanding. Application and transfer: giving students the agency to go deeper, apply their learning, or reinforce it with an activity. This is why the agent invites learners to teach concepts back, apply ideas to new problems, and connect what they're studying to real-world contexts. It's the kind of work that builds flexible understanding beyond a single test. We built this with learning science researchers, cognitive scientists, and educators in the room from the beginning. 👩🏽🏫For educators The best way to understand what Study and Learn does differently is to spend a few minutes with it. Open it at aka.ms/studyandlearn and try it with your Microsoft Education account in the M365 Copilot app on a unit you're teaching next week. Take the professional development course at aka.ms/studyandlearnmodule to get a deep dive overview, and earn a badge! Most educators tell us this is the moment the design clicks — noticing where the agent asks a question instead of giving an answer. A few things educators in our preview have found useful: Pointing learners to it for a specific moment in an assignment can be more effective than a blanket "you can use this." Something as simple as "if you get stuck on problem 4, ask Study and Learn to walk you through it after you've taken a first attempt" tends to shape how learners engage. The activities lend themselves to specific moments. Flashcards before a vocabulary check. Matching for a unit on cell biology. A quiz the day before a test, with the agent's per-answer explanations as a self-review loop. The agent itself is a useful AI literacy artifact. Some educators use its behavior as a discussion starter — "notice that it didn't give you the answer? Why do you think that is?" — to open up conversations about how to use AI well. A short framing for learners helps a lot. Naming up front that the agent will ask questions before helping, push them to try first, and quiz them on what they missed — and why that's the design — shifts how learners engage. One small tip worth passing on: the more specifically a learner can name what's tripping them up, the better the agent can help. "I can't picture what's happening here" gives the agent more to work with than "I don't get it." Feedback shapes what we ship next. There's an OCV form linked from inside the agent, and educator input has driven much of the roadmap so far. 🫱🏼🫲🏾The bet For decades, the students who got one-on-one coaching outperformed the students who didn't. That gap was a function of access — who had a tutor, who had a teacher with bandwidth, who had a parent at the kitchen table. AI is the first technology in the history of education with a real shot at closing it. That's the bet we are making. AI as a coach. Built on learning science. Built into the tools schools already trust. Available to every student, not just the ones whose families can afford it. Study and Learn is the first move. Open Microsoft 365 Copilot. Look in the left navigation. The coach is there as long as Copilot Chat is enabled. Get going at https://aka.ms/studyandlearn Resources: Enable Copilot Chat step-by-step video tutorial Educator professional development Support documentation Anoo Padte is Principal Product Manager for AI in Education at Microsoft.How can you stay competitive and relevant in an AI-Driven World?
In a world where AI tools evolve weekly and yesterday's skills can feel obsolete overnight, this blog offers a grounded, human-first guide for cloud and technology professionals who want to stay ahead not by chasing every trend, but by building the right foundations. Across six core themes, the post walks readers through understanding what AI truly changes in the workplace, committing to deliberate and structured learning through platforms like Microsoft Learn, getting hands-on with real Azure AI projects beyond just certifications, and doubling down on the human skills critical thinking, communication, and ethical judgment that AI simply cannot replicate. The blog also makes the case for community and network as a long-term career asset, and closes with a call to develop an AI mindset rooted in curiosity, adaptability, and a willingness to experiment and share openly. Whether you're a cloud architect, a security professional preparing for AZ-500 or SC-200, or simply someone navigating what this AI shift means for your career this post is written for you. Key Takeaways for Readers: Understand AI's real impact · Build a deliberate learning habit · Go hands-on with Azure AI tools · Strengthen human skills · Invest in community · Cultivate an AI-first mindsetAccess fixes released in Version 2604 (Build 16.0.19929.20090)
Bug Name Issue Fixed Values display in the wrong control when using a form as a sublist When a form was used as a sublist (subdatasheet), field values could display in the wrong control, showing data in incorrect positions. Values now display in the correct controls. Applications that use the Access Database Engine (ACEOLEDB) terminate unexpectedly on exit Third-party applications using the Access Database Engine (ACEOLEDB) provider could terminate unexpectedly when closing. The shutdown sequence has been corrected. Long Text field corrupted when a query updates a record while a user is editing it When a query updated a Long Text field on a record that was simultaneously open for editing, the field data could become corrupted. The record update now correctly handles concurrent access to Long Text fields. Rendering errors with Aptos (Detail) font Controls using the Aptos (Detail) font variant could render incorrectly, with characters appearing misaligned or garbled. The font rendering has been corrected. Standard colors in Access didn't match other Office apps The standard color palette in Access used different color values than other Office applications like Word and Excel. The color palette has been updated to match the rest of Office. Option Group with Vertical Anchor Bottom: option buttons show incorrect visual state after clicking When an option group control had its Vertical Anchor property set to Bottom, clicking an option button would not correctly update the visual state of the buttons. The visual state now updates correctly regardless of the anchor setting. Query Design: Insert/Delete Columns don't work when ribbon is set to Show Tabs Only In Query Design view, the Insert Columns and Delete Columns commands on the ribbon did not work when the ribbon display option was set to "Show Tabs Only." The commands now work correctly regardless of ribbon display mode. SQL View: Ctrl+K should toggle pretty formatting off/on In the Monaco SQL editor, the Ctrl+K keyboard shortcut did not toggle SQL formatting. Ctrl+K now correctly toggles pretty formatting on and off. Monaco editor incorrectly converts Unicode characters in SQL view When switching between Design View and SQL View, the Monaco SQL editor could incorrectly convert certain Unicode characters, corrupting the SQL text. Unicode characters are now preserved correctly. Importing text files with Unicode characters in the filename fails Attempting to import a text file whose filename contained certain Unicode characters would fail. File imports now handle Unicode filenames correctly. Added VarP and StDevP to the Totals query aggregate dropdown The VarP (population variance) and StDevP (population standard deviation) aggregate functions were missing from the Totals row dropdown in Query Design view. They have been added alongside the existing Var and StDev options. Added VarP and StDevP to the datasheet totals row dropdown The VarP and StDevP aggregate functions were missing from the Totals row dropdown in Datasheet view. They have been added to match the options available in Query Design view. Access hangs at shutdown when VBA holds temporary DAO field references Access could hang during shutdown when VBA code created temporary DAO field references. The shutdown process now correctly cleans up temporary field references. Full Screen Mode ribbon display option does nothing in Access Selecting "Full Screen Mode" from the ribbon display options had no effect in Access. This option now works correctly, hiding the ribbon to maximize the available workspace.
Building an Agentic, AI-Powered Helpdesk with Agents Framework, Azure, and Microsoft 365
The article describes how to build an agentic, AI-powered helpdesk using Azure, Microsoft 365, and the Microsoft Agent Framework. The goal is to automate ticket handling, enrich requests with AI, and integrate seamlessly with M365 tools like Teams, Planner, and Power Automate.Cross Tenant Mailbox Migration: NotAcceptedDomainException
This week I'm performing a new cross tenant mailbox migration. I have some experience with this kind of migrations, ( it's the third one I'm in charge of ), and with the new procedure, ( will paste the link with the instructions at the end of this article ), an Azure Key Vault is no longer required, so I was very confident and thought that I would no have any issue. But, as sometimes occurs, I was wrong The setup was quite easy, and the mail users configuration was like always, so no a big deal. But now comes the point... Once I launched the migration batch, half of the users started syncing correctly and the ther ones failed, ( neither a MoveRequest was able to start for them ). Once I checked the errors, I got the same for all the failed ones: " NotAcceptedDomainException: You can't use the domain because it's not an accepted domain for your organization ". Ok. No problem... ( I thought ). I work with Exchange since more than 10 years and this is a common error message. ( Again I was wrong ). I started to check the mail users, looking for some misspelled domain, missing alias, spaces, etc... Basically, the troubleshooting for this kind of errors. But from my perspective all looked good. So, I decided to reconfigure all the mailusers with a script, launch a delta sync, and resume the failed moverequest. But again, same error for all of them. Checked again, with PS, from source and target tenant, checked in AD, all the proxy addresses... Nothing, all was correct! Non sense... Ok. At that point I decid to compare some syncing mail users with some failed ones, looking for anything that could be a pattern. And "voilá"! The syncing users were all licensed in O365... The failed ones not! After assigning a license to the failed ones and resume the MoveRequest, all started to work smoothly. For sure, I would have saved many hours of work if the error message had been: " The user is not licensed ". But, yeah... It would have been too simple 🙂 Summarizing, make sure that the mail users have an O365 license before you start the migration batch. And remember, not always the error messages are what they seems to be 🙂 Cross Tenant Mailbox Migration procedure, ( Preview 😞 https://docs.microsoft.com/en-us/microsoft-365/enterprise/cross-tenant-mailbox-migration?view=o365-worldwide
