Copilot Studio

“I’m working on building some AI agents for our documentation team and wanted to understand how much we’re actually spending on tokens/credits. We have a pretty large amount of documentation: 50+ user guides, around 150+ pages each 30+ release notes, around 700–1,000 pages each I uploaded all of these as knowledge sources in Copilot Studio so the agent can search through them when answering questions. We’re using Claude Opus 4.7, and honestly, the results are really good. It can usually pinpoint ”

“The base model for MS Copilot Studio is a unbelievably stupid, lobotomized Chat-GPT. The free Chat model on OpenAIs website has more capabilities to analyze files and will answer questions better than Copilot. Gemini AI Studio is also better. Just test by asking the same questions on both platforms. You would not believe how limited Copilots model is. My experience with it: - It can't read data connected via Copilots internal file database or connectors to specific plugins without throwing awa”

“Ultimate AI Assistant | It has performed exceptionally well where I have required deep contextual understanding , multi deimension analysis or web research. Recently it helped me in the implementation of SAP Project. It helped to prepare documentation like business blue prints, conduct workshops. Understand business requirement with the help of reviewing the information by summarizing. | Pros: Doesn't require extensive coding skills | Pros: Best at integrating with other systems | Pros: Efficie”

“I like that Microsoft Copilot Studio allows me to build and manage customized AI agents to streamline my workflow and automate customer service inquiries. It securely grounds my AI assistant with my enterprise data and connects seamlessly with my system to handle complex business tasks. My favorite part is the balance it strikes between low-code accessibility and enterprise-grade sophistication. The native ecosystem integration, especially with Microsoft Teams and Power Automate, works exception”

“I was testing Copilot Studio deployments, so I set up separate dev and prod environments on my developer tenant. To publish Copilot agent , I turned on the Pay-as-you-go plan. That was on 06/22. I then deployed one agent from dev to prod. That's it — a single agent. Today I checked my Azure subscription and there's a bill for ~$47,000. I'm a developer by profession — I do not have that kind of money sitting around, and there's no way a single test agent should generate a bill like this. Somethin”

“Pay as you go without a budget cap? Slow clap...”

“The fact that it blew past your budget cap is what gives me fear. I’m hoping for you it is a billing error. Please do update about this.”

“I get the reason for the change (though the blog post banging on about harnesses will no doubt cause confusion among customers). But the sudden cliff edge seems unreasonable, surely there's something you can do that halts extreme use cases from smaller one's? I'm currently working through a course which covers some of the new Studio features. With this release all my labs are now locked up across my development environment and tenant. Can't publish any workflows or agents as I don't have "suffic”

“Braindead take. You really think non-devs are actually going to use VS code/ CLI tools? This platform isn’t made for the people who use those tools. It’s made for the finance manager who wants a year end true up agent to share with his team. I swear some people who shit on copilot are dummies.”

“It's Microsoft. The marketing doesn't meet the reality. They need to improve it. Especially knowledge source handling and their support team often can't help either. Such a shame as it has potential. But then so did the Surface Duo. Copilot Studio could do the basics but for something as simple as reading a SharePoint list I had to integrate Copilot Studio with Python and an OpenAI LLM. It should just be easier to work with connecting to Microsoft services because that's the USP. Bui”

“This is a nice advice for actual developers, but completely useless for non-IT-people. The simplicity of Copilot Studio outweighs the smaller range of capability in most small cases.”

“Microsoft Copilot Studio changes the enterprise game. | Where you have a very finite and specific use case, that matches well with the intended use case of the product, it tends to be a great fit. For example we created a plain language chatbot which can rewrite any given piece of text into plain language, referring to specific internal standards. | Pros: Agentic workflows | Pros: AI chatbots | Pros: Deploying via Teams | Pros: Multiple knowledge sources | Cons: Ability to combine multiple sourc”