Copilot, Microsoft’s generative AI technology brand, promises to be a big moneymaker for the company, with one analyst predicting it could generate $10 billion in annual revenue by 2026. Somewhat overlapping and confusing With the rollout, 40% of Fortune 100 companies were testing Copilot by the fall, according to Microsoft CEO Satya Nadella. Forrester pridectMeanwhile, 6.9 million knowledge workers in the US will use some form of Copilot in 2024.
Thus, Microsoft’s quest to develop technology continues.
During Microsoft Ignite 2023, the company unveiled three new Copilot offerings across its suite of software and services: Copilot for Azure, Copilot for Service, and Copilot in Dynamics 365 Guides. The company also launched Copilot Studio, a new platform that provides tools to connect Copilot for Microsoft 365 – Copilot in applications like Excel, Word and PowerPoint as well as the Microsoft Edge and Windows browser – to third-party data.
Azure’s co-pilot
Copilot for Azure, now available in preview, is Microsoft’s answer to Google Cloud’s recently announced Duet AI. Like Google’s AI, Copilot for Azure takes the form of a chat-based assistant for cloud customers, suggesting configurations for applications and environments and assisting in troubleshooting by identifying potential issues and solutions.
“Microsoft Copilot for Azure is integrated into the Azure platform, directly into the Azure portal where IT teams work,” Erin Chapple, executive vice president of product and core design for Azure, told TechCrunch in an email interview. “With a unified chat experience, they can easily ask questions, gain insights into workloads, infrastructure and cloud functions, and take action.”
Copilot for Azure can answer questions like, “Copilot, how many cloud resources do I have?” As well as more complex queries and requests such as “Copilot, which data store should I use in my request?” and “Copilot, create a command line interface to perform this action.” (Imagine OpenAI’s ChatGPT, but for Azure-related tasks.) Leveraging a set of generative AI models, especially large language models (LLMs), Copilot for Azure draws knowledge from technical documentation as well as users’ individual settings and policies.
“Using the power of LLMs, Copilot is thinking about the customer data that we have in Azure…that defines everything that has been deployed and running for that customer and all of their properties,” Chappell said. “We think about things like metrics and data monitoring so customers can ask questions and understand patterns in their data.
But given that generative AI is prone to making mistakes, can Copilot in Azure be trusted? Maybe – at least depending on your risk tolerance. Chappell says users can always ask Copilot in Azure how it arrived at the recommendation or explain its reasoning.
“Currently, we envision that customers will use Copilot for Azure for four activities: design (configure the right services for their application and environment), operate (answer questions, author…commands, and act on their behalf), and troubleshoot (coordinate data across Azure services to gain insights) and optimize (improve costs, scalability, security, and reliability through recommendations specific to their environment).”
Chappell claims that Copilot for Azure has already been used by more than 15,000 internal users and a “handful” of private preview customers. “We will continually add skills as we learn through inspection, and expand the ways they can help,” she added.
Co-pilot for service
The co-pilot for service is a different animal. Designed for customer service use cases, it integrates with customer relationship management (CRM) software including Dynamics 365, Salesforce, SAP, Workday, and Microsoft’s ServiceNow to answer sales-related questions and provide next step suggestions.
Copilot for Service—which can be embedded across the Windows desktop or deployed via Microsoft Teams—can answer common questions from customer service agents by drawing on company websites, knowledge articles, offline databases, and more. These same agents can ask Copilot for Service to provide them with account and status information from their CRM systems when they need it – and in theory – where.
The functionality seemed a bit limited at launch, but Microsoft promises a lot more to come as the service develops. The company says Copilot for Service will soon be able to provide proactive recommendations for creating or updating existing “knowledge assets” (such as articles) based on emails, customer service cases, Teams conversations, and more. It will enrich Outlook email summaries and drafts and meeting summaries in Teams with data from CRMs, giving users the ability to view and update CRM records directly in Outlook and Teams and automate tasks such as wrapping up a case, scheduling a meeting, follow-ups, and adding new contacts.
Readers may recall that Microsoft introduced its customer service-focused software, Copilot in Dynamics 365 Customer Service, in March with a comparable feature set. So what’s the difference between the two? Well, the bare minimum – except that Copilot for Service comes with Copilot for Microsoft 365 and can be customized via the new Copilot Studio. (More about Copilot Studio below.)
Microsoft says Copilot for the service will be available for public preview in early December, priced at $50 per user per month. (This is higher than Microsoft’s price for Microsoft 365 Copilot, which starts at $30 per user per month — which likely reflects High cost To run AI models at scale.) The company is targeting the first quarter of 2024 for general availability.
Copilot in Dynamics 365 guides
Copilot in Dynamics 365 Guides may be among Microsoft’s most ambitious Copilot releases.
Designed to work best on the company’s HoloLens 2 headset, with support for mobile devices to follow multiple preview periods, Copilot in Dynamics 365 Guides leverages generative AI to summarize information that might be useful to frontline workers and translate those summaries into overlays on equipment They use the province.
Using Copilot in Dynamics 365 Guides, workers can point to or look at a component and ask questions like “What is the torque limit for these bolts?”, “What direction does this part fit in?” “He showed me the steps needed to dismantle this filtration unit.” Copilot will recognize what is being pointed or indicated and provide answers, displaying instructions on the HoloLens 2’s head-up display.
The more information Copilot in Dynamics 365 Guides has access to (e.g., technical documentation, IoT sensors, service logs, operational data, notes, scripts, training content, etc.), the more industry- and company-specific questions you will have. can answer it, Microsoft says. The OpenAI generative AI models that power the service decode what’s on the HoloLens 2 camera and show graphs — or even read summaries aloud — to illustrate specific steps and solutions.
This is all interesting in a science fiction genre, like Minority Report. But I’m skeptical that the new co-pilot will work completely And so are Microsoft’s promises. Understanding images and text is difficult enough for generative AI, let alone dealing with less-than-ideal viewing conditions (dimly lit factories, for example) and specialized machines.
Microsoft’s decision to launch Copilot in Dynamics 365 Guides first in private preview with a “limited set of customers and capabilities” suggests that there are already some kinks to work out. The color intrigued me, though. I’m curious to see where this Copilot program goes — and how it measures up to the attempts being made by various startups in the field service space.
Co-pilot studio
Copilot Studio — the latest Copilot-related announcement today — is Microsoft’s attempt to make its Copilot products more extensible for business customers. Microsoft has been telegramming to fine-tune Copilot, the flagship feature of Copilot Studio, for some time, but the emergence of countless AI-powered chatbot builders — many of which can link LLMs to private company data — has no doubt increased the pressure. .
With web-based Copilot Studio, which is now available in public preview for Copilot subscribers to existing Microsoft 365 subscribers, an organization can give Copilot for Microsoft 365 and Copilot for Service access to data in its customer relationship management (CRM) systems, enterprise resource management systems, and other databases It stores data using pre-made connectors or ones it builds itself. (Microsoft plans to expand the controls in Copilot Studio to include additional Assistants at some point, but it’s starting with just those few.) With Copilot Studio, customers can also ensure that Assistants always respond a certain way to certain questions (for example, setup questions), And – arguably more importantly – creating and deploying their own tailored “copilots.”
“Copilot Studio plays an important role in Microsoft’s Copilot strategy,” Omar Aftab, vice president of conversational AI at Microsoft, told TechCrunch via email. “What Microsoft has heard from customers is that they want to customize and scale Copilots with their unique business processes and data for enterprise-specific scenarios. Now, with Copilot Studio, they can do that easily. Likewise, customers who want to build their own standalone, custom Copilot experience can now So, publish it on the channel of their choice.
Any user with the appropriate license can tap into Copilot Studio to create a personalized copilot — for example, an expense management chatbot — by describing it in natural language. Copilot Studio will provide a starting point and send the copilot into a “canvas” user interface with collaborative tools, including a commenting system and side-by-side markup views, that can be used to enhance it.
Customers can create co-pilots that filter to specific data sets for specific teams or users. Or they can connect Copilot to an automation, plugin, or third-party service to trigger actions or workflows.
Once completed in Copilot Studio, co-pilots can be deployed across a range of channels, including websites and social media platforms. The dashboard displays the operational status of copilots and who is building and customizing them, along with controls for securing copilot data using access policies and managing deployment environments.
“Copilot Studio exposes the complete, end-to-end lifecycle of Copilots within a single pane – you can create, deploy, analyze and manage Copilots all within the same web experience,” said Aftab. “And because it’s software as a service, everything you create becomes live right away.”
One wonders if Microsoft has the data center capacity to host countless dedicated copilots, Taking into account the The company’s recent struggles on this front. But Microsoft is nothing if not confident — it’s offering Copilot Studio as a purchasable capacity-based license, aiming to spur adoption of not only the Copilot ecosystem but also its ever-growing suite of Azure products.
For more Microsoft Ignite 2023 coverage:
This story was originally published at 8 a.m. PT on November 15 and was updated at 3:15 p.m. PT.