Azure AI Foundry Reviews

My main use cases for Azure AI Foundry revolve around employee productivity agents, building agents, and connecting it to the data ecosystem. Those are our primary use cases, and we are using that to build RAG architecture and RAG deployment for supply chain-related work as well.

Document Intelligence is the primary AI service we use within Azure AI Foundry for pretty much all the use cases, and we are leveraging it for speech recognition too. Document Intelligence is the primary service that we use.

The reason Document Intelligence is the most used one is that a bunch of our use cases go through tons and tons of documents. For instance, we recently initiated a project for all our supply chain contracts that Cox currently owns, which are close to around 60,000 documents, and then it has multiple sub-documents under that. Going through that and looking for some privacy language, whether it's defined as what we would anticipate or what we would expect, would have taken a couple of years for our supply chain team to do manually. With Document Intelligence, we just went through Foundry, enabled Document Intelligence, and we were able to get everything done in less than 90 days for the complete end-to-end solution we built on that. We see value there.

All of them are impactful, but RAG seems to be the most impactful for us.

RAG is the most impactful because it's the way we do the augmentation on the document and the way it processes the document. That's the primary thing. AI Search and RAG are complementary, so that's the primary use case for us.

We are developing AI agents.

I love the model inventory in Azure AI Foundry; it's one place for us to go review the models, click and deploy, making it a one-stop shop. We like that capability.

At the end of the day, Azure AI Foundry's capability benefits my company by allowing us to try out some new models just for research purposes right on the click. It's pretty easy to access, and we don't have to go deploy, find virtual machines, and do this; it's all in one click, and we can deploy it right from there. That's the primary benefit we see from that.

With regard to security, compliance, or governance features in Azure AI Foundry, this is something that we have started looking into, primarily using Microsoft Purview for our governance, data governance. There is this new module called DSPM for AI, and we are exploring it while trying to operationalize it with different policies and so forth, but we're still not where we want to be on the governance, AI governance side. It's a process and a path, and we are trying to work through that right now.

Azure AI Foundry can be improved from the governance perspective, as a lot can be done. The promising part is the recent announcement on the Foundry control plane. A couple of days back, there was an announcement regarding it bringing in some of the gaps that were on the platform, so it's a really positive direction in terms of where it's going. More governance is what is lacking, but the control plane will really play a big role there.

I have been using Azure AI Foundry for probably 18 months now.

Azure AI Foundry is a stable and reliable product. In the last six months, we probably had issues once in terms of stability, and it was a cloud issue, not Azure AI. Microsoft Azure had a service issue that impacted across the different regions, so that was the only thing we faced in the last six months.

Azure AI Foundry scales well with the growing needs of my company, and we are actually expanding its usage around the document augmentation project. That's how our 2026 roadmap is looking, with more projects coming in the pipeline to address those.

Fortunately, we did not have a lot of tickets put in on this AI piece of it. Outside of AI, we use Microsoft support a lot, and we are happy with the response and the collaboration that the account executive provides for us, bringing in his team and so forth.

I would rate the support from Microsoft at 9 or 9.5.

Positive

I was not using another solution to Azure AI Foundry before choosing it in the current organization, but I am used to Google's Vertex before coming into this organization, and comparatively, I prefer the Azure ecosystem much better than Google's, at least in my experience. Some of our divisions use AWS primarily, so we've been doing a comparison and contrast with AWS's Bedrock platform as well.

The deployment process with Azure AI Foundry is easy if you follow the Microsoft deployment pipeline guides that they have provided. However, there are some nuances; an enterprise having their own deployment pipeline now has to adopt this Microsoft pipeline, so we're trying to work through that. Otherwise, if you stick to the Microsoft-defined path, it works great, and there are no big issues there.

The biggest return on investment for us when using Azure AI Foundry is the complete productivity for our employees and our teams, bringing a lot of automation to the process. That's the primary thing, and it also makes our team members really excited to work on something new and pretty cool, so those are two things that boost morale.

We are happy with the pricing, setup cost, and licensing of Azure AI Foundry.

My advice for other companies that are considering Azure AI Foundry is that developers would love it because it's a well-connected ecosystem from data to AI, using all the services within Azure Foundry. The developers would love it, and from the administration perspective, if you're a Microsoft shop, it makes perfect sense to go into that ecosystem. I gave this review an overall rating of 8.

We are currently examining agentic data modeling and data lineage, focusing on the types of relationships between different data sets, jobs, scripts, data warehouses, and tables, and how we can build that into one knowledge graph. We are mainly exploring this area, but there are many different security use cases that we have been investigating with Sentinel and Sentinel Data Lake, including leveraging the MCP server and using agent-to-agent protocol for interoperability between different agents and tools.

We are deploying AI applications on the cloud, attempting to get some of them on the Security Store with Security Copilot. The ones that deal more with data are being deployed more on Azure. Agent 365 is exciting because it allows for better governance, providing visibility into agent metrics and traceability. While we do not have agents deployed yet, we anticipate having them operational within a month.

The security features primarily include governance, observability, and traceability. I recently participated in a lab session demonstrating how to perform tracing with MCP and agent-to-agent interactions in Azure. While you can definitely build something with open-source tools such as LangGraph and LangChain, Azure simplifies the process significantly because you do not need to create complex telemetry setups from scratch. Agent 365 is exciting from a security standpoint since agents can be treated similarly to any identity, such as a service principal. Furthermore, I believe Microsoft implements some of the best responsible AI practices, including guardrails and hallucination protection, which alleviates the burden of having to code those features yourself.

It is far less daunting than spending a year building or testing open-source solutions with extensive libraries, all of which is now available directly in Azure.

Azure AI Foundry streamlines the solutions and focuses on what I want to build and the results I want to achieve. There are numerous data scientists and AI engineers, but not everyone has an AI, automation, or data science background, so they benefit from Azure's integrated offerings, which Microsoft provides comprehensively.

Azure AI Foundry, alongside Agent Foundry and the classic Azure OpenAI components, offers ease of switching between them. However, I find that the online documentation can sometimes be confusing, requiring extensive searching through multiple articles to locate specific information. The tools are satisfactory, but understanding them might be challenging at times, which makes in-person events beneficial for conducting hands-on labs and enhancing my learning. I feel the documentation may benefit from clearer presentation on occasion.

I have been working in analytics, data science, AI, and automation for approximately five years, specifically in Azure AI Foundry for just about a year.

The only downtime I have experienced occurred when Azure was down overall, but I have never witnessed Azure AI Foundry being offline or having any crashes.

As a partner, we are not directly building these agents for our organization; we aim to develop them for our clients in partnership with Microsoft. Therefore, there is a return on investment because we have goals for Azure Consumed Revenue (ACR), and these models are integral to achieving those goals. I anticipate that as customers begin using these solutions, it will significantly drive our ACR and make both our organization and Microsoft happy, as the increase in ACR benefits Microsoft as well.

Azure AI Foundry makes it very straightforward, as I do not have to write thousands of lines of code; I rely on GitHub Copilot and Azure AI Foundry. I can easily set the instructions and temperature without needing to write a lot of Python code. It does not take much time to develop these agents, particularly with Copilot Studio, which allows for quick building, and I appreciate how Azure AI Foundry and Copilot Studio complement each other, along with M365 Copilot and Security Copilot. There are many copilots and agent tools, which can be somewhat confusing due to their multitude.

I utilize Azure AI Search frequently, especially with embedding models such as Ada. When dealing with unstructured documents that require processing or semantic search, I use Azure AI Search to bring that knowledge base in. Microsoft makes it straightforward, allowing me to avoid extensive coding for RAG, and the cost model is transparent, making the process much simpler than attempting to create these solutions independently, which is time-consuming and challenging.

My primary use is within Azure AI Foundry, as I operate from a Microsoft business unit. I predominantly focus on Microsoft solutions, but I do appreciate that Microsoft offers Anthropic models, which are useful for coding tasks. I find that Gemini provides a larger context window, and if I wanted to use a Gemini model, I could do so through the agent-to-agent protocol. However, I primarily stick to the Microsoft stack due to my familiarity and organizational requirements, with minimal concern for alternatives such as Google or Amazon.

I would rate this solution an 8.5 out of 10.

My main use of Azure AI Foundry is to create apps, our app backends, and integrating AI features into it, embedding AI features from there. We also use our tenant in Foundry to host the OpenAI models and then we basically have an API on a key. That is basically how we bring the AI usage in there.

We are deploying the AI application in the cloud. Our apps are hosted in Azure, at a different part of the tenant, different resource group, and all of that. We also use it in integrating it back into Copilot 365 as well.

I am not yet doing machine learning in Azure AI Foundry.

I am just connecting to the models in Azure AI Foundry. We have explored it, but we haven't done anything specific. It is all really early days.

The AI agent we are trying to create involves many agents. For example, we are creating research agents. We are a B2B transactional company, so with our Salesforce, we have 800 salespeople. They are always looking to see what is coming up for renewal, how they can go for alternative products where products are being discontinued. This is both hardware and software based products. We sell about 5,000 OEMs, 50,000 products every year to 45,000 businesses. There is a lot of data that is there, but it is very hard. We have built a simple agent that looks at the past orders in the last three months to see what is coming for renewal or anniversary dates, and then figures out what should be the top track an AM can have, and what should be the top products they can refer out to the customer. That is one example of an agent.

The most useful or impactful feature in Azure AI Foundry so far is just the ability to host our models in Foundry. They allow you to run models and have API keys there, which we are using in our apps as I said. In the future, we do plan to use more functionality, but right now it is pretty limited to just using the models.

The benefit of using Azure AI Foundry for the organization is saving time. So saving time and then making money. Now that we have this, many people were not doing this because it took them so long to do that research. That has been fixed. Because now they can see the data and talk to the customers, it is hard to benchmark that, but it obviously means that they are doing more renewals and they are able to sell more.

Assessing the stability and reliability of Azure AI Foundry is hard. When we were building it and trying to see which models to even choose, it was hard to keep track. I have seen some new tools for prompt evaluations and prompt improvement coming out, but it is hit and miss. You are really writing prompts now instead of writing code, which are language prompts, but you really need tools to test and figure out what the results came out and then have a measurable way to figure out if the prompt can be improved. Right now, it was all Excel or just try hit and miss. We are not really organized there. With code, you know what the binary result is, but with prompting, it is a lot harder. You definitely need tools that can help.

In Azure AI Foundry, I did not see many tools inside to test it that can really help us. I wish there were more tooling. Even in Visual Studio, something should take the prompt, run it on all the different models, and then show which one is a better result by itself. Right now, we had to do it by hand.

I have been using Azure AI Foundry since earlier this year.

Azure AI Foundry did not really reduce the time it took to create the app. These are new apps that we are building brand new, so there is no comparison really. We did not use AI to build the app, we were using human click-throughs, those kinds of things.

I did not use their customer service for Azure AI Foundry project, but we have used them in the past.

Neutral

I did consider another solution before Azure AI. We do work with other partners, but Azure is officially our main cloud. We are a Microsoft partner as well as a Microsoft customer and we use and host Databricks in Azure. We really did not look out. The only other solution we have used is Glean for enterprise search. But again, not for building anything that connects to beyond Office 365 data. Because this covers transactional data, we can only use Azure for this.

I was part of the deployment of starting to use Azure AI Foundry.

RAG is relevant, and RAG grounding and RAG are important. We did check them, but again, right now we are not using any of that with the database or anything.

The deployment process for Azure AI Foundry involves using CI/CD to deploy, so we have scripting with DevOps engineers. It is definitely easier to build versus deployment because then other teams are involved. Right now, because the model we are just using as an API, it is not that hard. We have a vault where we save the secrets for the keys and all. We then deploy it as a container app. That is the process that we went through.

Compared to human effort, Azure AI Foundry is very fast.

The security feature in Azure AI Foundry that has been useful is security. Because this is a separate tenant, we do have the normal Azure features trying to secure it, but we have not done much different. We always block the east to west and outward traffic. We always block, for example, to certain IP ranges and whitelist them. Sometimes we do the private links, so you have to get on the VPN to use Azure. That is usually the way of securing how we work with the cloud.

The AI service I mostly use is not really any specific ones just yet. I have not used all of them.

My overall rating for Azure AI Foundry is seven out of ten.

My main use cases for Azure AI Foundry involve deploying large language models, and we use it heavily for observability, evaluation, and fine-tuning. Those are the major ones.

Azure AI Foundry has improved decision-making in my clients' organizations, as they often use it for decision support, which is probably one of the primary use cases among many of my different clients that are working with Azure AI Foundry.

The features of Azure AI Foundry that I appreciate the most include the model catalog and the capability to deploy all of these different models, especially now that they've added Anthropic, which was the big one that was missing. I appreciate having all of that capability and having it stood up in an endpoint for me, and I really value the fact that it's in my Azure Enterprise Agreement, so all of the data protections that are there with my Enterprise Agreement are there when I deploy models.

The feature that has been the most beneficial for enhancing customer experience is the one that allows you to compare multiple models to one another and see how they perform against each other. That has been really beneficial when I'm trying to select what the actual model I should take to production is.

What's really valuable about the catalog of large language models is that it gives me a lot of different options, and it's beneficial because they exist within my Azure tenant and within the Enterprise Agreement of my clients, which is a huge benefit. Then the model comparison is really helpful because it helps me figure out what the right model is to actually deploy.

To improve Azure AI Foundry, I think I'd appreciate a more consistent story so it's easier for my clients to understand where Foundry fits in the ecosystem of Microsoft tools as well as what it can and can't do. I would also like to see the observability tooling improve a great deal. I feel it's inadequate compared to some of the other tooling that's out there, with an example being LangSmith, which I would say is the market leader in that space, so catching up in that area would be a great help.

I've been using Azure AI Foundry since before it was actually called Azure AI Foundry, so probably about four years.

I assess the stability and reliability of Azure AI Foundry as very stable and very reliable. The only time we really have issues is when we reach our capacity limits, which often appears to our customers as though something broke, but we're really just hitting capacity limits.

I have experienced downtime, crashes, or performance issues only when there have been general Azure outages, which are fairly uncommon.

Azure AI Foundry scales well with the growing needs of my organization, but we do run into occasional problems with not being able to get enough capacity because of the lack of GPU capacity in Azure. At some of our larger clients, that's been a problem, and we've had to escalate to support cases, which has been unnecessarily difficult.

To go into more detail, we've had to leverage our relationships both as a partner and as an MVP, and I've had a couple of times where I've had to get to the VP level of Microsoft before I could get the capacity I needed for my customers. That's gotten better in the last couple of years, but still, that is a common problem that we deal with.

I would evaluate customer service and technical support as terrible. Whenever I have to reach out to customer support, I end up waiting sometimes days for a response, and our 24-hour response time often turns into three to five days. We've had support cases that have lasted three to six months, so it's terrible, and we avoid using support at all costs.

There are certain expectations from an SLA perspective that my clients have, where we need a response within a certain amount of time, or at least we need to predict when we're going to get a response. Often we think we're getting one within a day, but it's a week later that we get a response, and that's pretty unacceptable, making it a real struggle to explain that to my customers.

Negative

Prior to adopting Azure AI Foundry, I was actually building my own models, so we kind of moved into a different space.

The factors that led me to consider a change were really just the way that the market has changed. A lot of the market moved from looking for machine learning solutions to AI solutions, and as we shifted over to that, Azure AI Foundry was a great place to do that work.

I would describe my experience with deploying Azure AI Foundry as very simple. It fits right in with the Azure subscriptions that my customers already have, making it very easy to get up and running and get started. It has all of the same challenges as any other Azure service, but that's to be expected, and that really depends on the regulatory needs of my individual customers.

What works really well when deploying Azure AI Foundry is being able to deploy as infrastructure as code. Leveraging Bicep and similar tools to deploy makes a huge difference, especially in reproducibility and repeatability.

I have absolutely seen return on investment with Azure AI Foundry. We've got many clients—probably between 10 and 20—that have all deployed solutions into production environments, and each one we've carefully measured ROI and been able to demonstrate significant ROI with them.

My experience with pricing, setup costs, and licensing of Azure AI Foundry is that pricing is very equivalent to what I'd expect across the industry, as many of these models are available elsewhere and the pricing is exactly the same, except I'm just getting the enhanced enterprise capabilities in Azure. Much of the other tooling, such as Azure AI Search and some of those tools, seems to be priced fairly and consistently with similar competing products, so we've been very comfortable with the pricing and our clients have been comfortable with it as well.

I most often default to Azure AI Foundry. We feel it is the market leader in the space, so unless there's something that is not available in Azure AI Foundry, that's where we start.

When evaluating the different options in the marketplace, what stands out positively is the market leadership of Azure AI Foundry because the capabilities are far ahead of everything else that's out there in my opinion at my company.

The advice I would give to other organizations considering Azure AI Foundry is that it is a market leader and this is the best product that you could possibly deploy in the space between all the cloud providers. However, be aware that especially with capacity, it's essential to plan your production capacity at the beginning of your project, because it may take you months to get the capacity you need. Make sure you have a support plan that doesn't rely strictly on the customer support of Azure. I would rate my overall experience with Azure AI Foundry as an eight out of ten.

My main use cases for Azure AI Foundry include deploying AI applications to perform document comparison, translation services, and a chat feature, helping the digital AI team at our company.

Currently, I am using some speech services, but Document Intelligence is probably one of the core services that I am using for my use cases today, as we have translation services that have been tremendously helpful in translating manuals and other materials that previously required manual effort.

I use Azure AI Search to search existing components and data that is connected to my AI solutions.

The most impactful services I utilize are Document Intelligence because that is where I see a lot of validation issues, problems, and discrepancies, especially regarding speed compared to what an actual human can do versus what Document Intelligence can produce.

I am definitely considering developing an AI agent, as that seems to be something that will be core to building workflows and continuing automation and AI initiatives.

I think we need to develop a roadmap and plan on how to incorporate Azure AI Foundry agent service at my company, and I believe it will be very useful to leverage that technology.

Azure AI Foundry has helped reduce the time taken for AI app and agent development cycles in my organization, but I think it is very new and there are a lot of unknowns. Right now, because of the UI and the complexity of it, it is complicated and cumbersome, and I think figuring out how to simplify that would probably make it a faster process.

The platform's effect on my management of privacy, performance, and compliance across different regions is quite complex because Azure AI Foundry does not make it very clear how to deploy. We set up a foundry in a region and it is hard to know how to deploy different models into other regions that Azure AI Foundry has created. We have a lot of back and forth just trying to figure that out and how to optimize our deployments.

What did not work well for us regarding Azure AI Foundry includes the security piece, being able to identify how to deploy to multiple regions, reducing latency, and managing tokens per minute. Determining the right amount of tokens and gauging that before going to production, as well as load testing, would be helpful, especially because we have load tested with the real AI and incurred high costs. Finding the right balance without high costs is essential.

The experience with pricing, setup cost, and licensing for Azure AI Foundry has been straightforward, as pricing involves tokens charged per token. The setup was fairly easy.

I have not seen a return on investment with Azure AI Foundry because we have not fully deployed it into production; we are looking to go live soon. We have not really utilized or been able to measure that metric yet, but from what we have looked at, we have probably seen a seventy percent reduction on the translation services for some things we have done that will be implemented soon. We have not captured metrics on the document comparison piece since we have not deployed it to production.

I have been using Azure AI Foundry for six months.

I assess the stability and reliability of Azure AI Foundry as relatively stable, likely giving it a rating of nine or ten.

Azure AI Foundry scales fine, as we can deploy as many OpenAI or Document Intelligence services as needed while only paying for usage, and we can load balance the traffic between those different services, which makes it straightforward.

My evaluation of customer service and technical support is through a cloud service provider, and I have not had a chance to talk to Microsoft representatives at this company, so it feels almost nonexistent.

I cannot rate customer service and technical support because we go through a cloud service provider.

Negative

Prior to adopting Azure AI Foundry, I was not using another solution to address similar needs for this project, but I used AWS PaaS and Anthropic models on previous projects.

I considered AWS, specifically Anthropic, and their Textract solution, which was similar to Document Intelligence before selecting Azure AI Foundry.

In my evaluation process, I found that Microsoft Azure AI Foundry is much more accessible compared to AWS on model selection and the capabilities of using Document Intelligence versus Textract were much better.

At this point in time, we do not even look at the observability features because we do not know what features are available, so we are just pinning it out to Application Insights and workspace analytics.

The most useful security, compliance, and governance features have been limiting token limits. Azure AI Foundry's Entra ID to Azure AI Foundry integration needs to be a little more explicit, but that is helpful because a lot of times, my team was looking at a deployment and using an access key, which is not secure. I think it needs to be more intuitive regarding access key and token versus using Entra ID to fully secure the solution, and it is not straightforward.

Regarding licensing, I did not really look at that piece.

My advice to other organizations considering Azure AI Foundry would be to understand your use case, find the regions with the lowest latency, and comprehend the complete project, including security, access keys versus tokens, and the type of services being used. Additionally, knowing how to load balance and spin up multiple services to avoid creating bottlenecks and problems with latency for your customers is crucial. I would rate Azure AI Foundry at a seven overall.

My main use cases for Azure AI Foundry are AI models and then building AI agents for our client base.

The features of Azure AI Foundry that I value the most are probably the management console, the configurability it gives us, and the visibility into what models we have and how we use those.

It is difficult to determine the direct correlation with enhancing customer experience, but probably the model management piece has had the most impact on us because we have hundreds of models deployed. That has allowed us to scale our solutions to support our customer needs.

These features have benefited our organization by allowing us to release new products and new offerings this year, accelerate our growth, and quickly drive our AI agents into customers' hands. That has been the primary benefit.

I think Azure AI Foundry can be improved with more integration with some of the low-code aspects of Copilot Studio. I'm seeing some of that at Ignite, but I haven't seen it myself in practice yet. Being able to take something from a low-code scenario into a pro scenario that we have an experience with in Azure AI Foundry would be great.

I have been using Azure AI Foundry for approximately two years.

I would assess the stability and reliability of Azure AI Foundry as very good. When I look at other SaaS platforms, it's as good or better than the other platforms.

I have experienced a couple of performance issues with Azure AI Foundry, but no significant downtime. We have high token usage at times, and so there are token throughput issues that we've experienced, and we get 429 errors when we're accessing some of the models. However, nothing that we can't engineer into our solution.

Azure AI Foundry scales with the growing needs of my organization very well. We have massively expanded usage of Azure AI Foundry, and it has been very smooth.

I would evaluate customer service and technical support as greatly improved. It was average before, but with our new customer team, it has been much better. On a scale from one to ten, I would rate customer service and technical support as a nine. We've recently gained a lot of traction, and I would rate it a ten if that continues, but the last two months have been great.

Positive

Prior to adopting Azure AI Foundry, I was not using another solution to address similar needs.

I would describe my experience with deploying Azure AI Foundry as good. We've been able to do everything we want to. We use Terraform rather than Bicep, so there have been some challenges in terms of getting that initially configured and integrated properly, but after that it was easy.

What worked well during deployment was our technical knowledge on our team. Not having that and not knowing what to do at first and how to integrate some of the more nuanced aspects of the product were the challenges we faced.

I have seen return on investment with Azure AI Foundry by being able to work with our customer representative and the models that are available to selectively choose which models we use.

My experience with Azure AI Foundry's pricing, setup cost, and licensing was a mess. It was difficult to get an understanding of how we could model out our pricing and cost over time without talking to someone. It has changed a lot, and a lot of that is outside of Microsoft's control with OpenAI models and changing prices over time. However, it has been difficult until recently to get that under control. We actually changed our customer support team on the Microsoft side recently, and that has greatly improved our visibility into this. It has still taken humans to interact and work with, which is great. So now we have a better relationship there, but I think from the portal side, there are real ways to go in terms of giving us visibility and control ourselves on some of the pricing.

Before selecting Azure AI Foundry, I considered OpenAI and third-party vendors that had their own models, things of Hugging Face environments, but Azure made sense for the work we were doing.

What stood out to me in the evaluation process when comparing these options, both positive and negative, was first and foremost, the security around Microsoft's platform. That was a huge factor for us in choosing, and then integration capabilities for other products that we already have in our portfolio.

Azure AI Foundry is a great platform to build commercial products on at scale. I would definitely recommend it for companies that have security and governance around the models and agents that they're putting together. I would rate Azure AI Foundry overall as a nine on a scale from one to ten, and I think there are some features that we saw at Ignite that would take it to a ten. Overall, I'm very happy and very pleased with the product.

My main use cases for Azure AI Foundry are developing agents that work with documents, including Q&A, RAG Q&A, and one that uses a code interpreter to work with data. I am developing agents and utilizing Azure Machine Learning for other use cases. I also utilize Azure AI Search.

Azure AI Foundry has helped reduce the time taken for AI app and agent development cycles by approximately 50% for one use case. What I appreciate most about Azure AI Foundry is the possibility to have everything in one place, including all the traceability, observability, the deployment of the agents, and the model catalog.

The most useful observability features are those for agents, allowing me to see step-by-step what the agent was doing and have an evaluation on that. Observability is the feature I value most.

The AI service I am realizing the most impact from is Document Intelligence. Document Intelligence impacts my company by processing every OCR and every document we use, as 70% is made using Document Intelligence to OCR the documents.

The most impactful use of AI Search for us is as a vector database for the indexes, where we have all our indexes in AI Search. AI Search impacts my company as it serves as the vector database and the central point for all the indexes, essentially being the central resource we use for indexing.

To improve Azure AI Foundry, I would integrate Copilot Studio and Azure AI Foundry because at the moment they are two different resources. What our customers ask for is one resource that brings together Copilot and Foundry, and that is what is missing now for a unified experience.

For Azure AI Foundry, there is no actual clear pricing structure, which can be confusing for customers to understand, as every feature that you activate has its own price, and it is not very clear sometimes to define the pricing.

I have been using Azure AI Foundry for approximately one year.

The stability and reliability of Azure AI Foundry platform are fine, but we have not used it with thousands of users at the moment.

It is probably too soon to tell how well it scales. At the moment, as I mentioned, we use Azure AI Foundry for proof of concept, and for scaling, we utilize Kubernetes and other solutions.

The customer service and technical support from Microsoft were great, especially considering it is a service that has just been released.

Positive

I was not using another solution similar to Azure AI Foundry before choosing it; we were using custom solutions or AI Studio.

The biggest return on investment for me when using Azure AI Foundry is the reduction of time, and to understand the ROI, it is possible to understand it by the time reduction. It is not possible to get it by the decentralization that Azure AI Foundry provides, so these are the two main reasons. Azure AI Foundry has been able to reduce the time for one case by half, going from six weeks of development to three weeks.

I did not consider another solution while using Azure AI Foundry because we are a Microsoft partner, so we are essentially required to use it.

We actually do not use much security functionality on Azure AI Foundry, so I think the question does not apply.

My experimentation and differentiation efforts are affected by the fact that we mostly use OpenAI, so for our use case, we did not choose other partners such as Mistral, but it is valuable to have a comprehensive model catalog.

My advice to other companies considering Azure AI Foundry is to use it for prototyping and to try it out now with the latest releases, as they integrated a comprehensive way for telemetry and integration with API management. Azure AI Foundry is growing, so I suggest giving it a try. I would rate this solution a 7.5 out of 10.

My main use cases on a day-to-day basis for Azure AI Foundry are to build proof of concepts, use cases, which include transcript summarization, document summarization, and knowledge retrieval.

The feature of Azure AI Foundry that I find most valuable is the playground.

I appreciate this feature because I get to choose different models and I get to load it with knowledge and then just try to test it out for a quick proof of concept.

The Azure AI Foundry feature that has been most beneficial for enhancing customer experience is the dashboard, which provides a way to quickly explore. I am able to use the playground and then generate skeleton template codes. We use that to quickly build out our apps and to try things out. That helps us create applications faster to help the customer.

I would assess the integration of Azure AI Foundry with existing cloud services as fairly good. My feedback is that it is not really out-of-the-box. Integration-wise, it does integrate with many Azure services, but in order to use it in Azure AI Foundry, you need to get permissions. Once you try it out, you hit into conditions and policy issues. So, in terms of experience, it is probably not very good. However, in terms of integrating with other components, I think the possibilities are excellent. From a usability standpoint, I do not think it is that user-friendly.

The first step to improve Azure AI Foundry is to get it approved and accepted by stakeholders. For a financial institution, Azure AI Foundry probably does not provide enough information for us to actually acknowledge that it is secure enough to adopt it. We cannot fully utilize it. We use it to try things out and to see what is there, but we cannot fully utilize it in production yet. Providing data on the internal workings of Azure AI Foundry would help customers like us feel more comfortable adopting it.

I have been using Azure AI Foundry for two to three years.

Regarding stability and reliability, I would say we encountered some challenges. There was a major outage with Azure. I would say it is fairly stable overall, not too frequent, but there was a major issue that was in the news before.

In terms of scalability, Azure AI Foundry scales fairly with the growing needs of my company.

When it comes to customer service and technical support for Azure AI Foundry, I would rate technical support seven out of ten. I am pleased that at our support level we are able to reach them within twenty-four hours, but sometimes it takes a while to get the right people.

Positive

We have considered other solutions that meet the same needs before choosing Azure AI Foundry, but I cannot share specifics.

Without sharing names, I think the biggest difference between those solutions and Azure AI Foundry is to minimize technology creep. That is the biggest reason why we choose to stay or lean towards Azure AI Foundry.

The deployment process for Azure AI Foundry has been very challenging. There are many issues with policies, and because we have very stringent controls, we need to know exactly how Azure AI Foundry works before we can deploy it.

The biggest return on investment for me when using Azure AI Foundry is the savings in cost for implementing our own observability, visibility, evaluation, and building our own infrastructure to do proof of concepts. Azure AI Foundry speeds up the development time, from the way to identify issues to track how models are used across the board, without purchasing other vendors. That is the biggest ROI.

Regarding the pricing, setup cost, and licensing of Azure AI Foundry, I would say it is fair, but I think it gets more expensive. We are looking at open source models and deploying local LLMs to reduce the cost. Right now, there is not anything significant, but we are exploring reducing cost by using those AI models.

Azure AI Foundry has not improved decision-making in my company yet.

My advice to other companies that are considering Azure AI Foundry is that if you are a full Microsoft shop, Azure AI Foundry is a no-brainer. However, I think it is good to keep your options open since it is a competitive space. I would rate this product seven out of ten overall.

My main use case for Azure AI Foundry is deploying custom AI agents.

What I like the most about Azure AI Foundry is customizing the agent and deploying a custom agent that is only for our company, not a shared AI.

The feature of being able to pick the right models has been the most beneficial for enhancing our customer service because some AI models are more expensive but slower, while others are faster and cheaper, allowing us to pick the right model for the right task that we are trying to solve.

Having Azure AI Foundry as a tool has benefited our organization significantly because our team has five data scientists, and this type of tool makes everything much faster. In the past, if we wanted to do machine learning and classification on our projects, we had to spend three months, but with the generative AI integration, we can finish things in a week.

Azure AI has improved decision making in our organization considerably.

I would appreciate it if Microsoft could improve Azure AI Foundry by releasing new features immediately because sometimes we have to wait for weeks or months to use them. Since these technologies are changing rapidly, once we decide on the platform and choose Microsoft, everything goes with Microsoft, which makes it difficult to move things around.

I have been using Azure AI Foundry for two to three years.

I have experienced some downtime, crashes, and performance issues.

I evaluate customer service and technical support positively because we have the enterprise license, which allows us to prioritize serious issues, ensuring that Microsoft support responds quickly.

I would rate my customer service and technical support experience as around eight. I give it an eight because sometimes the first person I talk to might not know everything, leading to an escalation to the second or third level, which results in some process overhead on the time we need to spend.

Prior to adopting Azure AI Foundry, we used AWS to address similar needs.

I would describe my experience with deploying Azure AI Foundry as smooth, as I don't have any major issues. The deployment model for Azure AI Foundry is mostly on the cloud.

We don't really focus on the financial aspect, as our company is a non-profit, but we mostly focus on engagement, working with the government and the industry. I cannot say we've seen a significant return on investment.

I would say the pricing, setup costs, and licensing for Azure AI Foundry are very expensive, but still cheaper than hiring an additional position because the AI reduces many workloads that we have to do, enabling us to use Azure AI Foundry to solve some of the tasks instead of hiring another person.

Before selecting Azure AI Foundry, we considered solutions from AWS, Elasticsearch, and other companies, but we ended up using Azure AI Foundry because it functions as a comprehensive platform, not just one software, providing all the tools that we need.

Azure AI Foundry scales effectively with our growing needs, serving our requirements successfully. We have expanded usage of Azure AI Foundry considerably. The process of expansion has been smooth, mostly just involving increased payment. I assess the stability and reliability of Azure AI Foundry as ten out of ten. My advice to another organization considering Azure AI Foundry is that Microsoft offers many AI models available that are very easy to use compared to any other cloud services. I would rate this product eight out of ten overall.

Our main use case for Azure AI Foundry is that we built a chatbot and use Foundry as a control plane where we configured the content moderation and model precision, along with many guardrails.

The Azure AI Foundry feature that has been most beneficial for enhancing customer experience is the integration of Azure Purview with Foundry because AI is data, context, and intelligence. Now that we have this integration, I want to ensure I can tag the data and model throughout the AI life cycle.

I measure this improvement by looking at the deployment of the chatbot for an insurance customer, which was providing responses very slowly. We tried everything in Foundry, but it was not improving, so we deployed Redis Cache in front of it and made a copy of all the questions from the chatbot, which made the response faster. We made architectural changes within Foundry and integrated with Redis Cache to improve the speed of the chatbot. It's similar to how your phone with Maps works—if you stay in a hotel and look up the location once, the second time it automatically asks if you want to go there. We implemented caching before Foundry and the chatbot.

The feature of Azure AI Foundry that I prefer most is the guardrails, as it is much easier than the one that Bedrock in AWS provides. The fact that it is integrated with Defender for Cloud allows me to see the metrics in Defender for Cloud, which is the CISP.

This feature benefits a company because I can get the posture of AI across the cloud, which was a missing part. People talk about the system, but I need to do the overall posture, which means infrastructure, apps, and context.

I can only assess the integration of Azure AI Foundry with existing cloud services for Azure. While I find it pretty simple, I think there is more that could be done in Foundry that is missing, as there are third parties such as AIM Security and PromptFeed that are not integrated in Foundry as of now.

I would improve Azure AI Foundry by adding more functions within Foundry itself, as right now it is quite basic in what it does. I would also enhance integration with Azure policies and Azure functions where orchestration can be done.

I have been using Azure AI Foundry for just over six months.

I think the stability and reliability of Azure AI Foundry are good, as clients are mostly experimenting and doing small-scale testing and proofs of concept, and at this point, it is working well.

I have not come across customer service and technical support from Microsoft for Azure AI Foundry, but from my past experience, it has sometimes been hit-and-miss. I have not tested it for Foundry, but in the past, you sometimes get good service and sometimes it has been circles.

Neutral

The biggest return on investment for someone who uses Azure AI, from my point of view, includes the playground and guardrails, as these are the two biggest investments. The playground is where you can deploy the model and test, and guardrails serve as the protection mechanism.

The main differences between Azure AI Foundry and other tools are that Foundry is a catch-all, one bucket, while in AWS, there are Bedrock and SageMaker, which are more spread out and do a lot more than Foundry.

Azure AI Foundry's data visualization capabilities play a pretty basic role in optimizing my operations, as I would prefer to use a third-party tool such as Tableau that provides even more visuals.

I have not utilized Azure AI Foundry's predictive analytics feature.

Azure AI Foundry has improved decision-making in my company.

I would describe the experience of the deployment of Azure AI Foundry as not challenging, but I think it is still in a growing phase and not enterprise-grade in terms of doing a lot of things. However, I need to understand how intuitive it is to configure and set up.

My advice to other companies considering Azure AI Foundry is that it should be a foundation, and then you have to supplement it with something else. Start with the foundation, which is Azure AI Foundry, and then build on top of that, as Foundry is not the only solution. I would rate this review overall at 6.5 out of 10.