Introducing Tyk AI Studio – Welcome to the future of AI governance powered by AI chat, gateway and portal

blog-header_generic-servicemesh

Everyone is talking about AI, including Tyk customers. They’ve told us they want to deliver better AI adoption within their organizations and get greater value from that adoption. Naturally, Tyk has delivered – in the form of Tyk AI Studio

We revealed Tyk AI Studio in all its glory at the 2025 LEAP 2.0 API governance conference, with Tyk’s Budha Bhattacharya hosting a hands-on demo led by Head of Engineering Leonid Bugaev and Presales Architect Carlos Villanúa Fernández. We’ve summarised some of the highlights of the session below, so read on to discover:

  • The pain points of integrating AI solutions – and how to overcome them
  • How Tyk AI Studio empowers AI adoption
  • How to monitor and control the costs of your AI usage
  • The value of working natively with AI

AI pain points 

The arrival of OpenAI heralded a new era – one where people began using it without any knowledge of what they were doing or sharing with outside of their organization. This shadow AI usage, which lacked any kind of oversight, quickly became a privacy and compliance nightmare. 

At the same time, organizations were grappling with workflow integration headaches, trying to integrate AI with Jira, Zendesk, HubSpot and a whole range of internal systems. Businesses wanted more than just simple queries – they wanted things like chats that they could use to tap into data in a safe and human achievable way. 

Then there was the issue of cost. All tooling comes with a cost, and it’s essential to control those costs and monitor usage across departments, for AI just as for any other tool. 

The final AI pain point relates to security. Organizations that were opening data into chats suddenly needed access control to protect that data and prevent external authors from accessing it. Even within a company, not every department should have access to all data. So, businesses needed a way to control who could access what, while still harnessing all the potential that AI promised. 

Empowering AI adoption 

Tyk AI Studio fits into this still-evolving situation by empowering AI adoption. It supports teams to explore AI tools proactively, without a fear of doing something wrong or sharing information that they shouldn’t. 

By filtering the information that can travel outside of the organization, Tyk AI Studio champions data privacy. Anything that shouldn’t be shared will be filtered and kept within the boundaries of the business. 

Tyk AI Studio also helps streamline integration while providing robust security, ensuring the right people can access the right data when integrating AI with Jira, HubSpot and the like. It provides interactive capabilities, meaning you can use chat functionality to complete objectives such as creating a task on HubSpot or raising a Jira ticket. 

From a cost perspective, Tyk AI Studio ensures you know how much you’re spending on your tools, empowering you to control AI expenses and monitor usage. 

All of this perfectly aligns with what customers have been telling Tyk when it comes to AI integration. They’re looking for a centralized portal with usage monitoring, audit trails and compliance reporting, and integration capabilities with tools like Jira Zendesk and HubSpot. 

Introducing Tyk AI Studio

AI presents so many opportunities that, if you can’t keep up with the market, you’ll be out of the market. Within Tyk, it was important to us that our engineers (among others) have access to the latest AI tooling. Our goal was to enable them to be as efficient as possible, while also being as secure as possible. 

A dose of realism is important here. Employees use AI. If you tell them not to, they’ll still use AI, you just won’t have any control over it. That’s why it’s crucial to provide them with an official means of interacting with AI – such as Tyk AI Studio. 

Introducing Tyk AI Studio blog image 1

Tyk AI Studio has three main types of users: 

  • Administrators – who can see analytics that show what’s happening with AI usage and costs
  • Users – who use the chat to communicate with private data using models
  • Developers – who need API keys to work with the latest SDKs, code editors and so on

Introducing Tyk AI Studio blog image 2

You can use the Tyk AI Studio dashboard to dive into the details of how your teams are using AI. The above screenshot provides an example of this in action, showing token usage over time. You can then look at each user at a granular level, broken down into total tokens used and the total associated cost. 

Tyk AI Studio lets you analyse usage of each LLM, from the cost of using each model to factors such as the use of prompt questions and how effective they are (this can be a huge factor when it comes to code editing). You can also set a monthly budget for each LLM, with a budget assigned to each developer. Any attempt to exceed the budget triggers a notification to the administrator.  You can also set price details per model, with highly customizable options when doing so. 

The value of working natively with AI

While other vendors have developed AI proxies that sit on top of your container management system, Tyk has taken a different approach. We’ve created a separate product that works with AI in a native way. This allows for a whole range of different notions. 

Rate limiting is one example of why the Tyk approach is so advantageous. With AI integrations, it doesn’t make sense to rate limit by the token amount because you have at least four different token types. Working natively with AI makes it easy to manage this in a way that makes much more sense. 

It also makes it easy to add new models, setting a monthly budget and privacy level for each of them. 

Introducing Tyk AI Studio blog image 3

You can set privacy levels to apply to every interaction your LLM has with other tools and data sources. 

Introducing Tyk AI Studio blog image 4

The above example shows the different privacy level between Tyk’s public data – Tyk Documentation – and our internal Stack Overflow, which contains some private data and therefore requires different treatment. If a user tries to use OpenAI (for example) with the Stack Overflow data, they’ll receive an error message telling them they can’t use that data source with that model. 

Tyk AI Studio puts a range of other abilities at your fingertips. You can work with both vector data sources and file-based sources, ingesting data for use in a variety of ways. 

There’s also a hugely helpful tools feature, which supports you to carry out a wide range of functions, including offloading work to agents, while always taking account of the privacy level of any data involved. 

Introducing Tyk AI Studio blog image 5

Tools in Tyk AI Studio are based on the OpenAPI spec, making it easy to convert an HTTP service to a tool that you can then use in the chat. This approach means the tool can be parsed automatically and Tyk AI Studio will know how to use it – once you add a proper description. Want to us a different model? No problem. Tyk API Studio includes a conversion layer to handle protocol format and the like. 

Introducing Tyk AI Studio blog image 6

Introducing Tyk AI Studio blog image 6

The powerful part is that you can add tools in a matter of about ten minutes. You can ask chat to generate the OpenAI spec for you. You can also combine multiple tools in the same chat. For example, you can ask to find a Zendesk ticket, extract information from Jira tickets and combine everything in a summary. It empowers your users to be more creative and to interact very differently with all the services. 

You can add to this with filters. These essentially use scripting language for a range of tasks. You can define some of your logic, anonymize or block certain data, call HTTP APIs, call another LLM and protect against prompt injection, for example. 

Exploring the AI portal 

The AI portal within Tyk AI Studio lets you use it as your IDE. You just connect it to your AI proxy and use it with API tokens generated from the portal. It can expose API endpoints for the proxy, for use with an Anthropic SDK or an OpenAI SDK – any model that Tyk AI Studio supports. 

Developers can access the portal and self-serve. They can request access, which will notify the administrator, who can then approve the request, set the budget and so on, with everything tracked and auditable, with logs that you can export.

Empower your AI adoption with Tyk AI Studio 

Ready to find out more about empowering your AI adoption with Tyk AI Studio? You can speak to the Tyk team or arrange a personalized demo to find out more.