Forge LLMs API is now GA: Here’s how you can build AI-native apps on Atlassian

Today, we’re announcing the Forge LLMs API is now generally available for all developers. Large language models are now everywhere—embedded in the apps we use every day. They provide the intelligence that reduces manual effort, surfaces information, and helps us work more effectively. The Forge LLMs API lets any developer bring that intelligence directly into their Forge apps, quickly, easily, and all from inside the Atlassian trust boundary.

It works by giving apps access to large language models through a single developer-friendly API. You can add powerful AI features to your app while Atlassian handles the model infrastructure and heavy lifting, and it doesn’t require data egress from the Atlassian Platform. Apps that use the Forge LLMs API can qualify for Runs on Atlassian because app data stays contained within Atlassian cloud the entire time.

Since the Preview launch in June, adoption has taken off. More than 100 unique apps using Forge LLMs have launched in production since Preview, and twenty of those are public Marketplace apps already serving customers today. These apps are solving real problems for Atlassian customers: risk management, diagramming, workflow automation, and content quality. And with LLMs running on Forge, customers get the AI features they need with the trust they require.

With this launch, Forge is becoming the platform where developers build AI-native experiences on the Atlassian platform. We’re investing in new capabilities that make it easier for developers to build intelligent apps and agents in Atlassian’s ecosystem. The Forge LLMs API is now part of that broader set of AI building blocks, alongside custom Rovo agents and skills, with more to follow on our roadmap this year.

Here’s what’s available, how developers are already using it, and how to get started.

Amazon Bedrock is now natively built into Forge

Under the hood, the Forge LLMs API runs on Amazon Bedrock, AWS’s foundational layer for generative AI. Bedrock is now embedded in Forge’s core infrastructure, giving developers convenient access to large language models.

“With Amazon Bedrock integrated into Forge, developers can build AI-powered Atlassian apps faster while keeping data inside Atlassian’s cloud boundary. It’s a meaningful step toward making enterprise-ready AI easier to build, deploy, and trust.”

Jamie Simon, Director of Partner Sales & Development, Asia Pacific, AWS

Today, that model selection includes three Claude model tiers: Sonnet, Opus, and Haiku. With GA, we’ve upgraded support to Sonnet 5, and Opus 4.7 and 4.8 are now available. Bedrock’s support for a wide range of models means we plan to expand this selection in the future. Developers can call these models using a simple SDK—all it takes is a few lines of code and you don’t need to manage API keys.

But let’s talk more about data management. Because models are running inside the Forge platform via Bedrock, customer data never travels to a third-party AI service. Your app can qualify for the Runs on Atlassian badge, which helps enterprise customers find apps that meet strict data egress requirements.

This removes two hard problems that developers face when building AI features: standing up your own AI infrastructure, and making sure enterprise security teams trust a third-party AI service.

Real apps, real impact

Forge LLMs has been in Preview since June, and over 100 apps using it have launched to production, with over 350 more in development. When we look closely at these apps, there’s clear evidence that the Forge LLMs API is helping developers ship AI features faster, with new solutions going live within weeks.

It’s also important to note who is building. It’s no surprise that Marketplace partners are at the head of the pack, weaving Forge LLMs into both new and existing apps. But customers are also building their own apps, using Forge LLMs to power bespoke, internal use cases. Both types of developers are finding new ways to deliver more value to their end users with AI.

See what developers are building

We already see patterns emerging in the apps that are hitting the Marketplace. Developers are using Forge LLMs to make complex tasks simple and surface information right where users need it. Many apps are tackling workflow automation and creating structured content out of open-ended user input, which AI is especially suited for. We’re especially excited to see domain-specific use cases taking shape, from software testing to procurement.

Here are just a few of the Marketplace apps already powering AI experiences with Forge LLMs:

• Jira Workflow Toolbox by Decadis AG – No-code workflow automation for Jira
• Smart Terms Glossary by Creativas: Creates an enterprise glossary of terms across Jira and Confluence
• Procuro – Complete Procurement Management in JSM by Deviniti – Handles the end-to-end procurement lifecycle inside JSM
• Shoppie – Procurement & Shop for JSM by Caelor – Users can browse and purchase Assets in an intuitive e-commerce environment
• VibeTest by AppdashAI – Turns Jira issues into runnable tests, executable by your AI agent
• Smart Metrics Ai by SmartFlow Studio – Surfaces advanced engineering metrics, insights and bottlenecks, directly within Jira

“Shoppie transforms request management in Jira Service Management into a shop-like experience that people actually enjoy using. With Forge LLMs, we were able to ship AI-powered translation for Assets in just a few days, without having to build or operate our own AI infrastructure. That let us stay focused on delivering customer value, not platform complexity.”

Marin Varvodic, CEO, Caelor

Get started today

With the move to general availability, Forge LLMs are now a core part of how developers build on Atlassian. Both Marketplace partners and customers are already building and shipping to production, and GA makes it official, fully supported, and ready to scale.

If you’re already building apps that call an external AI service, Forge LLMs is now a great alternative, with the added benefit of keeping data inside the Atlassian trust boundary. And if you’re building a new app, the Forge LLMs API might be just what you need to make it smarter, faster, or more useful.

All you need to get started is adding the Forge LLMs SDK to your app—that’s it. If you’re already building on Forge, you don’t need new credentials, a separate AI account, or changes to your app’s billing setup. Visit the docs to learn more, and sign up for the livestream on August 4th for a walkthrough of real examples.