BeyondATC has released a new update video detailing the latest advancements in its AI-powered air traffic control solution for Microsoft Flight Simulator. The video highlights the progress made in large language model (LLM) integration, improvements to AI traffic, and the introduction of an interactive tutorial system.
LLM integration moves to Early Access
The most significant development is the migration of BeyondATC’s custom LLM from the Experimental branch to the Early Access branch. This means that all users now have access to the LLM-powered interactions, which have been well-received during the experimental testing phase.
According to BeyondATC, the LLM implementation has performed exceptionally well, with minimal impact on system performance and overwhelmingly positive feedback from the community. Users have noted a substantial improvement in the way the ATC understands and responds to commands, resulting in more dynamic and engaging exchanges. The LLM now allows for more natural interactions, with controllers responding in a conversational yet professional manner.
The move to Early Access does not indicate that the LLM is entirely bug-free, but the team believes it is stable enough for general use. The update is now available to all users for free.
Advancements in AI traffic behavior
BeyondATC continues to refine AI traffic management, addressing challenges such as nose-to-nose conflicts on taxiways. The update introduces predictive movement logic, allowing aircraft to anticipate player movement and adjust their taxi behavior accordingly. In practice, this means that AI planes will now hold their positions or receive explicit ATC instructions to prevent unnecessary standoffs.
The update also brings enhancements to takeoff sequencing, ensuring that aircraft clear the runway without unnecessary delays. This includes the ability for departing aircraft to take off immediately when given clearance while another aircraft is on short final, eliminating unrealistic hold-ups. Additionally, backtrack functionality has been implemented, allowing aircraft to use backtrack runways correctly before departure.
A unique aspect of this update is the incorporation of LLM-driven radio chatter. When air traffic is light, AI pilots may engage in longer, more natural-sounding transmissions, adding to immersion. However, during peak traffic, call-ins will be automatically shortened to maintain realistic radio discipline.
New interactive tutorial system
For new users, BeyondATC is also introducing an interactive tutorial system designed to guide pilots through the basics of using the program. The tutorial includes narrated step-by-step instructions covering IFR and VFR operations, with an emphasis on making ATC communication more accessible. The developers have also injected a touch of humor into the tutorial, making it both educational and entertaining.
Roadmap and future plans
With the core LLM and AI traffic framework now in place, BeyondATC is shifting focus to the next stages of development. The team plans to release a community survey in the coming weeks to determine which features should be prioritized next. Additionally, VFR support remains a key goal, with work continuing alongside ongoing bug fixes and refinements to traffic sequencing and vectoring.
BeyondATC says that while new features are being developed, maintaining and improving existing functionality remains a top priority. As the team continues to refine and expand the software, they encourage users to provide feedback and participate in the upcoming feature voting process. For now, simmers can enjoy the latest enhancements by updating to the newest version of BeyondATC.
