ai16z's Shaw has announced the integration of Hierarchical Task Networks (HTNs) into its AI platform, Eliza.

On January 6, Xu, founder of ai16z (AI16Z) to X to indicate that the team will also fund the development of HTNs integration into Eliza Second version. His response came as a response to fellow X member @lordOfAFew, who pointed out the various challenges that AI agents face on a daily basis.

Loaf is one of our largest contributors to the Eliza repo.

We will be integrating hierarchical task grids into Eliza v2, and we are funding this development.

Huge for games, wallets and any kind of complex series of actions planned. https://t.co/GV1pUwQ5yR

- Shaw (@shawmakesmagic) January 6, 2025

Challenges of artificial intelligence agent networks

AI agents are the future of all AI companies, Xu said at Delphi Podcast. He also stressed the need to make AI customer networks strong. However, currently is the hardest part for both of them I have an agent The network selects targets over a long time frame, especially when agents are first selected. Natural language cues constrain agents to follow simple directions. It is possible for agents to trade with infrastructure by responding to signals, but this does not give them freedom and independence.

One of the most significant challenges facing any agent framework is the problem of time-horizon goals. Agent development generally evolves through three stages:

Stage One: Agents rely on natural language prompting to carry out straightforward tasks. For example, “Trade 1 ETH on…

— 🍞☁⛩️ (realms/acc) (@lordOfAFew) January 6, 2025

This presents the biggest problem because it prevents agents from acting on their own to reach such complex long-term goals. For this to work, complex task networks and the ability to explore and change strategies on their own are needed.

Achieving Stage 3 autonomy requires hardware, firmware and software changes, which is a difficult task. Only through rapid improvement in technology, such as HTNs, can agents handle complex, long-term tasks.

HTNs are systems that break down difficult tasks into smaller, manageable parts, which are then grouped together in a tree-like structure. Even when things are not clear, AI systems will be able to plan and execute a complex set of tasks quickly and effectively.

For example, in the gaming industry, HTNs can allow NPCs to act more strategically, thus delivering dynamic and immersive experiences for players. For digital governorHTNs may simplify the steps involved in the process, such as payment authorization, Fraud detectionand balance adjustments.

How can HTNs help Eliza Labs?

The addition of HTNs to Eliza v2 means that the platform will finally be able to, among other things, perform actions that involve multiple steps, update plans in an instant when changes are made, and handle operations in an industry-level way without breaking a sweat. . HTNs represent a clear manifestation of Eliza Labs' ultimate goal, which is to create AI systems that not only react, but go forward with intelligent, proactive planning that fundamentally bridges the gaps to solve real-world problems.