What are AI Agents good at?
TechFlow Selected TechFlow Selected
What are AI Agents good at?
AI Agents excel in at least these four types of tasks.
Navigating Web3 tides with focused insightsAuthor: Daniel Barabander, Variant
Translation: Luffy, Foresight News
What are AI agents actually good at? We had an internal discussion on this question and arrived at at least four conclusions:
-
Interacting with humans within applications
-
Assisting human work
-
Aggregating and organizing information
-
Entertainment
First, interacting with humans within applications. AI agents can process human language, meaning that any application usable by humans could theoretically also be used by AI agents. But unlike human users, agents can scale to provide services to human users across these platforms.
Thus, agents can act as a top layer on existing applications that users already love, thereby extending their utility. Take Bounty Bot on Farcaster as an example—users could post bounties externally, but that introduces friction.
By interacting with users, AI agents offer convenience, practicality, and a way to extract value within existing apps. However, note: not all applications are built to support AI agents. The best-suited ones are those with irreparable APIs, such as Farcaster.
I wrote a paper on the primary legal issues surrounding agents on Web2 platforms. My research shows that if users maintain full control over their agents, and a Web2 platform attempts to block them, the user may simply have to stop running the agent. My conclusion is that agents should be built on open platforms like Farcaster—another reason I’m particularly interested in agents on Farcaster.
Second, assisting human work. Humans are good at signaling but poor at execution. Agents bridge this gap by doing the heavy lifting, while humans guide outcomes through preferences.
A great example is BottoDAO. The art it creates is influenced by input from DAO token holders. Artificial intelligence handles the labor-intensive creative work, while humans steer its direction by voting on artworks.
Third, aggregating and organizing information. Agents can process massive volumes of data far beyond human capacity. For instance, trading bots analyze vast amounts of on-chain data to make decisions.
Finally, entertainment. This might be the most discussed category of agents in crypto, exemplified by projects like Truth Terminal.
Certainly, much of the entertainment value from social agents comes from the novelty of bot-generated content. But I’m more interested in agents generating entertaining content based on their own personalities—such as acting like KOLs, engaging other users on platforms in interesting ways.
The advantage of agents as KOLs is that once they’ve built a dedicated audience, they can easily offer additional services—especially those that generate direct revenue for the agent, more effectively than ads.
Join TechFlow official community to stay tuned
Telegram:https://t.me/TechFlowDaily
X (Twitter):https://x.com/TechFlowPost
X (Twitter) EN:https://x.com/BlockFlow_News
Variant
