Reading：https://docs.google.com/spreadsheets/d/1JSmI-YDvbdxXJ1Tud2sK8mly33zh1VmcH8i5wUgSYs0/edit?usp=sharing
TG Channel：https://t.me/SleepinRain
Binance registration, reduced fees: https://www.binance.com/join?ref=MPQ4JG9S
Paid group: 3SOL/year, interested Twitter DM
I've mentioned $FET before, mainly complaining about @Fetch_ai's lack of market education or their previous lack of a clear product development direction. This directly resulted in the previous hype around $FET being mostly centered around the AI narrative and the merger.
But it must be said that, looking solely at the previous price performance, $FET has been a very successful Token - it surged by dozens of times in 2023 due to the hype around the AI narrative, and in 2024 it is pushing the narrative of the three major project mergers.
However, Fetch has now found the right product landing direction and has launched a large language model (LLM) called ASI-1 Mini.
Previously, Fetch was too far from the market and too far from the users. Now Fetch is moving closer to the user end.
Let's chat about Fetch's latest update⬇️
1. First of all, ASI-1 Mini is a large language model, with the distinctive features being "adaptive reasoning" and "context-aware decision-making". Let's focus on "adaptive reasoning". "Adaptive reasoning" provides Agents with four types of reasoning services: multi-step reasoning, complete reasoning, optimized reasoning, and concise reasoning. Through its underlying core reasoning engine, it will adjust the depth of reasoning based on different needs to improve the efficiency of reasoning - this is also one of the reasons why the hardware reasoning efficiency of ASI-Mini is higher.
2. The architecture above Fetch's ASI-1 Mini reasoning engine is MoM (Mixture of Models) + MoA (Mixture of Agents), where MoM is about selecting the appropriate model for reasoning based on the Agent's needs, and MoA is the Swarm AI Agent cluster narrative, i.e. multi-Agent collaboration.
3. The main application direction of ASI-1 Mini is Agentic Workflows. Simply put, Agentic Workflows is about using multiple AI Agents to collaborate, execute, manage and optimize workflows to automate and improve the efficiency of complex tasks. These complex tasks include data processing + data analysis (supporting text, images, code), forecasting (sports, Polymarket), on-chain interaction (DeFAI), etc.
4. ASI-1 Mini is the first product of the ASI Alliance (the project formed by the merger of $FET, $AGIX, $OCEAN). There will be more updates in the future. In the future, they plan to expand the context window to 10M Token (which can be directly understood as the amount of data that can be processed) to adapt to scenarios such as financial transactions, legal documents, and large databases.
5. Fetch's ASI-1 Mini is focused on using high performance + low cost to lower the threshold for Agents to run - in their words, to allow the "community to directly invest in, train and own AI".
The most core point is that the community needs $FET to access ASI-1 Mini - which is equivalent to adding an empowerment to $FET. This is also what I mentioned earlier, that Fetch is moving closer to the user end, and ultimately the AI narrative will have to land in practice.
And some recent updates on $FET (recently they've been doing Token buybacks and burnings, in the next article I want to share my thoughts on this): On January 10, the ASI Alliance burned 5M $FET, and the next burn will be three months after January 10, i.e. on April 10.
The Earn and Burn mechanism works like this⬇️
Earn: Users can earn $FET by using ASI products or participating in ASI ecosystem activities, such as developing AI Agent products through ASI Train. This process will generate fee income for the ASI Alliance;
Burn: A portion of the fee income will be used to buy back and burn $FET Tokens. The previous Fetch announced buyback and burn target should be 100M $FET.
Finally, let me share with you an information stream I've been following: https://t.me/cyra_alpha (full grasp of Web3, Web2 AI updates, I'm @0xPrismatic teacher's water boy).
