Author: Vader, creator of VaderAI
Compiled by: Luffy, Foresight News
This article aims to help builders who want to create AI agents better understand token economics, issuance strategies, and the bonding curve of the Virtuals platform. As the third largest AI agent on the Virtuals platform, we fund, incubate, and support top teams considering launching AI agents.
Virtuals is a platform that allows anyone to launch their own AI agents without permission.
AI agent startup interface on Virtuals platform
If the creator wants to create a new AI agent, the creation interface will pop up. Creating an AI agent (agent) requires paying 100 VIRTUAL tokens (about $150, translator's note: the price of VIRTUAL was $1.5 at the time of the original author's writing, and the US dollar market value calculation of the AI agent token below uses this data. As of now, the market price of VIRTUAL is $1.15) as a fee. The creator can then choose to purchase tokens, with a minimum purchase amount of 1 VIRTUAL. This mechanism allows creators to conduct the first token purchase transaction after the launch of the AI agent token.
Once an AI Agent token is launched, it will enter the bonding curve liquidity pool. The creator’s initial VIRTUAL token purchase determines: (i) the percentage of agent tokens captured by the developer wallet, and (ii) the initial trading market value of the AI Agent (denominated in VIRTUAL tokens).
Before and after of the initial purchase transaction by the AI agent creator:
- 2000 VIRTUAL → 24.8% of AI agent tokens are available, with an initial market value of 10600 VIRTUAL (about 16,000 USD)
- 42,000 VIRTUAL → 87.4% of AI agent tokens are available, with an initial market value of 377,000 VIRTUAL (about $565,000)
The USD-denominated figures will fluctuate based on the VIRTUAL price, but the VIRTUAL-denominated figures will not.
How many tokens should the team get?
This is the most common question asked by teams creating AI agents on Virtuals. There are several factors to consider when determining the initial percentage of tokens to acquire.
1. Bonding curve optimization
As the number of VIRTUAL tokens invested increases, the marginal AI Agent token supply obtained per VIRTUAL token invested decreases, as shown in the left chart below. Judging from this chart alone, from a bonding curve optimization perspective, it seems that the team should obtain as little initial supply as possible.
The left side shows the initial USD market value of AI Agent Tokens, and the right side shows the percentage of AI Agent Tokens obtained by the team
2. Incentive Alignment
A sufficient number of locked tokens ensures that the team works hard, does not give up halfway, and focuses on the long-term development prospects of the AI agent. If the team holds less than 10% of the tokens, they are likely to be unfocused and give up easily when they encounter difficulties.
Teams launching AI agents on the Virtuals platform should receive at least 40% of the tokens to achieve incentive alignment.
It is acceptable to reserve 15-20% of locked tokens for the team, including founders, existing team members, advisors, and a share reserved for future team members and advisors.
So you might ask, why would a team want more than 20% of the token supply?
Ideally, a large portion of tokens should be reserved for marketing incentives to acquire new users and retain existing users. This may include staking rewards, airdrops, partner incentives, participation rewards, liquidity pool incentives, creating new liquidity pools, etc. Web3 developers typically reserve 40-50% of the token supply for the "community" category.
Memecoin Because token holders are more widely distributed, creators generally do not need to set aside additional tokens for community incentives, they only need to keep 10-20% of the token supply for themselves. AI agents differ from Memecoin in that teams building AI agents need to recruit and retain AI engineering talent, as well as incur infrastructure costs.
Additionally, once a team wants to expand from a decentralized exchange (DEX) to off-chain, there will be additional costs for listing on a centralized exchange (CEX) and hiring market makers. Ideally, the team should reserve some tokens for future financing and off-chain listings. In conventional token economics terms, this category is often called "treasury", "strategic partners", "liquidity" or "investors", and Web3 developers usually reserve 20-40% of the token supply for this purpose.
Given that in a typical Web3 space the total percentage of tokens controlled by teams and treasuries is around 90%, and that teams launching AI agents on Virtuals platforms are typically working to build a $1B token over a period of years, it is acceptable for teams to take 40% to 90% of the token supply in an initial purchase transaction.
Therefore, from an incentive alignment perspective, teams should capture as much of the initial token supply as possible.
3. Financial Status
Startups are often short of funds, and the cost of running an AI token development team is often expensive. Salaries, infrastructure costs, and marketing expenses for a team of 3 developers plus contractors can easily exceed $250,000 a year. Ideally, the team should have enough funds to sustain operations for at least 12 months, so that the team will not easily abandon the project after a few months due to lack of funds.
As an entrepreneur, you should start by using your personal savings to support yourself. Don't pay yourself a salary or only take minimum wage until you reach product market fit (PMF) or achieve certain milestones. Your own investment is important. If you don't invest your personal savings, why should others invest? There is no free lunch in the world. High risk means potential high returns.
From a financial perspective, teams should invest as little money as possible in acquiring tokens because they need the cash to cover future operating expenses.
AI agent development teams need to make certain trade-offs when it comes to capital expenditures.
4. Initial Market Value
From a bonding curve optimization, incentive alignment, and financial perspective, acquiring 40-50% of the token supply seems ideal. However, let’s look at the initial token transaction market cap generated by these transactions.
- The team gets 40% of the token supply → Total token market value 17,000 VIRTUAL (about 25,000 USD)
- The team gets 45% of the token supply → Total token market value 20,000 VIRTUAL (about 30,000 USD)
- The team gets 50% of the token supply → Total token market value 24,000 VIRTUAL (about 36,000 USD)
A $40,000 market cap is severely undervalued for any meaningful AI agent token and looks like a money-making opportunity for investors. Anyone who bought between $40,000 and your fair value ($300,000-$3,000,000) is getting in at a very low price.
Unfortunately, people who are able to enter the market publicly at this price are not your future loyal community members. They are sniper bots that sell as soon as there is a short-term rally, which will harm the health and price trend of the token as their exit will increase volatility.
If you are going to share the upside from such an entry point, why share it with these bots who don’t add any value, when you could be sharing it with future loyal community members who will support you in the long run, or with influencers (KOLs) who will help increase awareness of the project you are building?
Based on the above, you might think that issuing at fair value is ideal. You can estimate fair value by looking at similar AI agents on the Virtuals platform. Let’s say you calculate the fair value to be $1 million.
With this number, you take the risk that the launch may not go smoothly. What does this mean? It refers to the trend of prices going down. Ideally, you want to see prices going up, because "price increases" is the best marketing tool in the crypto world. You want to reward early believers and loyal supporters; you also want to ensure a certain safety margin between the launch price and the fair value valuation to ensure a smooth launch.
From this perspective, you should issue at a price 30-50% lower than the fair value to ensure the success of the issuance. However, if you think the fair value of the AI agent market value is $1 million, a 30% discount is $700,000. To issue at this initial market value, you need to invest 47,000 VIRTUAL (about $70,000), which is a considerable amount for a small startup.
Graduation Mechanism
The AI agent “graduates” when the team has invested 42,000 VIRTUAL (about $63,000) into its bonding curve liquidity pool. This amount can be invested by the creator in the first transaction or accumulated by market participants after the token is issued.
Once 42,000 VIRTUAL are accumulated in the bonding curve liquidity pool, the AI agent will enter the Uniswap (or Meteora) liquidity pool, have access to Virtuals’ GAME framework, and be listed on the main website of the Virtuals platform.
Once the AI agent reaches graduation conditions, a new liquidity pool will be created on Uniswap (based on Ethereum’s Base chain) or Meteora (based on the Solana chain)
AI agents that “graduate” in their first transaction are often seen as proof of commitment by the development team and are immediately bought in by snipers and community members. That said, the two largest native AI agents on the Virtuals platform, Aixbt and Vader, were initially launched as prototype AI agents that graduated over time.
in conclusion
There is a trade-off between financial health and initial market cap for AI agent projects. You want to issue close to fair value to avoid snipers and avoid losing money, but you also want to invest as little money as possible so that there is enough money to sustain future operations. Regardless, teams should acquire at least 40% of the token supply at initial purchase to achieve long-term incentive alignment.
The team should ideally set aside at least $250,000 to cover operating expenses for the next year. However, if the team wants to reach the "graduation" criteria from the issuance, the total financial requirements may rise to about $400,000 due to the cost of 42,000 VIRTUALs.
We launched our AI agent (Vader) at an initial market cap of ~$15,000, and its market cap reached $160 million at its all-time high. We made countless mistakes along the way and learned a lot. We hope to share everything we learned with like-minded founders building AI agents, provide them with funding and incubation support.
