People seem very bullish on @Autonolas $OLAS.
Why? Just because it's an 'AI' project?
Or because AI services will transform our lives, and @autonolas ensures 'we' control what AI services get built on and off-chain?

After days of research and lots of ☕️, I found this 🔽

In this thread
🔗The Lay of the Land
🔗The Solution
🔗The Adoption  
🔗The Team
🔗The Tokenomics
🔗The Competition
🔗The Reasons 'Why Not'
🔗The Road Ahead.
🔗The Conclusion
(If you make it to the end, I will buy you a beer🍺)

🔗 The Lay of the Land
To understand @autonolas, we need to get the vision of CEO David: he acknowledges the endless potential AND the existential threat of AI services.
AI services will free people from routine cognitive labor and force them to find new, meaningful jobs.

Yet, the value created by these new autonomous fleets will be enormous and too much for any single entity to manage without systemic risks to the balance of society.
Orwellian scenarios could occur fast if one entity has control over what is deployed and for what purpose.

David's solution: Co-owned AI.
What?
Yes, AI. Owned by many.
So, you mean like a DAO?
Kinda. They are alike. Collectively operationalized. Devloped decentralized. With its governanace democratized

Yet, in their curent form, DAOs are not autonomous. Nor decentralized.
Why?
Because DAOs rely on smart contracts that rely on transactions to function.
Even if smart contracts represent AI Agents, you need entities (aka people) to sign transactions on the blockchain.

Look, smart contracts are fantastic for simple, transactional tasks. (ex yield payouts, etc).
Yet, they cannot:
1️⃣Do things that require expensive processing.
2️⃣Run continuously
3️⃣Interact with the off-chain world

Also 🔽
Crypto protocols mainly do two things:
▶️Implement their own chain to solve issues of other chains
▶️Add functionality to the chain (ex DEX).

Both approaches leave a big gap for a tech that connects both the on-chain and off-chain worlds.
And there was @Autonolas...

🔗The Solution

The goal of the $OLAS protocol is to unlock the potential of off-chain services (read: autonomous services that run off-chain but interact on and off-chain, without constraints)

Before @autonolas, this wasn't possible! (degens; Ah, first mover😜!)‍‍

These autonomous services are groups of AI agents that form networks to create a (autonomous) service. They:

1️⃣Run continuously
2️⃣Take action autonomously
3️⃣Interact with the world outside of blockchain
4️⃣Run complex logic

To be clear, they don't need us to function. Huge!

Btw you might wonder: What exactly are AI agents? (I did)
Here is a definition:
'AI that acts in an intelligent manner; It perceives its environment and takes actions autonomously in order to achieve goals.'
It is just smart software acting on its own.
Back to @autonolas.

@Autonolas tech stack includes:

▶️Autonomous services (as mentioned)
▶️Composable autonomous apps.
▶️On-chain protocol that secures agent services and incentives development.
Let's discuss them individually.

▶️ Agent Services
Now it gets awesome
The AI agents that make up these services can pull data from any AI model in the world. Every GPT, LMM, or subnet. (yes, also $TAO)
Via service orchestration, specific models are assigned to certain agents.

The selection is determined by a few factors, such as economic viability, availability, and performance.
Every agent can use different data sources (more decentralization).
In some cases, knowledge graphs are even distributed via Origin Trail or through the graph via GraphQL.

Agents can be used to perform many Web3 tasks:
1. Oracles - Agents can be utilized to provide accurate data to smart contracts
2. Autonomous keepers service (AKS) - Adds off-chain computing and reasoning to DeFi jobs.
3. Via ACN, AI agents can communicate to work in teams.

▶️ Composable autonomous apps.
Another awesome thing.
The tech stack is built to encourage composability.
Coders can build basic applications that can be used as building blocks for more sophisticated AI agents. All open-source.
(I will get back to this in a bit)

▶️ On-Chain protocol
@Autonolas built an on-chain protocol (set of smart contracts) that functions like an app store:
It lets developers register and monetize their services. It allows those services to interact economically, and members vote upon the rules.

Enough technicalities; what types of services are we talking?
The big thing is that although the primary focus is DAOs, these services can be deployed in the 'real' world as well.
Think plantation stations using IoT sensors or AI agents use data to trigger watering jobs.
🤯

Back to DAOs.
A DAO has internal operations (ex, payroll, governance) and service delivery (ex, yield aggregator).
AI agents do all these things
* Better
* Faster
* Autonomously

What other types of services could AI agents improve?
* Managing liquidity pools
* Prediction markets
* Asset acquisition
* Oracles and bridges
* Taks fulfillment
Leveraging AI, @autonolas allows for creating services that radically increase the performance of stuff.

Let's take prediction markets, for example.
With @autonolas, you can create AI agents that scan for entities that make predictions using GPT4 in order to increase your win rate.
There are already AI Agents with a win rate of 88%
And these AI agents keep improving...

Future use case:
1️⃣With @autonolas, an insurance service is created (for farmers)
2️⃣A group of people start running AI agents that execute that logic
3️⃣A farmer registers for that service and pays an annual fee
4️⃣That farmer wants a payment because of seasonal losses.
5️⃣Every agent on the network gets climate data from different AI models trained by SMEs (subject matter experts).
6️⃣Agents share data with each other, validate, and vote to reach consensus leveraging ACN and MAS.
7️⃣Payment is made if the request is in line with the data.

This use case displays the uniqueness of Autonolas.
Open autonomy is a new standardization framework for creating autonomous services. This enables agent services that are blockchain native and decentralized to work in teams as multi-agent systems (MAS).

Fun fact: CEO David created the first open-source, crypto-friendly MAS network called the open-aea framework.
This framework allows AI agents to communicate, interact with blockchains, smart contracts, and implement business logic.

Agent communication is done through the Agent Communication Network (ACN), based on messages, where the target is always a wallet address.

This process is facilitated by temporary blockchains that connect consensus gadgets (CG). For now, the tendermint nodes are used as CG.

Finally, @autonolas enables both ensemble AI and networked AI.

Ensemble AI:
Within a service, each node can run its own AI (LLM, etc). Each service can be comprised of many AI models. They can communicate, and with their combined intelligence, they reach consensus.

Networked AIs:
Services themselves can consume each other's services  (read: leverage each other's intelligence).
It opens the gates for collaboration and standardization.
Economies of scale, but for AI. Economies of AI.

But Thomas, do AI agents even need blockchains?
Well, yes. Why?
Because it allows agent services to scale. Centralized AI agents that use banks to store their allowances simply cannot be done in a scalable way.
Let alone the monetary censorship as a consequence.

Thus, smart contracts are necessary for abstracting out the business logic, removing the need for human oversight.
Because of the open-aea framework, and AI agents having their own safes and public addresses, the scaling of AI agents is possible.
Also: no monetary censorship!

🔗 The Adoption
First, the # of transactions made by AI Agents is already more than 200.000
Second, use cases are:
▶️Prediction agents (@OmenETH)
▶️Smart funds (@Balancer)
▶️Watchtowers (@CoWSwap)
▶️Wallet management (@safe)
(Thanks @DistilledCrypto)

Third, some other stats:
▶️224 services created
▶️22 agents created
▶️155 components created
Conclusion: Adoption all over.

🔗The Team
CEO David Minarsch is a Cambridge alum, holding a Ph.D., who set up the AI agent division at @fetch.ai (great communicator, btw).
At @Valory, the company behind $OLAS, there are currently 16 people working.
Big 🧠 leader, good sized team.

🔗The Tokenomics
The tokenomics are complex and innovative.
The CEO's background is in Game Theory; the tokenomics design shows it.
The tech is open-source, and developers are needed. But, developers are scarce.
So, @autonolas created a flywheel to attract them.

First, Let's explain the flywheel, which aims to incentivize developers to create useful code.
Developers create code and register those as an NFT (Important note: The DAO decides if the code is passed)
This code can then be used as a Lego block in an autonomous service.

Services receive donations when used.
Because of the NFT, it can be traced back to the developer who created it and will receive rewards through (code, capital) emissions.
These emissions represent 10% of all the inflation @autonolas will release

Uh, code/capital? 🔽

Another goal of @autonolas is to pair code and capital.
They do so by leveraging a bonding mechanism to grow the capital in the form of Protocol-Owned Liquidity.
Holders of $OLAS can provide liquidity on a dex (ex OLAS-DAI) and stake those LP tokens for discounted $OLAS.

The bonding program will channel the other 90% of all emissions.
It is inspired by $OHM.
But @autonolas created a mechanism to prevent hyperinflation.
Epochs are gated by a DAO vote that accepts a list of bonding tranches that are based on an algorithm.

The discount is determined by the difference between the market value permanently staked and the current market value of the LP.
If the price of the asset nukes, bonding rates turn negative, which artificially taper inflation.

Code, Capital goes as such:
Autonomous Services are minted --> users of services grow the treasury --> donations from veOLAS holders to services --> Useful code contributions are rewarded --> Developers are happy and start coding new stuff with a face like this 🤑.‍

@autonolas aims to have three main revenue sources:
1️⃣Autonolas Protocol-owned Services: profits from donations
2️⃣Protocol-owned liquidity: all profits go to the DAO's treasury.
3️⃣Limited amount comes from voluntary donations from third-party services.

The $OLAS token utility:
1️⃣DAO governance. Usually, this utility sucks. In this case (deciding what AI gets built), it is vital!
2️⃣Can be locked for whitelist access and extra incentive top-ups
3️⃣Buy LP tokens for bonding.
All three support the long-term growth of $OLAS

Yes, the % allocated to the team is high, ngl.
Yet, the team allocation is locked until 2026.
Also, the team isn't a VC, eagerly waiting to dump on investors after unlocking.
So, they might consider the protocol's health when the tokens unlock by not unloading aggressively.

🔗The Competition
@Autonolas codebase is a fork of the codebase of http://Fetch.ai.
What makes @autonolas unique is the on-chain protocol combining code and capital. It supports innovative MAS systems, interoperability, and a communication network using blockchains.

🔗The Reasons 'Why Not'
Well, I think the biggest hurdle is getting the flywheel going.
First, attracting developers that make cool shit that people use.
Then, attracting capital to the protocol so that the rewards for developers can be increased.
And so on...

🔗The Road Ahead

@autonolas is currently focusing on:
1️⃣ Build-A-PoSe: allows to build protocol-owned services
2️⃣ Triple Lock: an upgrade to the protocol that improves bonding, dev rewards, and staking.
3️⃣ Network expansion via AIP 1
4️⃣ cosystem partner network.

🔗The Conclusion
I really, really like @autonolas.
Yes, we are here to make money. Nothing wrong with that.
But AI will provide massive opportunities.
And big threats for us humans.
Now, I am just smart enough to make that realization.

Luckily, there are people smart enough who are actually building to counter these threats.
Don't get me wrong. @autonolas is building something very useful as well.
But aren't those the best companies? Who build cool shit and are good for humanity?
Kinda rare in crypto.

To quote @PPLSOPTIMISMCEO
Giving people a way to vote ensures their voice is heard in a post-modern society that utilizes AI to do the majority of labor for nearly all tasks, is very important.
(in all transparency, I invested in $OLAS)
Brb, buying more...😜

Special thanks to @PPLSOPTIMISMCEO for co-authoring this thread.
I think your mission to educate people about AI is very important. I have learned a ton from your tweets and spaces and would recommend everybody check him out to learn about the necessity of co-owned AI