Aptos infra, tooling, AI, and public goods: all you need to know for Hack Holland
Table of Contents
On May 30, Pontem co-hosted the Move Tuesday Twitter Space in preparation for the Hack Holland hackathon. Together with our partners Aptos Labs, Concordia, Sentio, Switchboard, and Panther Protocol, we talked about Aptos infrastructure, tooling, AI, and public goods. Find the best insights in our detailed recap.
Get ready for Pontem-sponsored Infra sessions at Hack Holland
On June 5-7, Amsterdam will welcome Hack Holland – a major global hackathon organized by Aptos Foundation. It will bring together the best teams building on Aptos and everyone who is interested in learning Move, the blockchain programming language of the future.
Pontem Network is the proud sponsor of the Infra track at Hack Holland – the part of the event that deals with infrastructure, tools for Move developers, public goods on blockchain, and AI. Participants will have a chance to showcase their ideas, network, code new projects, and win prizes. As part of the event, we’ll also host an online workshop on starting out with Move, headed by our CTO Boris Povod.
We believe this is the most important track of all, as the growth of the Aptos ecosystem ultimately depends on infrastructure and tooling. Luckily, Aptos is supremely developer-friendly and already offers teams coming from other ecosystems high-quality tools for building dApps with Move.
Those who are used to Solidity find Move to be a revelation – clear, secure, and easy to master. And of course, the community of Move builders is always there to help and guide the newcomers.
As a sort of a prologue to Hack Holland, we’ve just co-hosted a massive Twitter Space. Listen to the recording here or read the recap that we’ve prepared with our partners. We’ll describe what our infra tools do, how oracles work, and what we can expect when AI and the blockchain industry collide.
Pontem infrastructure for Aptos and Move: from a wallet to an EVM implementation
Infrastructure (or just infra) is an umbrella term for all sorts of solutions that allow blockchains to function: nodes, virtual machines, dApp development platforms, bridges, wallets, and so on.
Infra partially intersects with tooling: software that developers use to build dApps and new blockchains. Tools include wallets, contract debugging tools, code editors, libraries, and so forth.
As an end user, you’re probably familiar with our Pontem Wallet, Liquidswap, and NFT collections. But we are also one of the leading providers of Aptos infra solutions – let us list them for you.
This is the first-ever browser-based code editor for Move. This is probably the easiest way to start building on Aptos and other MoveVM-based blockchains. No need to download or set up any software: simply go to https://playground.pontem.network/ to access your workspace. The tool is completely free.
With Move Playground, you can write and deploy any Move smart contracts. Take your first steps as a beginner and create a new token – or build a full-scale dApp. We’ve included some documentation, and you can ask additional questions in the Pontem Tech Support chat on Telegram.
IntelliJ is one of the most famous families of integrated development environments (IDEs). These are platforms that maximize the productivity of software developers working in languages like Python, Java, C, and C++. The developer of IntelliJ is JetBrains.
With Pontem’s IDE plugin, coders can build Aptos dApps using their familiar IDE environment. Just like Move Code Playground, the plugin is 100% free. It’s compatible with all major IntelliJ IDEs (IDEA, PyCharm, CLion, etc.)
The plugin includes features like Move code auto-formatting and statement handling, symbol support (with quick search, renaming, and autocomplete), and on-the-go error checks. If all this means nothing to you, simply believe us: all these things make developers very happy.
What if you could take a dApp built for Ethereum, pass its code through a magic box, and get working code for a dApp on Aptos? That’s what ByteBabel does: it’s the first tool that compiles Solidity code into Move binary code.
ByteBabel allows developers to port projects written for EVM chains into the Aptos ecosystem without having to re-write the code. It’s a revolutionary way to try EVM on Move, with all of Aptos’ processing capacity (100,000 TPS) and low fees (around $0.01 per transaction). The plugin supports all Ethereum tools: MetaMask, Truffle, Remix etc.
Pontem is also working on an industry-first implementation of MoveVM for EVM chains. Stay tuned for details.
AI and blockchain: 4 interesting use cases
Neural networks are the hottest trend of 2023, with many professionals actually worrying that ChatGPT, Midjourney, etc. will replace them in the workplace. The crypto industry was quick to jump on board this trend, with AI-related tokens like AGIX, OCEAN and FET rallying over 100%.
Is there really a use case for AI on blockchain, though? And if so, how can AI help the development of the Aptos ecosystem?
First of all, there are areas where AI and blockchain don’t mix well:
- ML model training: it can’t be done on-chain, because you need huge computational and memory resources.
- dApps with a high security margin: imagine an AI-driven blockchain payment bot that is closed-source (a black box). Perhaps you’d trust it with $100, but not with $1 million. We can’t entrust important financial or security tasks to neural networks whose behavior we can’t control. In any case, we are still far from being able to run financial or payment bots on-chain.
As for areas where you can merge AI and blockchain, the participants pointed out the following.
Certifying content origin
The experts speaking at the Move Tuesday also agreed on another great use case for blockchain in AI: certifying content provenance. It’s becoming hard to distinguish content and images produced by AI from those created by humans – but we need a way to do this. Was a specific image drawn by a popular digital artist or by a neural network imitating her style? Is an article really from the New York Times or a fake?
Blockchain is great for attesting to facts and tracing the origin of things. So perhaps we can come up with a blockchain-based watermark system to trace a digital item back to its creator (signer). It’s not the same as the truth, but at least we’ll be able to prove who made what.
Simple coding and debugging
AI can already write pieces of code – though the result always has to be reviewed by human programmers. Once we have natural language models able to work with the Move language, we could use them to generate ideas for simple frameworks, testing, and debugging.
Distributed Intelligence (DI)
In this approach, AI processing tasks are distributed among many users (nodes). It’s similar to how blockchain works, and there are already many “decentralized AI” projects in crypto. Dr. Anish Mohammed, co-founder and chief scientist at Panther Protocol, thinks that it could be a bridge between the current centralized AI and on-chain AI. However, the same security margin problem remains, as DI systems are hard to control and update.
Assisting ZK systems
No AI model can really verify a zero-knowledge proof, and in any case you can’t trust AI to do such work independently as the security margin is high. But it’s an interesting area of research.
This is all very exciting, but there three serious problems around AI that need to be solved:
2) explainability and source of truth;
3) information storage and security.
Pontem is actually planning an online expert panel about AI, where we’ll discuss all these issues, so stay tuned.
Tooling, oracles, and data analysis
In this section of the livestream, the experts from Pontem Network, Switchboard, Concordia, Sentio, Panther Protocol, and Aptos itself discussed such complex but essential problems as tracing transactions, verifying external data, and collaboration on tooling development.
As a new network, Aptos doesn’t yet have a rich set of tools that are available for EVM chains, but luckily Aptos Foundation has an active grants program for proposals in this area.
Block explorers and real-time on-chain data
Here we’d like to give special kudos to our partners from Sentio, who are doing amazing work with their Aptos block explorer and on-chain data dashboards. We use them to track activity on Liquidswap and to monitor adoption across the Aptos ecosystem.
It is important to be able to see inside each transaction – not just the entry function that was called on a smart contract, but also various contracts’ calls on each other and opcodes executed by a virtual machine.
This can help identify how an exploit was conducted, or why a transaction didn’t work as intended. It’s also important for everyone in the ecosystem to have access to such data for reasons of transparency and independent review.
There are already such tracing tools for EVM chains, and hopefully they will emerge on Aptos soon, too.
Collaborative infra building
Concordia’s Thomas Ruble stressed that a healthy blockchain ecosystem should consist of reusable, modular, specific programs: liquidation bots, risk management algorithms for loan collaterals, staking protocols, DAO management tools, archival nodes, public data warehouses, etc.
Ideally you want teams to work on what they are best at and then freely share the results with each other. This is similar to the Linux culture, where every Linux user has hundreds of small free programs inside their OS, all built by different devs.
The issue is that building such tools requires money and resources. Some, like liquidation bots, are monetizable, but others are not – like governance protocols. Once again, Aptos Foundation has a program of grants, and ideally the Aptos team would like to see competing proposals for the same products. The ecosystem as a whole is already very dev-friendly, so hopefully we’ll see a collaborative primitive-building system emerge.
Transaction scheduling and multiple fee-payers
Another set of primitives that Aptos needs, as pointed out by Switchboard founder (@DoctorBlocks), are protocols for transaction scheduling and fee-paying models. Users should be able to decide how to bring information on-chain, how to reveal it, and how to control gas schedules.
Multiple fee-payer models are particularly interesting and valuable. The idea is that different entities can pay for different parts of a transaction. It’s different from multisig, where several addresses approve the same transaction. In the multi-payer scheme, there are no multiple signatures: you can still be the sole protagonist of your transaction, but someone else can sponsor the fee. For example, a game can pay the fees for the players to attract users, etc.
Aptos already has a clever multiagent model, which can in theory accommodate multiple gas payer schemes. Still, there are exciting challenges:
- access control (which entities are allowed to pay for parts of a transaction);
- protection from replay attacks;
- authority verticals
- timely signature aggregation;
- making sure the right entity gets the reward.
Switchboard is already building this functionality for its oracle product, and all interested teams are welcome to contact them for more insights.
Many dApps (such as games) use randomness, which needs to be verifiable. For this, third-party randomness beacons are often used. The challenge arises from situations where parties can collude with each other. Thus, having more sources doesn’t guarantee a random outcome.
Aptos is working on a built-in randomness solution to solve this problem: it’s yet another thing to look forward to on the vast Aptos roadmap.
Oracles: the 3 models
Switchboard is a premier oracle project on Aptos with its build-your-own oracle product. Its CEO Mitchell Gildenberg described the three main models that blockchain oracles follow:
1) Push button: oracles put the data (usually asset prices) they want to share on-chain. The feed creator pays for on-chain updates. You could describe such services as “eager” to share their data.
2) Pull-based: the oracle streams data to the selected L1 blockchain, which should have a mechanism to verify each price update. Users can pull that data as they need, but there is a risk of bottlenecks on the front-end that supplies prices.
3) Commit-reveal: Mitchell Gildenberg’s favorite model. A protocol points to a certain data feed, and the oracle pushes a price update after a short delay. You could call such oracles “lazy”, as they only share info when asked.
The advantage is that this eradicates MEV windows. (See our article on memecoins for more info on the costs of MEV for an ecosystem.) No matter how transactions are ordered in the mempool, no MEV actor will be able to use price updates to their advantage.
As usual, there are some considerations to take note of:
- Syncing across multiple blockchains: making sure that all involved chains learn about the updates supplied by an oracle. For example, if ETH “lives” on 20 different chains, ideally you want an “Eye of Sauron” to be scanning everything all the time so that the price stays more or less the same across the blockchains.
- Trust: who can validate that the price points provided by a third party are true? And how can we trust oracles to report the data correctly? Switchboard is already working on a solution to this, called Trusted Execution Environment (TEE) Oracles.
It’s time to start building with Move
The speakers covered many other interesting points during the discussion, so we do recommend that you listen to it in full here. If you are interested in building on Aptos, register for Hack Holland in Amsterdam on June 5-7.
Remember that Pontem will host an online workshop on coding with Move, led by our CTO Boris Povod. We always welcome teams that build on top of Pontem Wallet and Liquidswap, so don’t hesitate to contact us if you need help and guidance.