Name: Sahara AI
Uploaded: 2025-02-17T15:29:47.429Z
Duration: 2 min 33 s
Description: A decentralized AI blockchain platform ensuring security, equity, and accessibility for AI assets.

sahara-ai

A decentralized AI blockchain platform ensuring security, equity, and accessibility for AI assets.

Sahara AI

Hello and welcome everyone to the workshop. Uh my name is Aaron Scales and I'm head of product at Sahara. Um I have been building software products since 2008. I've experienced at companies like Goldman Sachs, LinkedIn, and Chainlink Labs. Um throughout my career I've been focused on mostly on platforms and integrations across both web 2 and web 3. I'm really excited to be here today sharing more about Sahara AI.

I want to begin by uh being very upfront with you. Sahara AI has recently come out of stealth mode and we're very excited to share our vision with you. Our products are just beginning to launch for early end user testing. That means we don't have a platform ready yet for developers. But this workshop and this introduction here is a great opportunity for you to learn more about our vision and what we aim to achieve. Let's get into it.

So first of all, what is Sahara AI? Sahara AI is a purpose-built blockchain platform for AI development that enables the creation, storage, deployment, and trading of AI assets in a secure decentralized environment. What we have in mind for the future is a future where AI is equitable, transparent, ethical, and accessible to all through AI asset sovereignty and open technology. What do that mean a little bit more tangibly? Well, our mission is to build a decentralized AI platform that prioritizes AI asset sovereignty, enabling, uh excuse me, ensuring security, equity, and accessibility for all users.

We are backed by uh many different firms and I'm just showing the slide here to show you that we have the means and the resources we need to achieve this vision. Before we get into some more details about what Sahara AI aims to do, I think it's important to look at the macro landscape. We're entering a stage of AI development, uh where you start to look at the solutions around us today which are primarily centralized and you note certain things about them. Um and we in stark contrast to that have a different vision of the future.

So going point by point, beginning with data privacy. In centralized AI, you have challenges with sharing your data directly with the central entity and this.

Means that your data could potentially be misused and access by unauthorized parties in the future. For decentralized AI, we imagine that data would remain on Prem or in your own private nodes where you have full control over it. Building upon that, if we look at the dimension of security, centralized storage is a prime target for attackers. Um, but by using private nodes and decentralized management, you can likely reduce many of those risks. 

I think that monetization and cost efficiency today might be better in centralized AI, but I think in the future decentralized AI is going to be offering more cost effective transparent pricing. Finally, on this slide, I want to talk about data ownership. So today, when you use centralized AI, oftentimes these terms can grant them access or rights to that data for various purposes. In contrast, in our vision of decentralized AI, you would retain full ownership and control of your data. 

I have a few more dimensions I want to cover here. Um, when it comes to scalability, there are all sorts of risks involved with centralized infrastructure, whereas we've seen that decentralized networks really allow for more dynamic resource allocation and scaling without facing those bottlenecks. Additionally, in a centralized environment, you're limited to the customizability of your solutions. You're going to be using the models and tools that they provide you. Um, in contrast, decentralized solutions open up a wide range of tools, data sets, and models that might be available for your development. 

Another key concern having to do with AI development is interoperability. How easy it is to, you know, interact with those assets across different platforms. In centralized AI, again, you're limited by what the platform offers, but in decentralized AI, you have more interoperability with various systems so that you can integrate and collaborate across lines. 

In terms of how you can access additional data to power your AI, centralized AI today is quite limited. You're only going to be using the data sets or maybe some additional training data you.

Provide to a centralized entity. Uh, in contrast, we believe that having an open marketplace is going to lead to better outcomes. And by using decentralized lauras, um, you can have models that are more effective. 

So to summarize, centralized AI platform compromise a lot. They face security risks and your ability to monetize your work is severely limited. There's a lot of benefits of decentralization. First of all, you should always have control over your AI assets. Second of all, contributors should be fairly rewarded for their participation in the AI economy. And lastly, blockchain technology ensures that there's traceability of how data is used and how AI models are interacted with. 

All this together leads to a future that is a collaborative economy. One with broad participation, equitable wealth distribution, and one where all stakeholders are benefiting from and are influencing the platform's evolution. 

So I W to I want to try to make this a little more real for everyone here. Uh, here is a user journey that we imagine becoming true with Sahara AI in the future. Let me set the stage a little bit. 

So to begin with, we have a company Tech Solutions. They are a small medium business in the business of customer service automation, and they want to develop a custom AI agent using their own proprietary data sets. 

And at this company, there's a CTO, Jane, and she's got some challenges. Excuse me, she is interested in maintaining the privacy of her data. She's interested in, uh, meeting all the regulatory guidelines, and she's also interested in using additional data sets to train and enhance their AI. 

Lastly, there's a lot of pressure for Jane to monetize solutions and help her business be more profitable. So what can Jane do with Sahara? 

First of all, Jane would sign up for Sahara AI, setting up a profile, completing the onboarding, and getting access to docs, tutorials, anything she needs to better understand the solution and how to go about building. 

Then Jane can use Sahara AI's SDK to set up encrypted private nodes. And this is where they're going to store all the proprietary data within their own environment and enable secure data.

Transfer to AI infrastructure offchain. Then Jane is going to be able to create an agent from scratch using Sahara's no code tools. This is going to enable her to also keep her data secure during training and inference using decentralized lauras. But Jane doesn't want to just use her own company's data. She wants to build the best AI agent, and to do that she thinks she should train her model on additional data sets from the marketplace. So she goes there, finds relevant data sets to customer service automation, and is able to pull them in for training to enhance the AI agent she's building while keeping her own proprietary data secure.

At this point, Jane and team are ready to deploy the AG for the actual use case, which is customer service, and integrate it with their existing systems. At this point, they're ensuring that data security is maintained throughout the entire process, and they can monitor their performance improvements over time. So that completes the usage of this AI. But beyond that, Jane is finding that the company's data set is quite valuable, and she thinks it could be licensed and used by others to also improve their models. Maybe even she can build her own model and launch in the marketplace as well.

So using Sahara's technology, Jane is able to license her data for training purposes without ever revealing the exact data itself. Throughout this, she's also able to maintain onchain attribution and guarantee compensation for the data that she licenses. So with that use case in mind, what are the building blocks that let us get to that vision? Well, really right now, Sahara is very much focused on the AI marketplace, and this is our vision for a hub where users and AI developers can come together to create high-quality AI assets that they need to develop their integrations.

So beginning with training data, your AI is only as good as its training data. We think it's really important to focus on a process and a marketplace that ensures quality above all else. So we're using techniques like onchain majority voting. We're using honeypots and reputation.

Systems to penalize malicious activity and make sure that only um only users that are that are doing this correctly and are actually providing uh accurate responses are used in the training data. This Marketplace is also where we're going to facilitate the secure trading of data models and all AI assets on chain. This is also where we're going to manage the licensing and access control of those assets. 

So so next I want to talk a little bit about more from the developer point of view. Once you have your model and you've trained it um you know what are the other pieces of of tech and product that you need to make this a reality. Well we're imagining that uh you're going to need a software development kit that will help you interact with the Sahara API. 

Um the Sahara API is really what abstracts away all the specific AI capabilities of uh everything from training um deploying serving uh and and Reporting on the success of different AI models. And so the SD is going to help developers do all that without needing to write a lot of code from scratch. The end goal here is that you're going to be able to deploy Sahara agents which are going to actually provide those inferences and then that inference generation can be written back on chain. 

So let's let's look at the the chain side of this in a little bit more detail. Um I can't today offer too many specifics because some of this is still under wraps. You're going to have to stay tuned for more soon but uh what we are using our blockchain for is uh as a platform where we're going to manage all the AI assets all the transactions all the trades all the uh training and um task completions. 

This is where we're going to reward everyone uh for their contributions and and some of the key things we're putting on chain associated with that would be things like Sahar Sahara's identity management um all the voting having to do with quality but also governance of the protocol itself. The actual task creation and management for everything we're doing in our Marketplace um that reward distribution I've spoken about to make sure that it happens in a trustless way.

Um pre-compiled refers to some tweaks that we plan to make to our blockchain to make it even more efficient for doing AI operations. Um and then there's also some fundamentals on the blockchain such as the ability to participate as a validator with staking and also support for Ethereum. Uh and uh Ethereum compatibility and Solidity smart contracts. 

So related to the chain is a concept of the Sahara data layer and this is really important because uh traditionally a lot of these assets associated with AI development happen to be really large. They're too big to put on chain directly but we still need to be able to reference them in a way that is verifiable. So that's why we're planning the Sahara data layer. 

Um onchain we'll have those immutable records of user operations, all the attributes and metadata that we need. But offchain we're going to use decentralized storage for large data sets and models and we'll use integrity mechanisms uh that that ensure that we know uh what is offchain accurately. 

Um we also plan for data integrity mechanisms that are going to prevent uh duplication which is definitely a huge concern of ours uh as it comes to AI asset Providence, sovereignty, and trading. So together um all those solutions and that architecture that I just spoke about enable a really cool future. 

We already talked a bit about what's possible for customer service automation but I also want to highlight some other use cases we're imagining that in the future. Um NPCs and gaming don't have to be boring and just say something from a limited set of responses. You could imagine a game developer using a Sahara AI agent to create a much more dynamic experience where the NPC is able to um come up with new responses on the fly based on situations and context and all the training that's gone into that model. 

So we think this is going to be a much more rewarding experience for players. Um in a completely different use case we can look at DeFi use cases like portfolio management. In the future I want you to imagine an AI agent uh built on Sahara helping users manage and optimize their portfolios by doing.

Real-time analysis and all that training data um was was based on the Sahara Marketplace and our collection of defi strategies and other information that would make these agents successful. Uh lastly I want to drive home the concept of data privacy. There are many high stakes domains that require us to have full control over our private data and that's why we're also working on privacy features that would enable things like Health Data to be shared um or I'm sorry Health Data not to be shared but to be accessible during AI inference Generation. 

Um related to this uh we are very interested not just in the Privacy aspect of AI but also in the verifiability so that the model that is making differ recommendations about high stakes decisions in your life is verifiable on chain with full traceability of which version of that model uh existed and how it was trained. Now I want to give you a high-level picture of what's going on on our road map um in very broad Strokes. Right now we are focused on the limited release of our AI Marketplace. This like I said before is really the foundation of AI development being able to bring together great data with the ability then to train models and then putting those models on the marketplace for others to use as well. That's going to be the basis of a thriving AI uh ecosystem and development platform. 

What we're trying to do on the Chain side is develop our test net uh first in a centralized and then a more decentralized fashion and prove out the robustness we need to operate all these AI use cases that I spoke about previously. We are targeting a mainnet launch early next year and shortly following that we expect the AI capability the AI capabilities of our platform uh to move along quite quickly as we have our sdks our apis and our um actual agents able to be deployed onchain and accessible uh to web 2 users as well through apis. I want to highlight some of these other opportunities for web 3 and AI Developers. First of all if you're a data contributor and uh this could be anyone who has sub uh subject matter expertise in a particular area you can monetize your data.

Using Sahara AI developers are going to be able to generate revenue from licensing and usage fees and token holders can earn rewards through staking. Uh this is an opportunity for AI developers to actually create data projects on chain and then reward the users that help contribute that data. Um additionally there's going to be opportunities for folks to contribute on the technology side. 

Um we are expecting certain aspects of our platform to be open source and we'd love the community to get involved in helping us build this. Lastly, um I just want to stress that our commitment here is really to uh to the community and to humanity. We are looking uh to build very ethical AI solutions and this is going to require constant collaboration not just with the user but um in an economic sense as well trying to promote knowledge sharing and continuous improvement of our platform. 

I really appreciate your time today. Thank you.

Note, that there may be slight differences in text and video. Watch video for the original content.

Sahara AI Video