Introduction
Gnosis built the original prediction market with Omen and was later handed over to DXdao to continue development. Omen was down-prioritized by DXdao and there has been no active development since 2022. Gnosis, along with Gnosis AI and OLAS has been using Omen’s code to deploy their own frontend (aiomen.eth.limo) to test the efficiency of the prediction markets driven by AI agents and also as a way for letting AI agents predict these events.
The past quarter, we have been building a new frontend as a way to make the prediction market better. The vision of this new dapp is to visualize how well these AI agents can create markets for real world events and how well these AI agents are able to predict these markets.
Motivation
The past couple of months, there has been a great amount of traction in prediction market area with Polymarket raising 70M in investment and there has been considerable user growth in polkamarkets, limitless.exchange and manifold.markets. At Dappcon, there has been plenty of talks and discussions around AI, the kind of risks it poses, the challenges it has and how AI and blockchain could become an unstoppable force that needs to be kept a close watch. Gnosis has also been ahead of its time in terms of all the products they launch and this vision aligns well with the Gnosis AI and the investment they have made on AI agents.
During Dappcon, we showcased the new UI we had started building for prediction markets and asked for feedback from Martin. We explored a few possibilities around how we could build a frontend for Omen, where we see the predictions that AI makes and the research that it does when making a prediction. We will also be able to see how each AI agent is performing, comparing them against each other.
Proposal for Omen frontend development
We would commit to developing the features listed in the vision and roadmap and would be actively communicating to the Gnosis community. We would also work closely with Gnosis AI and OLAS teams to be able to explore new improvements in the AI agents that create the markets and predict the outcomes.
GnosisDAO will be managing the liquidity for all markets that are being created. This is the status quo right now as all the markets created by the AI agents are being given a budget to add liquidity. In case, there are markets that are good to have based on current events, Gnosis DAO will be managing and maintaining the liquidity.
Gnosis AI will also look to create these markets against Swapr V3 to get deep liquidity on the markets and the outcome.
As part of the proposal, we would also be looking forward to receiving a grant of 400K in WXDAI and 600 GNO.
We would love to receive control of omen.eth to be able to deploy the frontend on the Omen’s ENS and using Omen’s brand. We would create a new multisig to control this that would be a 3 of 5 signatures, with the 3 signers being Venky, Leo and Diogo from Presagio’s team and 2 from Gnosis.
We also would work towards the Gnosis and Omen team agreeing on a roadmap to continue post MVP.
Roadmap
Vision
To build a dapp that gives insights into how AI agents predict the events, how often they are successful in predicting the events and what research they do to reach there. The vision is to provide a visualization of how well these AI agents are predicting the markets, what difference in prediction score each model presents and how that affects the agents predicting.
MVP version:
Base features:
-
The dapp should be able to show all the binary markets and the spread for each market. (Completed)
-
In the markets page, any user should be able to buy the outcome tokens with the wallet connected.(Completed)
-
Searching and filtering of markets. (Completed)
-
There should be a way to see the activity history and a graph depicting the fluctuations in the outcome of a market. (Completed)
-
Show markets that are based on different collateral tokens. (Completed)
-
Redemption of outcome tokens once the market is resolved. (Completed)
-
“My bets” page with a list of all the markets that the user has predicted. (Completed)
AI specific features:
-
Connect with Gnosis AI APIs to show relevant articles related to the market. (Completed)
-
A filter for all the markets created by the AI agents. (Completed)
-
Visualize whether a particular transaction is created by one of the AI agents. This is based on the list derived from https://dune.com/hdser/omen-ai-agents (Completed)
-
Display the resolution status and oracle, based on reality.eth oracle answering process (In Progress)
-
If the transaction is from an AI agent, link the research article, the AI model used and the confidence rating that had been used by the AI agent to arrive at the decision. This is done by integrating and fetching info from autonolas mechs.(Yet to start. This needs collaboration with Autonolas and their AI mechs)
-
Display AI agents performance across various markets. We will link the agents performance to take the user to AI agents’ overview (Yet to start)
MVP plus:
We also had a call with Martin and Stefan to create a market for the Presidential election to create a market based on SDAI as collateral. This will offer a unique differentiator for the predictors and the LPs to generate yield on top of their predictions. With SDAI generating almost 8 to 10% , this will be a huge differentiator to other prediction market products.
Future versions:
-
Expansion from binary market to multi category market. This will also need the support from OLAS and Gnosis AI team to be able to create categorical markets and also to be able to predict outcomes from those markets.
-
Permissionless market creation. The possibility for anyone to use the new Omen dapp to create a new market (binary and categorical) Anyone can come in, ask a question to the web app and get a bunch of research done by AI through PMA agent and then get a new market created based on the question.
-
AI agents overview. This should display how each AI agent is performing, what markets they are predicting on, the model they use and how accurate their predictions are. We can also show the Brier score and compare the agents’ performance against the overall agents performance.with a leaderboard that lets us compare why a certain AI agent is performing better compared to others, helping us get insights into the prompts and the model that is being used to research.
-
Limit orders. Explore the possibilities of placing limit orders to enter or exit at a certain outcome price.
-
Explore predictions on DAOs by adding predictions to the KPI goals.
-
Explore the possibility of discussion forums for each market. We can also have a special tag for “AI specialists” and “AI thought leaders” to provide their inputs on any market and why they predicted a specific outcome.
-
Predicting and betting on markets with any token by enabling integration with Cowswap
-
Exploring options to convert the ERC1155 conditional tokens to ERC20 bet tokens
Project update:
Presagio is already functional and is used by Gnosis AI team as an alternative to aiomen.eth.limo (a fork of old Omen) and can be accessed here:
The designs: https://www.figma.com/design/VcGkl9Wyc416IeyeRM8Z5U/Omen-2.0
The codebase: https://github.com/SwaprHQ/presagio
Funding:
As part of the proposal, we would also be looking forward to receive a grant of 400K in WXDAI and 600 GNO (will be used as a validator for 1 year)
The funding will be used for the following:
Development, testing and maintenance of the product - 70%
Marketing - 15%
Infrastructure costs - 3%
Project / Product management - 12%
- Transfer omen.eth ENS to a new multisig for frontend deployment and management.
- We would also like to work closely with the Gnosis AI team to understand and explore the possibility to make the agents get better at creating markets and predicting them.
The Presagio team comprises Leo, Diogo, Rorry, Nicki, Akash, and Venky. As a cohesive unit, we have collaborated for over a year to deliver notable projects such as Swapr’s CLAMM(https://v3.swapr.eth.limo ) on Gnosis chain and Stackly (https://stackly.app) deployed on Mainnet, Gnosis chain and Arbitrum, all built atop CoWswap infrastructure.
Looking for feedback from the community.
Thanks,
Venky