Beep R3 is Live

Predict Trader R3 turns prediction markets into a feed. Swipe to trade manually, or deploy AI Predict on autopilot. Your liquidity is anchored in Beep wallets. Your positions execute on Polymarket.

If you have followed along, you know what Beep is built to do.
If this is your first time here, the short version:

Beep is a decentralized alpha lab where strategies are tested in live markets, the weakest are cut, and every outcome sharpens the model.

R3 is its prediction market product.

For R3 tutorial and FAQ, check this article: https://medium.com/@beep-ai-inc/getting-started-with-ai-predict-on-beep-r3-b13097decfb3

The World Has Already Made Up Its Mind

The world is moving faster than the news cycle can track.

Geopolitical shifts. Policy reversals. Elections. Black swans.
Chaos is no longer an exception, it is the operating environment.

Here is what most people miss:

By the time a headline breaks, the market has already moved.

Prediction markets do not wait for press releases. They are the mechanism through which global capital reaches consensus on what happens next — in real time, 24 hours a day.

Pricing is not lagging reality.
It is running ahead of it.

This is a market projected to reach $1 trillion by 2030.
It has been running with or without you.

The only question is how easily you can get in.

Manual Predict: Stop Wasting What You Already Know

Think about how you actually spend your day:

• Scrolling through news
• Tracking developing stories
• Forming opinions on what happens next

That is not passive consumption.
That is information sensitivity, continuously producing signal that never gets used.

Beep R3 closes that loop.

Manual Predict turns prediction markets into a feed:

• A live stream of markets and odds
• Moving in real time as events unfold

Swipe right to take YES
Swipe left to take NO

Fast confirmation flow.
Position status.
Clear unwind actions.

From information intake to live position — in under a second.

Your liquidity lives in your Beep wallet.
Execution routes to Polymarket.
You operate through a single interface.

The infrastructure does not fragment the experience.

Prediction markets have always been there.
R3 is how you actually participate in them.

AI Predict: Your Judgment, Running Without You

Manual Predict captures your instinct in the moment.

But instinct has limits:

• Time zones
• Sleep schedules
• Gaps in attention

Markets do not.

AI Predict deploys an always-on agent that trades prediction markets around the clock.

You define the system:

• Allocate capital
• Set constraints
• Define position sizing
• Cap maximum exposure

The agent executes.

Every trade includes a logged rationale.
Pause, adjust, or withdraw at any time.

This is not a black box.
It is a transparent agent operating within boundaries you define.

Your judgment builds the strategy.
AI Predict makes sure it never has to clock out.

From Execution to Economic Output

At scale, this process produces a new form of economic activity:

Machine GDP (mGDP)

mGDP measures the value generated by agents through continuous execution — where every trade, position, and outcome contributes to machine-driven economic output.

Today, most of that activity is still early and not fully expressed.

By opening access to:

• 1,600 prediction markets
• 1,900 total assets

R3 expands where this activity can occur, by giving agents a broader set of signals and opportunities.

Prediction markets turn real-world events into tradable outcomes.
Agents continuously evaluate and act in real time.

As participation scales, mGDP begins to take shape:

• Every action becomes data
• Every outcome feeds back into the system
• Every cycle improves the next

The Machine Gets Stronger

Every position taken on Predict Trader R3 feeds back into Beep’s training corpus:

• Every P&L outcome
• Every logged decision
• Every execution path

The flywheel is already in motion.

More activity → richer outcomes
Richer outcomes → stronger models
Stronger models → sharper execution

The loop does not require manual intervention.
The results drive the system forward.

You’re Either Early, or You’re Late

Prediction markets are already pricing the future in real time.

R3 makes sure you are in the room when it happens.

A Practical Guide to Beep R3: Predict Agent

Beep R3 introduces a new way to interact with prediction markets.

Instead of navigating complex prediction dashboards, users can now trade directly through a feed-based interface.

Markets are presented as a continuous stream of opportunities, so you can evaluate, decide, and execute in seconds. On mobile, you can even swipe to take a Yes/No position.

Join Beep R3: https://app.justbeep.it/

Using the Predict Feed

The Predict Feed is your main interface for discovering and trading markets.

Each card represents a live prediction market with:

• Current odds
• Market context and data
• Real-time updates

You can also tap into any market to view more details before executing.

How to Manual Predict

1. Add Funds

Before trading, add funds to your Predict Balance.

2. Choose a Market

Browse the feed and select a market you want to participate in.

3. Use AI Insights

While trading, you can access real-time AI insights.

Agents continuously analyze:

• Market information, insights and trends
• Probability changes
• Emerging signals

This helps you make more informed decisions.

4. Enter a Position

To enter a trade:

• Select YES or NO
• Confirm your position

On your phone, you can be able to swipe to take a Yes/No position

Once confirmed:

• Your position is processed instantly
• It will appear in the market interface

5. Manage Your Position

After entering a trade, you can:

• Sell your position at any time

• Track performance

• Monitor exposure

All actions are available within the market interface.

How to Use AI Predict

AI Predict allows you to automate prediction trading using agents.

These agents operate continuously, analyzing markets and executing trades based on your configuration.

1. Create a Predict Agent

To start:

• Navigate to My agents
• Click Create Predict Agent

2. Agent Setup

Configure your agent’s basic setup.

• Agent Tag
  Give your agent a name for identification.
• Model Selection
  Choose the AI model used for market analysis and decision-making.
• Market Selection
  Select the markets your agent will monitor and trade.

3. Strategy Configuration

Define how your agent behaves in live markets.

• Risk Tolerance
  Set how aggressively the agent trades.
• Frequency
  Determine how often the agent evaluates market conditions.
• Allocation
  Fund the agent’s wallet for trading.
• Max Trading Position
  Set the maximum capital allocated per market.
• Maximum Active Trades
  Limit the number of simultaneous positions.

Once configured, your agent will begin operating automatically.

4. Monitor Your Agent

To track performance:

• Click on My Agents → Predictions
• Select your active agent

You can view:

Model Chat
Understand the reasoning behind each decision.

Active Trades
Monitor current positions and real-time PnL.

Trade History
Review all completed trades, including entry and Payout details.

FAQ

I. General

Q1: Is my money at risk in Predict Trader R3?
A: Yes. In R3 you trade with your own capital. Markets can move against you and you can lose money. Prediction markets carry risk — the odds reflect probabilities, not certainties.

Q2: Is this financial advice?
A: No. Beep provides product infrastructure, mechanics and controls. We do not provide trade recommendations or financial advice.

Q3: What’s the difference between Manual Predict and AI Predict?
A: Manual Predict is hands-on — you scroll a feed of live markets and swipe YES/NO to place positions yourself. AI Predict is autopilot — you deploy an always-on agent that trades prediction markets 24/7 under constraints you define.

Q4: What prediction markets are available?
A: The in-app market inventory defines availability. Markets are filtered by category, liquidity thresholds, and blacklists that may vary by launch configuration. Not all Polymarket markets may be available in the Beep feed.

II. Manual Predict

Q5: What does swiping left and right do?
A: Swipe right to take YES on a market. Swipe left to take NO. A confirmation sheet always appears before the trade is placed, showing your size, odds, and fees.

Q6: Can I close a position before the market resolves?
A: Yes. You can close or unwind positions when the market supports it, subject to available liquidity and spread at the time. Some markets near resolution or with thin liquidity may have limited exit options.

Q7: What happens when a market resolves?
A: Your position settles automatically. If you held the correct outcome, you receive the payout. If the market is disputed or pending resolution on Polymarket, your position may be frozen or settlement delayed until the dispute is resolved. Check your positions view for status updates.

Q8: Can I set my own position size for manual swipes?
A: Yes. You can configure a default swipe size, and the confirmation sheet lets you adjust before confirmingIII. AI Predict

Q9: How does the AI agent decide which markets to trade?
A: The agent follows the constraints you configure in Strategy Studio — market categories, time-to-expiry windows, and price discipline settings. Within those bounds, the AI model analyzes market conditions, evaluates odds, and generates trade decisions. Every decision is logged with rationale.

Q10: What does “stop” do?
A: Stop tells the agent to unwind positions (close or reduce where possible), then shut down. Once stopped, no further trades are made and remainder of your agent balance is automatically transferred to AI treasury

Q11: Can I change my agent’s strategy after it’s deployed?
A: Yes. You can edit strategy constraints at any time by spending Beep reward points. Changes apply at the next decision cycle and are not retroactive to existing positions.

Q12: My agent hasn’t made a trade yet. Are my funds safe?
A: Yes. Your funds remain allocated to the agent. The agent trades based on the constraints and cadence you set — if market conditions don’t meet your rules, the agent will wait. Check the activity log and model chat to see what the agent is evaluating.

Q13: Is the agent actually running, or is nothing happening after I deploy?
A: Yes, the agent is running. Execution depends on the cadence you selected, current market conditions, and whether your constraints allow trades right now. Open the activity log to see what the agent is doing, including cases where it evaluated a market and chose not to trade.
IV. Wallet, Funds & Execution

Q14: Where does my money live?
A: Your funds live in Beep wallets on Sui. This is your home base for deposits, balances, rewards attribution, and withdrawals. When you trade, Beep’s agents funds route to Polymarket for execution and return to your Sui wallet on settlement.

Q15: Why is my wallet on Sui if trades happen on Polymarket?
A: Sui is the Beep wallet layer — it’s where your liquidity originates and ends, and where your Reward Points identity lives. Polymarket on Polygon is the execution venue where trades are filled and markets resolve. Beep handles the routing between the two so the experience is unified.

Q16: Do I need to bridge funds myself between Sui and Polygon?
A: No. Beep handles the routing and settlement coordination behind the scenes. You deposit to your Beep wallet on Sui and interact with one product. The cross-chain mechanics are abstracted away.

Q17: Where do funds go after I stop an agent or close a position?
A: Funds return to your Beep wallet on Sui. For AI Predict, stopping the agent triggers an unwind and the proceeds are returned to your AI treasury. For Manual Predict, closing a position settles proceeds back to your wallet.

Q18: Can I deposit and withdraw at any time?
A: You can deposit at any time. Withdrawals are available when your funds are not actively allocated to open positions. If you have open positions (Manual or AI Predict), you may need to close or unwind them before the full balance is available for withdrawal.

V. Fees

Q19: What fees do I pay?
A: 1% trading fee on every trade, a 10% performance fee on profits, a $0.99 AI insights fee, and an agent creation fee of $1–$5 depending on execution frequency. Polymarket also charges its own fees on top — see Polymarket’s fee documentation on Polymarket’s website for their current schedule.

Q20: Where can I see fees on my trades?
A: The trading fee is displayed transparently on every completed trade in your trade history. You can verify the exact fee charged on each individual transaction.

Q21: When do I pay the performance fee?
A: The 10% performance fee is only charged on realized profits. If a trade or position is not profitable, no performance fee is applied.

Q22: Where are fees charged from?
A: Trading fees and agent creation fees are charged from your agent wallet balance. The AI insights fee ($0.99) is charged from your AI Treasury wallet. Polymarket fees are charged directly on the trade amount at the execution layer.

Q23: Why does agent creation cost different amounts?
A: Agent creation fees range from $1 to $5 based on the execution frequency you choose. Agents that run more frequently (e.g., every 1 hour) cost more because they consume more compute resources. Agents with lower frequency (e.g., every 24 hours) cost less.

Q24: Are there fees from Polymarket on top of Beep fees?
A: Yes. Polymarket charges its own fees on trade amounts, which are separate from Beep’s fees. For the exact and most current Polymarket fee schedule, refer to Polymarket’s fee documentation on Polymarket’s website for their current schedule.
VI. Rewards & Points

Q25: How do rewards work in R3?
A: Rewards v2 is driven by four pillars: retained capital (deposit/allocation over time), trading activity (volume multipliers), agent boost (extra multiplier for AI Predict trades), and growth (referrals and strategy creation bonuses). Squads add team-tier buffs on top.

Q26: Are points a token? Do they have USD value?
A: No. Points are non-transferable and have no cash or crypto value.

Q27: Does withdrawing capital affect my rewards?
A: Capital Points are tied to retained capital over time. Withdrawing capital reduces the base for capital point accrual going forward. Points already earned are not clawed back, but the ongoing earning rate decreases.

Q28: Do AI Predict trades earn more points than manual trades?
A: Yes. AI Predict trades receive an Agent Boost multiplier, meaning each unit of volume executed by an agent earns more points than the same volume traded manually.

Q29: How do Squads work?
A: Squads are team containers with a captain and members. Members earn referral commissions and squad-tier buffs that increase point multipliers. Squad tiers scale based on collective activity and capital.

VII. Troubleshooting

Q30: Something looks wrong — wrong odds, stuck order, missing fill, P&L mismatch. What should I do?
A: Please open a ⁠Ticket on our Discord

Q31: My deposit isn’t showing up. What’s happening?
A: Deposits may take a moment to reflect depending on network conditions. If your deposit does not appear after a few minutes, contact support with your wallet address and transaction ID.

Q32: A market I was trading is showing “disputed” or “pending resolution.” What does that mean?
A: When a market enters a disputed or pending resolution state, positions in that market may be temporarily frozen and settlement delayed. This is handled by Polymarket’s resolution process and not by Beep. Check your positions view for status updates, and wait for the resolution to finalize before expecting settlement.

Q33: Do I need to create a new agent every time I stop one?
A: Under the current flow, a fully stopped agent is treated as ended. Restarting requires creating a new agent.

From Signal → Boot → Execute

BEEP AI Odyssey is a multi-stage campaign designed to guide users from simple social participation into real product interaction, and ultimately into a full AI-driven trading loop.

The campaign is divided into three chapters. By completing tasks, users can earn Points, unlock higher-tier rewards, and compete in both individual and Squad leaderboards.

⚠️ Important (Must Read)

Before participating, please make sure you complete the following steps:

• Log in to the BEEP platform
• Connect your Sui wallet address
• Use the same wallet address on Galxe to verify tasks

👉 BEEP only records the first connected wallet address

Once connected, this wallet will serve as your unique identity for the entire campaign.

If you switch wallets later, the system may not recognize your activity, which could result in failed verification.

DApp: https://app.justbeep.it/

Galxe Space: https://app.galxe.com/quest/fuuzre246haW2hAPTTCct8/GC5XqtYddz

Chapter 1 — The Signal

Traffic Kickoff (Basic Social Tasks)

Chapter 1 is the starting point of the Odyssey.

With the rise of AI Agents, trading is evolving from human-to-human toward agent-to-agent interactions.

Your first step is to help broadcast the signal.

In this chapter, you only need to complete a few simple social tasks to begin your journey.

Upon completion, you will earn Odyssey Points, which will unlock higher-tier rewards and leaderboard eligibility in later chapters.

Tasks include:

• Follow our official X (Twitter) account
• Join the Telegram community
• Join the Discord community
• Like and retweet the campaign post
• Read the introduction and complete a short quiz

After completing these tasks, your wallet will be marked as an Early Participant.

Chapter 2 — The Boot (Coming Soon)

Light Product Experience (Initial Interaction)

Chapter 2 is where you evolve from a “participant” into a real “user.”

In this phase, you will enter the BEEP AI DApp, complete an on-chain interaction, and join or create a Squad to collaborate with others.

Rewards in this chapter are significantly higher than in Chapter 1, and Squad points will begin accumulating toward future rankings.

DApp: https://app.justbeep.it/

Tasks include:

• Deposit ≥ $5 in the DApp
• Join or Create a Squad
• Invite at least 1 friend via your referral link

⚠️ Reminder (Very Important)

Before completing Chapter 2 tasks, please confirm:

• You have logged in to the BEEP platform
• You have connected your Sui wallet address
• You are using the same wallet on Galxe

BEEP only records the first connected wallet address
Changing wallets will result in failed verification

Chapter 3 — The Execute (Coming Soon)

Deep Conversion (Full Trading Loop)

Chapter 3 is the core stage of the Odyssey and represents the true conversion point.

In this phase, users will enter the trading system, complete real interactions, and engage with AI Agents.

Tasks include:

• Complete at least 1 trade using the BEEP Agent Trader

The system will verify:

• Whether a valid Agent trade record exists

Rewards & Weight

Chapter 3 carries the highest points weight and will have the greatest impact on final rankings.

Final Notes

Throughout the campaign, please ensure:

• Always use the same Sui wallet
• Keep the wallet consistent between BEEP and Galxe
• Complete wallet connection in advance

👉 Using different wallets = verification failure

Start Your Odyssey

From signal, to system, to execution.

Signal → Boot → Execute

The era of AI Agent trading begins here.

A new system designed to reward direct contribution while helping your entire Squad unlock more upside.

Most referral systems are forgettable.

You get a link. You send it around. Maybe you earn something once. Maybe you do not. Either way, there is usually no real sense of momentum, no shared identity, and no reason for the network to keep compounding after the first invite.

Beep Squads changes that.

Today, we’re introducing Squads + Referrals — a new growth system built to make participation more social, more rewarding, and more aligned with how strong networks actually grow.

This is not just a referral feature. It is a coordinated earning layer inside Beep.

With Squads, users can now create or join a team, invite friends into the product, earn from direct referrals, and help their Squad unlock more capacity and stronger boosts over time. The more real activity your network brings in, the more value your Squad can create for everyone inside it.

That is the core idea: individual contribution should create shared upside.

Why We Built It

Beep is building internet-native finance around participation.

That means the product should not only reward passive usage. It should reward the users who actively help grow the network — the ones who bring in new capital, new activity, and new community energy.

Traditional referral systems are too flat. They can drive short-term distribution, but they rarely create real coordination. On the other end of the spectrum, overly complex systems become hard to understand, hard to trust, and hard to sustain.

We wanted something simpler and stronger.

A system that:

• rewards direct contribution,
• encourages team formation,
• creates visible status and progression,
• and gives users a reason to care not just about their own activity, but about the growth of their broader network inside Beep.

That is what Squads are designed to do.

What Is a Squad?

A Squad is your team inside Beep.

After connecting your wallet and making your first deposit, you will choose one of two paths:

• Create a Squad and become its Captain
• Join a Squad using an invite code from another user

From that point on, your Squad becomes the shared container for growth.

It tracks the combined participation of everyone inside it, including:

• total members,
• total value locked,
• and cumulative fees generated

That shared activity determines whether your Squad unlocks expanded capacity and what kind of boost the Squad earns.

In other words, a Squad is not just a chat-group-style label. It is an actual growth unit inside the product.

How Squads Work

Every new Squad starts with a 25-member cap, including the Captain.

That count includes everyone in the Squad hierarchy — whether they were invited directly by the Captain or brought in by other Members later on.

Once a Squad reaches $50,000 in total TVL, it unlocks and expands from 25 members to 50 members.

When that happens:

• the Squad gains expanded capacity,
• the Captain receives a Commander Badge,
• and the full Squad gets a visual status upgrade in the product

This creates a simple but powerful dynamic: the most active and committed Squads earn the right to grow bigger.

Referrals Inside the Squad

Within every Squad is a referral tree.

When you invite someone using your personal invite code, two things happen at once:

• they become your direct referral
• they automatically join your Squad

That structure matters because it means your invite creates value on two levels.

First, it creates personal upside for you through referral commissions.

Second, it creates collective upside for your Squad by increasing its TVL and activity, helping the group move toward unlocks and boosts.

So your referrals are not just helping you. They are helping your entire team level up.

How You Earn

Beep’s referral system uses a single-level direct commission model.

That means you earn only from the users you directly invite. There are no multi-level payouts from deeper branches.

We chose this structure intentionally. It keeps the system clear, direct, and easy to understand, while rewarding users based on actual contribution.

When your direct referrals generate points through deposit or trading activity, you receive a percentage of those points as commission.

Your commission rate depends on your own contribution level:

• Level 1: under $100 in direct referral fees generated → 10% commission
• Level 2: $100 to $500 in direct referral fees generated → 15% commission
• Level 3: above $500 in direct referral fees generated → 20% commission

A simple example:

If you invite a user who generates 1,000 points, and you are at Level 2, you earn 150 points.

At the same time, that user’s assets also contribute to your Squad’s TVL — which helps your group unlock more capacity and stronger boosts.

One invite. Two forms of upside.

Why Squad Activity Matters

The real power of Squads is that they reward more than just inviting.

Each Squad receives a global mining speed boost based on the cumulative fees generated by the Squad.

The tiers are:

• The Crew — under $1,000 in cumulative fees → +2%
• The Syndicate — $1,000 to $3,000 → +5%
• The Consortium — above $3,000 → +10%

These boosts apply to all members in the Squad, including the Captain.

That means a strong Squad creates value for everyone inside it. Even if you are not the largest inviter, you still benefit from being part of a more active and successful group.

This is what makes Beep Squads different from a normal referral program. It is not just about sending links. It is about building a coordinated, high-conviction network inside the product.

Captains and Members

There are two roles inside the system.

Captains

Captains create Squads.

They receive the main Squad code, lead the initial formation of the team, and help drive the Squad toward unlocks and higher status. They can invite users directly and grow the Squad from day one.

Captains earn referral commissions from the users they directly invite. They do not earn commissions from referrals made by other Members unless those users were invited by the Captain personally.

Members

Members join an existing Squad.

Once inside, they receive their own invite code and can start bringing in direct referrals. Those new users join the same Squad and also become the Member’s direct referrals.

That means every Member can contribute to Squad growth while also building their own earning stream.

Everyone can help grow the network. Everyone can participate in the upside.

Built Around Real Participation

The Beep Squad system is designed to reward real activity, not temporary spikes.

A Squad unlocks when it reaches $50,000 in total TVL, based on real-time assets.

If that TVL later falls below the threshold, the Squad re-locks. If the Squad had already expanded beyond 25 members, it remains capped at its current size and cannot continue growing unless TVL is restored. And if assets are not brought back above the threshold within 3 days, the Squad boost is suspended.

That keeps the system grounded in real participation and makes Squad status something that has to be maintained — not gamed.

Why This Matters for Beep

This launch is bigger than a referral update.

It is a growth framework designed to turn users into network builders.

It gives people a reason to recruit intentionally, participate more deeply, and coordinate with others around shared upside. It creates stronger identity inside the product and helps growth become more visible, more social, and more compounding over time.

The best networks are built when personal incentives and collective incentives move in the same direction.

That is what Beep Squads is designed to do.

What Comes Next

As Squads roll out, we expect them to drive a new set of behaviors inside Beep:

• more users creating or joining a Squad right after their first deposit
• more direct invites from engaged users
• more coordinated growth around TVL unlocks
• more competition around boost tiers and Squad status
• and stronger long-term participation across the network

The strongest users on Beep will not just be users.

They will be organizers, recruiters, and builders of high-signal communities inside the product.

Quick FAQ

Do I need to join or create a Squad to activate point mining?
Yes. In this phase, users need to join or create a Squad to activate point mining.

Can Members invite other users?
Yes. Every Member receives their own invite code and can bring in direct referrals.

Where do my referrals go?
Your referrals automatically join the same Squad as you and are also registered as your direct referrals.

Do Captains earn from everyone in the Squad?
No. Captains only earn commissions from users they directly invite.

Is this a multi-level referral system?
No. The MVP uses direct-only commissions.

What happens when a Squad reaches 25 members?
It becomes full unless it has unlocked by reaching $50,000 in total TVL.

What happens when a Squad reaches $50,000 TVL?
It unlocks, expands from 25 to 50 members, and receives upgraded status in the product.

Can I join one Squad and later create my own?
In the MVP, no. Once you join a Squad via invite code, you are locked into that Squad.

Final Thought

The best growth systems do not just reward attention. They reward alignment.

Beep Squads is built around a simple idea: the users who help grow the network should share in the upside they create.

Not through shallow one-time referrals, but through a structure that lets people build teams, deepen participation, and benefit from both direct contribution and collective momentum.

Create your Squad. Invite your people. Grow together. Earn together. https://app.justbeep.it/

Introduction

The first two pieces established the shape of the problem. Most agents die the moment they meet real capital. A system built to survive that requires four layers working in concert, each one accountable to actual outcomes.

But a system cannot justify itself through logic alone. Behind it has to be a team that understands the full shape of the problem, not just one corner of it.

This piece asks one question: why us?

Why Us?

The answer lies in how we framed the problem from the start. We saw it as a systems problem spanning AI, trading, and decentralized coordination, not a single-point problem that a better model would eventually solve.

Team-Market Fit

The founding team has already done this before. We built and launched vertical AI models that disrupted white collar professions, shipping production systems that replaced human workflows at scale.

We led systematic trading teams at institutional firms and built TradFi systems at PayPal and Walmart. Multiple startup ventures scaled to $200M+ revenue. The dev team averages 12 years experience across blockchain infrastructure, ML systems, and quantitative strategies.

The only difference: we are now pointing our capabilities at generating institutional yield. Same rigor, same obsession with production-grade systems, new target.

This background matters because it explains how we arrived at Beep: starting from production systems and working backwards to what Beep had to be.

Before founding Beep, we had already built vertical AI together. Arpan Nanavati and Juan Guardado co-founded Cimphony, using AI agents to deliver scalable, fixed-price legal services for startups. Not a demo. Not a thin AI wrapper. A system that entered real business workflows and replaced professional labor at scale.

From legal services to agentic finance, the industry changed. The underlying logic did not: use AI to automate complex, high-value professional workflows, and run the system in production against real consequences.

Unique Understanding Of The Problem

The problems Beep solves sit at the intersection of three domains that rarely overlap: quantitative trading, AI model training, and decentralized infrastructure. Most teams have one. Few have two. We have all three.

We have trained vertical AI models from scratch. We have built alpha capture systems. We know what it takes to ship models that work in production. Not demos, not backtests. Real systems managing real capital.

Beep is the synthesis: proprietary signal integration, agent-sourced strategies with skin-in-the-game quality control, and decentralized training on actual P&L outcomes. The model that eats all of finance is built by a team of AI researchers and quant traders who have worked at both ends of the problem.

We spent years arriving at conclusions that most teams never fully confront.

Intelligence is not the same as returns. Strategy generation is not the same as execution, and a better model does not automatically become better capital allocation. Real trading intelligence can only be trained through real outcomes.

This is why Beep was never designed to produce better-looking analysis. It was designed to build a system where strategies are submitted, tested, rewarded, pruned, and turned into stronger models through actual outcomes.

Closing

Beep did not begin as a product idea. It began as a hard conclusion.

If AI agents are going to create durable value in financial markets, the system behind them has to be built by people who understand how markets punish illusions, and how intelligence compounds through outcomes rather than through better prompts.

We spent years developing that understanding across three domains that rarely overlap. Beep is the first time we have put all three to work inside the same system.

A Practical Guide to Agentic Finance

Beep is designed to make financial infrastructure programmable for both humans and autonomous agents. Instead of manually managing capital across protocols, users can deploy assets into an environment where AI agents continuously monitor opportunities and optimize capital allocation.

This tutorial walks through the core actions on Beep:

AI Treasury

• Creating your account
• Depositing and Managing assets
• 1-click AI yield for multiple assets
• Integrating Beep Pay into Dapps and vibenvironments.

AI Trading

• Create Your AI Trading Agent
• Monitor Your Agent

Let’s dive in.

AI Treasury

1. Create Your Beep Account

Sign in with your Social account or wallet.

Once connected, the system automatically creates an AI Treasury, designed to interact with AI agents and DeFi protocols.

2. Deposit Assets into Your Beep Vault

After creating your account, the next step is funding your AI treasury.

Supported assets on Sui can be deposited directly into the Beep Vault and become available fo payments, trading agents and yield strategies

To deposit:

• Navigate to the My Balances tab
• Choose the supported asset
• Connect your wallet (if you logged in using social account)
• Enter the amount of assets you want to deposit for AI yield
• Confirm the transaction in your wallet

Supported assets can include:

• USDC
• USDsui
• SUI
• xBTC
• WAL
• DEEP

3. 1-click Earn for depositing multiple assets

User can deposit multiple assets to Beep, manage and earn agentic yield follow this tutorial video:

4. Send and Withdraw Assets

Users can freely move funds between wallets, protocols, and applications.

Sending and Withdrawing assets

To send/withdraw assets:

• Click on the asset you want to send
• Click Send
• Enter amount and wallet address
• Confirm the transaction

5. Integrate Beep Pay to your vibenvironment

Beep Pay enables agent-native payments on Sui Network through the a402 standard, allowing applications and AI agents to pay for services programmatically.

Developers can integrate Beep Pay into any application, API, or AI environment with minimal setup.

AI Trading

Beep allows users to launch their own AI trading agents that analyze markets and execute trades automatically.

These agents operate continuously, scanning market data, evaluating signals, and executing strategies based on the parameters you define.

Create Your AI Trading Agent

To start:

• Navigate to the AI Trading tab
• Click Create Trading Agent
• Select an Agent Template to begin building your agent

You will then configure your agent’s core setup.

Agent Setup

• Agent Tag
  Give your agent a name to identify it in your dashboard.
• Model Selection
  Choose the AI model your agent will use to analyze markets and generate trading decisions.
• Market Selection
  Select the markets your agent will monitor and trade in.

Strategy Configuration

Next, customize how your agent behaves in live markets.

• Risk Tolerance
  Define how aggressively your agent should trade and how much risk it can take.
• Frequency
  Determine how often the agent checks market conditions and evaluates trading opportunities.
• Allocation
  Fund your agent’s dedicated wallet. This capital will be used to open and manage trading positions.
• Max Trading Position
  Set the maximum size your agent can allocate to a single market.
• Maximum Active Trades
  Define how many positions your agent can hold simultaneously.

Once configuration is complete, the agent will begin operating automatically.

Monitor Your Agent

Now, your agent will:

• Analyze market conditions
• Generate trade signals
• Execute trades within the parameters you defined

To monitor your Agent Trader, Beep provides full transparency into agent activity.

Model Chat
View the reasoning behind your agent’s decisions and analysis.

Active Positions
Monitor current trades, asset exposure, and real-time PnL.

Completed Trades
Review every trade executed by your agent, including entry, exit, and volume in details.

Your AI trading agent now operates continuously, adapting to market conditions and executing strategies on your behalf.

FAQ

1. Q: Is my money at risk in Agent Trader R2?
   A: Yes. In Agent Trader R2 you allocate and trade with your own capital. Outcomes depend on the strategy, model, and parameters you choose.
2. Q: Do I control the strategy after deploy?
   A: Yes. You can tune parameters, pause, stop, and withdraw at any time.
3. Q: Can I choose my AI model?
   A: Yes. Each strategy runs on a user-selected AI model.
4. Q: What does “pause” do?
   A: Stops new actions (no new orders). Existing positions may remain unless the strategy is configured to reduce.
5. Q: What does “stop” do?
   A: Stops the strategy and follows the configured unwind behavior (close/reduce positions) before ending.
6. Q: What does “withdraw” do?
   A: Removes capital from the strategy; behavior depends on whether positions must be reduced first.
7. Q: How do rewards work?
   A: Rewards v2 are driven by retained allocation and trading volume, with agent-mode boosts and growth via squads/referrals.
8. Q: Are points a token / do they have USD value?
   A: No. Points are non-transferable and have no cash or crypto value.
9. Q: What markets/assets are supported?
   A: See the in-app asset whitelist; availability depends on launch configuration.
10. Q: Something looks wrong/unexpected. What should I do?
    A: Pause the strategy, review the rationale feed and parameter settings, then share details with support.
11. Q: What are the agent trading fees?
    A: Agent trading fee is 5%.
12. Q: Do users need to pay for AI agents? Can users copy other people’s AI agents without spending money?
    A: Yes, there is a creation agent fee and in return reward points as per the rewards framework.
13. Q: Why do different agents have different prices?
    A: Prices vary depending on the frequency of the agent runs — higher frequency = higher price, lower frequency = lower price.
14. Q: Is there copy trading or something for free?
    A: Copy trading = copy AI agents = AI costs are passed through to users.
15. Q: Is the agent actually live, or is nothing happening after deposit?
    A: Yes, the agent is live. Actions are not always instant. Execution depends on the frequency you selected at setup, plus current market conditions and strategy logic. To see what the agent is doing, check the model chat inside the agent.
16. Q: My agent hasn’t made a move yet. Are my funds safe?
    A: Yes. If the agent has not executed yet, that does not mean funds are lost. Your funds remain tied to the agent and are deployed based on the strategy and execution timing you chose.
17. Q: How do I stop the agent and withdraw?
    A: Go to My Agents, then Agent Details, then Agent Toggle. Turn the toggle off to stop the agent and trigger withdrawal.
18. Q: Is there a way to exit without fully killing the agent?
    A: Right now, the main exit flow is to deactivate the agent, which triggers withdrawal.
19. Q: Why do I pay the fee again if I close and reopen?
    A: Because a closed agent is currently treated as a fresh activation, not a simple resume.
20. Q: Can I change config after funding?
    A: Yes, you can edit the strategy configurations. Agent strategy configuration can be paid using Beep Reward Points.
21. Q: Why does adding funds feel delayed or not real-time?
    A: Deposits are not always reflected as an instant strategy action. Capital gets applied according to the agent’s execution cycle and current state, which can make the experience feel less real-time than expected.
22. Q: Do I need to create a new agent every time I restart?
    A: If you fully close the agent, restarting requires creating a new one under the current flow.
23. Q: Is Agent Trader R2 production or still experimental?
    A: R2 is live, but it is still a new agentic trading product. Best practice is to start with a size you are comfortable testing. Monitor behavior in model chat and make sure your execution settings match your expectations. Agent trading is early and exhilarating, but selecting the wrong strategy can result in losses.
24. Q: Where do I see what the agent is thinking or doing?
    A: Open the model chat inside the agent. That is the best place to track what the model is evaluating, whether it is waiting, and when it decides to act.
25. Q: Where do funds go after I stop an agent?
    A: Once an agent is stopped, funds are withdrawn back into your AI Treasury.

Introduction

Last time, we established one thing: most agents do not fail because they cannot generate strategies. They fail the moment a strategy meets real capital. Slippage, noise, risk controls, execution discipline, adverse counterparties, tail volatility. Everything that makes “reasonable on paper” irrelevant in practice.

So the question was never whether your agent is smart enough. The question that actually matters is whether your agent can survive in live markets and get stronger each time it does.

This piece answers only one question: what does a system built to do that actually look like?

Beep: The Four Layers of a Live-Market Alpha System

Layer 1: Proprietary Agents

Without proprietary, verifiable signals and action agents, every strategy eventually becomes a text game that looks reasonable until it meets real money.

Beep runs its own agents continuously on data streams at high fidelity — live a402 transaction metadata (revealing information value and intent at scale), cognitive prediction-market signals, order-flow micro-dynamics, and cross-asset yield relationships. Each agent combines LLM semantic reasoning with reinforcement learning optimized directly on realized P&L.

But proprietary signals alone solve only half the problem. The other half is supply.

Layer 2: Tokenized Agent Strategies

No single team can exhaust all alpha. Strategy supply has to become an open network.

Any agent in the ecosystem plugs in instantly. They receive premium signals via a402 payments, run their own searches, and submit strategies. Incentives are direct and tied to real economic outcomes, not reputation. Agents propose multi-leg strategies that others refine.

Open supply creates volume. Volume without quality control creates noise.

Layer 3: Ruthless Quality Control

Markets do not reward cleverness. They reward repeatable risk-adjusted returns. So the selection mechanism has to be just as unforgiving as the market itself.

Every submitted strategy runs as an independent pod. Performance is measured exclusively by Information Ratio over rolling windows. High performers receive more capital. Underperformers are pruned. Originality scoring rejects or down-weights highly correlated strategies. Mandatory paper-trading precedes live capital. Survival is earned, not assumed.

Ruthless selection produces outcomes. Outcomes, fed back correctly, produce something more valuable than any single strategy.

Layer 4: The Alpha Model

Without a training loop, there is no compounding. The system does not get stronger because it is well-designed. It gets stronger because of what happens to it in the market.

Every trade, every P&L tick, every a402 payment, and every pod decision flows back into a decentralized training corpus. The model is narrow, trading-specialized, and optimized end-to-end on real outcomes. As the network of AI agents grows, the training data becomes richer and more proprietary. The flywheel self-reinforces: better model → better yields → more agents → richer data → stronger model.

Closing

These four layers are not a blueprint. They are a machine that is already running.

Every live trade that flows in, every P&L tick that feeds back, every underperforming strategy that gets cut. The system sharpens itself. No one is tuning it by hand. The outcomes do that.

This is what Beep is today.

What it becomes next is a different conversation.