KREXA

The credit layer for the agentic economy.

Compute, payments, and borrowing — bundled into one atomic transaction.

0

Active agents

0

Repayment rate

0

Protocol revenue

Live

On mainnet

THE PROBLEM

Agents need three things to operate.
Today, each is a separate, broken system.

Problem 1

Agentic Payments

Agents can't hold cards or sign ToS. No payment rail is built for autonomous software.

Coinbase x402 & Stripe ACP launched May 2025 — proving the gap exists.
Visa: "Machine-to-machine payments will exceed human payments by 2030."

Problem 2

Borrowing Capital

Agents need capital to trade, pay for APIs, and scale — but have no credit identity or collateral history.

a16z (2025): "The missing primitive for autonomous agents is under-collateralized credit."
No DeFi protocol underwrites agent-specific risk today.

Problem 3

Compute Lending

AI inference costs $55B+ in 2026. Agents pay upfront or stop. There's no "borrow now, repay from earnings" model.

Bain & Co (2025): AI infra spend will hit $990B by 2028.
9M+ paying OpenAI users; avg enterprise bill $85K/yr.

1 TX Krexa bundles compute + payments + credit
into one atomic transaction.

x402

Stripe ACP

Visa

Google AP2

WHO IS THE BORROWER?

The developers and operators behind every AI agent.

Not the borrower →

Cursor ($1B+)

ElevenLabs ($330M) Sierra ($100M+)

Trading Agent

Prediction market trader

50%+ APY — verified on-chain

Needs capital BEFORE first trade

Float gap: $500–$20K

Voice Agent Operator

Dental receptionist AI

$500/clinic/mo · API costs $150 upfront

Clients pay NET-30

Float gap at 20 clinics: $6–8K

Coding Agent Operator

Vertical dev shop

$2K/client/mo · $800 API costs upfront

Clients pay NET-30

Float gap at 15 clients: $12K

One pattern: costs hit before revenue arrives.

THE PRODUCT

Three surfaces. One credit engine.

krexa.xyz

Web dashboard for operators

Register agent · Connect billing

Monitor credit · Track repayments

Onboard in 2 minutes

MCP Server

46 MCPs · 75+ APIs on credit

Agent calls any tool — Krexa pays

Jupiter, OKX, Pyth, SNS, Pay.sh

Plug into any AI framework

CLI + SDK

krexa draw --amount 5000

krexa pay --to 0x... --usd 100

krexa score --agent 0x...

Programmatic access to credit

Agent calls an API → Krexa pays → Agent earns → Revenue Router auto-repays

One atomic transaction. Compute + payment + borrowing — bundled.

RISK FRAMEWORK

Both undercollateralized. Both scored by subscription history.

Low Risk

16% APR

Standard repayment — agent acts, earns, repays on schedule

Use cases: Service agents, coding agents, data agents, voice operators

Enforcement: Revenue Router auto-splits incoming payments

Whitelisted actions only — agent can only call approved APIs and services

Determined at registration · Foolproof constraints

High Risk

Smart Contract Framework

On-chain enforcement — PDA custody, auto-liquidation, position limits

Use cases: Perps trading, prediction markets, arbitrage, leveraged strategies

Enforcement: Smart contract controls every transaction — agent never holds keys

Whitelisted actions only — can only execute pre-approved trade types

Higher credit line · Structural enforcement

Both tiers scored by compute subscription history — not trust, not collateral.

Risk is determined at registration. Actions are whitelisted. The code enforces it.

TRACTION

Building in public.

0

Active agents

0

Repayment rate

0

Volume routed

0

Credit deployed

0

Protocol revenue

0

MCPs connected

Solana

Jupiter

OKX

SNS

Pay.sh

Monad

BUSINESS MODEL

Blended yield across two risk tiers.

Low Risk Tier

16% APR

Service agents, coding agents, voice operators

Standard repayment · Whitelisted actions

Cost of capital: 4–6%

Net spread: 10–12%

~70% of portfolio · Lower yield, lower default

High Risk Tier

24% APR

Perps, prediction markets, arbitrage, leverage

Smart contract enforcement · Auto-liquidation

Cost of capital: 4–6%

Net spread: 18–20%

~30% of portfolio · Higher yield, structural protection

Blended Portfolio Yield

~19% APR

Net spread after cost of capital: ~13–15%

+ Protocol fee 10% of every tx

+ MCP usage margin on 75+ APIs

+ KrexaScore queries (at scale)

More agents → more credit → more fees → better data → more agents

THE VISION

Credit is the GTM. Credit bureau is the endgame.

NOW

Compute-backed credit

Revenue Router generates data

Identity seeds as byproduct

LIVE

12–18 MONTHS

KrexaScore = reference signal

Platforms query before routing

Free — distribution first

IN PROGRESS

DESTINATION

Credit bureau for agents

Every query = revenue event

Network effects compound

THE VISION

Equifax + Experian + TransUnion = $0
We're building the agent equivalent.

PARTNERSHIPS

The financial OS composes with everything.

DeFi & Trading

Jupiter

OKX

Meteora

Pyth

Phoenix

Identity & Infrastructure

Solana

SNS

Pay.sh

Partners

HeyAnna OOBE NoahAI

Solana Network State

Expanding

Monad — this month + more coming

46 MCPs · 75+ APIs on credit · 5 Anchor programs · Multi-chain

🔒

🔓 ALPHA

Your OpenAI bill is your credit score.

$0 AI compute spend globally

$0 Enterprise Anthropic bill (avg)

$0 SMB Anthropic bill (avg)

0 Paying OpenAI business users

0 Companies $1M+/yr on Claude

Connect OpenAI / Anthropic

Verify spend + trend

KrexaScore +60 pts max

Credit line = 2.5x spend

$2K/mo→$5K credit

$10K/mo→$25K credit

$50K/mo→$125K credit

The richest credit signal in history —
verifiable, consistent, impossible to fake.

TEAM

The team building the agent credit layer.

YK

YATHARTH KHER

Co-founder · Protocol Architect

· Built 5 Anchor programs live on Solana mainnet

· ADGM Fintech Talent of the Year

· 15 hackathons — UAE, US, UK

· AI + blockchain researcher since 2020

@yatharthkher

MB

MOHIT BANSAL

Co-founder · Strategy & Capital

· Chief of Staff to Sandeep Nailwal (Polygon)

· Architected $75M raise at Avail

· Facebook x Polygon integration

· IIT Bombay · FRM · CFA Level 2

@mohitb026

THE ASK

$0

Pre-Seed · $20M valuation · First close open.

"Before the data speaks for itself."

0

Active borrowers

0

Monthly revenue

0

Default rate

16-week targets

LET'S BUILD

Let's build the agentic economy together.

Scan to connect on Telegram

yatharth@krexa.xyz

krexa.xyz · @krexa_xyz · @yatharthkher

Keyboard Shortcuts

Presenter Notes

The credit layer for the agentic economy.

Agents need three things to operate.
Today, each is a separate, broken system.

Agentic Payments

Borrowing Capital

Compute Lending

The developers and operators behind every AI agent.

Trading Agent

Voice Agent Operator

Coding Agent Operator

Three surfaces. One credit engine.

krexa.xyz

MCP Server

CLI + SDK

Both undercollateralized. Both scored by subscription history.

Low Risk

High Risk

Building in public.

Blended yield across two risk tiers.

Low Risk Tier

High Risk Tier

Credit is the GTM. Credit bureau is the endgame.

NOW

12–18 MONTHS

DESTINATION

The financial OS composes with everything.

Your OpenAI bill is your credit score.

The team building the agent credit layer.

YATHARTH KHER

MOHIT BANSAL

Let's build the agentic economy together.

Keyboard Shortcuts

Presenter Notes

The credit layer for the agentic economy.

Agents need three things to operate.Today, each is a separate, broken system.

Agentic Payments

Borrowing Capital

Compute Lending

The developers and operators behind every AI agent.

Trading Agent

Voice Agent Operator

Coding Agent Operator

Three surfaces. One credit engine.

krexa.xyz

MCP Server

CLI + SDK

Both undercollateralized. Both scored by subscription history.

Low Risk

High Risk

Building in public.

Blended yield across two risk tiers.

Low Risk Tier

High Risk Tier

Credit is the GTM. Credit bureau is the endgame.

NOW

12–18 MONTHS

DESTINATION

The financial OS composes with everything.

Your OpenAI bill is your credit score.

The team building the agent credit layer.

YATHARTH KHER

MOHIT BANSAL

Let's build the agentic economy together.

Agents need three things to operate.
Today, each is a separate, broken system.