The AI API Intelligence Layer
One query. Optimal routing. Transparent costs.
We're building the brain between AI agents and the world's APIs—the intelligence layer that transforms chaos into clarity. Think Stripe for the API / Agentic AI era.

by Saurabh Sethi

Founder and CEO

1

The Founders
We've lived this problem
Saurabh Sethi
Co-Founder & CEO
Built identity products at GSMA, Zumigo and iProov. Spent years integrating fragmented fraud vendors.
  • Former Innovation Director, GSMA Mobile Connect
  • Managing Director, India for Zumigo Inc.
  • Global VP Product at iProov (closed Series B $70M in 2022)
  • Deep expertise in fraud prevention systems
Rakesh Rohilla
Co-Founder & CTO
20 years building telecom APIs and Infrastructure. Knows why enterprise integrations fail.
  • Former Technical Director, Airtel
  • Built API infrastructure at scale
  • Expert in enterprise system architecture

+ A dedicated team of 4 engineers. We iterate, build, ship daily.

2

AI Agents Have Access to 17,000 Tools
They Have No Idea Which One to Use
Imagine you need to check the weather. Your phone has 50 weather apps installed. One is best for your city. Another excels at severe storm warnings. A third has the most accurate 10-day forecasts. But you have no idea which is which—so you just pick one randomly and hope it's right.
This is exactly what's happening with AI agents today. The Model Context Protocol gave them access to 17,000+ tools. But there's no intelligence layer telling them which tool is actually best for the job.
When AI agents connect to 150+ tools, their context windows consume 150,000 tokens just loading tool definitions. Anthropic documented a 98.7% reduction (150,000 → 2,000 tokens) using intelligent tool selection. The problem is exponential.
17K+
MCP Servers Live
Active and growing daily
0
Intelligent Routing Layers
Until now
We Need
Intelligent understanding - Capable Routing - Continuous Learning

3

Enterprises Are Drowning in Data APIs
Fraud teams alone deal with more than 50 vendors. KYC operations require another 30-plus integrations. Each API brings different pricing structures, latency profiles, geographic coverage, and reliability guarantees. The complexity compounds exponentially as teams attempt to manually orchestrate these services.
No Intelligent Routing
Teams manually select vendors based on guesswork and outdated assumptions, not real-time data or performance metrics. Selection becomes political rather than optimal.
No Cost Visibility
Black-box solutions hide which vendors they actually use and what they truly cost. Finance teams receive opaque bills with no audit trail or optimisation path.
No Failover Logic
When one API fails or degrades, the entire workflow breaks. Manual intervention required. Customer experience suffers whilst engineers scramble to reroute.
No Learning System
Every query starts from zero with no memory or optimisation over time. The system never gets smarter, faster, or cheaper despite processing millions of calls.

The Result: Projects take 3–6 months to go live. Integration costs often exceed the actual API costs. Meanwhile, fraud patterns evolve in hours whilst your frozen stack struggles to keep pace. The gap between threat velocity and response capability widens dangerously.

4

Outris: The AI API Intelligence Layer
Outris sits between your AI agents—or your existing systems—and the world's data APIs. We handle the genuinely hard problems: intent resolution, vendor selection, cost optimisation, automatic failover, and continuous learning from every interaction.
Parse: Natural Language Understanding
Natural language or structured input becomes crystal-clear intent. We understand what you're truly asking, not just the words you use.
Plan: Intelligent and automated Routing
We route queries to the best vendors based on geography, data freshness, cost, and reliability. Automatic failover. Parallel processing. Optimised every time.
Execute: Unified and seamless responses
Unified data format. Compliance-ready. Audit trails included. Results delivered in seconds, not hours. Everything your team needs in one clean response.
Learn and Add: Automated API Additions
Every query improves our routing intelligence. And we add new vendors without any code/engineering effort. No deployment cycles. Just continuous expansion.

5

Three Ways to Access
Same Intelligence. Three Interfaces.
We've built a platform, not a point solution. Access our intelligence layer however best fits your existing technology stack and team workflows. Maximum flexibility without compromising power.
REST API
Who uses it: Enterprise backend systems, existing fraud stacks, and production workflows
How it works: Standard POST /workflows/fraud-check with JSON payload integration
Value delivered: Drop-in replacement for direct vendor calls, adding routing intelligence instantly
Natural Language Dashboard
Who uses it: Fraud analysts, compliance teams, operations staff without technical backgrounds
How it works: Type questions in plain English, receive structured results with full attribution
Value delivered: Zero technical knowledge required. Instant productivity.
MCP Server
Who uses it: AI agents including Claude, GPT, Cursor, and custom agent architectures
How it works: Agent calls orchestrator_query("is this phone sketchy?") and receives intelligent routing
Value delivered: Transform 300 disparate tools into one intelligent, context-aware tool

MCP is a delivery mechanism. REST is a delivery mechanism. The natural language interface is a delivery mechanism. The intelligence layer itself is protocol-agnostic, working seamlessly across all three whilst maintaining consistent routing logic and learning.

6

See It In Action (Traceflow Workflow)
Real Query. Real Results. Seconds Not Hours.
Here's precisely what happens when a fraud analyst types a simple question: "Check if this phone number is risky in India"
Execution Time: 2.1 seconds
Parallel orchestration across 5 vendors
💰 Customer Price: ₹3.50 ($0.042)*
Enterprise pricing (Experian pilot)
🔧 Our Vendor Cost: ₹1.02 ($0.012)
Transparent vendor attribution
📊 Gross Margin: 71%
vs. 73% on individual API sales
🎯 Vendors Orchestrated:

3-5 APIs → 1 unified result

Current enterprise customers pay ₹0.40-0.50 per individual API call, and test multiple vendors to reach a final solution.
TraceFlow bundles all needed API's from a single NLP command, and into one comprehensive fraud investigation at ₹3.50 — delivering 8-10x more intelligence at 7x the price point. vs. $0.15-0.30 charged by black-box competitors like Sift/Sardine.

7

Why Fraud First
We Started With the Hardest Vertical
If we can orchestrate intelligently in fraud detection, we can orchestrate anywhere. Fraud represents the most demanding test case—combining real-time requirements, fragmented vendor landscape, and strict compliance obligations. Success here proves the platform works everywhere.
Highest Stakes
  • Real-time decisions with immediate financial consequences
  • False positives directly cost revenue and customer relationships
  • False negatives enable fraud, leading to losses and reputation damage
  • Every millisecond of latency matters in production
Most Fragmented
  • More than 50 vendors operating globally
  • Zero standards across different API implementations
  • Data quality varies wildly between providers
  • Significant geographic coverage gaps requiring intelligent routing
Strictest Compliance
  • Complex PII handling requirements across jurisdictions
  • Mandatory audit trail obligations for regulatory review
  • Sophisticated consent management requirements
  • Data residency rules varying by country and industry
Fraud detection is our wedge—the hardest problem that proves our architecture. Once solved, the same intelligent orchestration extends naturally to KYC, AML, network APIs, and any regulated data workflow. The technology scales; the challenges get easier.

8

Traction
Real Customers. Growing Revenue. Zero Churn.
Individual APIs (Current Revenue Base)
  • 2.85M calls/month
  • ₹0.40-0.50 per call
  • $15K MRR
  • 73% gross margin
TraceFlow Workflows (Emerging Premium Tier)
  • In pilot with Experian
  • ₹3.50 per investigation
  • $2K-2.7K/month (unbilled, production starting Q1)
  • 71% gross margin
  • 7x higher ARPU vs. individual APIs
Platform Performance Metrics
94%
Intent Resolution Accuracy
vs. 30% for raw LLM routing
60%
Average Cost Savings
vs. direct vendor calls
85%
Cache Hit Rate
On repeat queries
Product Mix

TraceFlow Opportunity
When customers see the total value that TraceFlow workflows bring (₹3.50), we achieve 7-8x ARPU expansion with same vendor costs. Experian pilot validates premium pricing for packaged intelligence.
More workflows and fully agentic ai orchestrator in the pipeline
  • 3 fraud vendors fully integrated and production-ready
  • 20+ additional vendors in active integration pipeline
  • Enterprise customers in financial services and fintech
  • Deployment times reduced from months to days

9

The Platform Path
Fraud Is the Wedge. The Platform Is the Prize.
Same architectural foundation. New domains unlocked. Expanding total addressable market with each phase whilst leveraging existing infrastructure and intelligence.
1
Now: Fraud & Identity APIs
Phone verification, email validation, device intelligence currently live with paying customers. Three vendors integrated and delivering production results.
2
6 Months: KYC, Compliance, AML, API Marketplace APIs
Document verification, sanctions screening, PEP checks represent adjacent workflows with identical orchestration requirements. Targeting 15+ vendors for integration.
3
12 Months: Network APIs & Beyond
CAMARA, Open Gateway, and carrier APIs leverage founder expertise in telecommunications infrastructure. Targeting 30+ vendor partnerships.
4
24+ Months: Universal AI API Orchestration
Any agent, any API, total governance. Building the trust infrastructure for the entire agentic AI economy. 100+ vendors on self-serve platform.
We're building the trust infrastructure for the AI agent economy—the intelligence layer that makes 17,000 tools feel like one whilst maintaining complete transparency, optimising costs, and improving with every interaction.

10

Massive Opportunity
Market Sizing
1
2
3
1
SOM: $200M
India + UK/EU + (SEA+MENA) fraud prevention in fintech & e-commerce. Our realistic 3-year target within immediate addressable segment.
2
SAM: $2.5B
Global fraud & identity verification API market. Expanding across KYC, AML, compliance orchestration in regulated sectors.
3
TAM: $50B+
Enterprise API management & orchestration across all industries. The full platform opportunity as AI agents proliferate.
The fraud API market alone is experiencing 25% annual growth, driven by increasing digital fraud sophistication and regulatory pressure. Our expansion into adjacent workflows multiplies this opportunity.
As enterprises deploy more AI agents, the need for trusted orchestration layers becomes critical infrastructure—not a nice-to-have, but foundational.
Growth Drivers
  • Fraud API market growing 25% annually as digital transactions increase
  • MCP protocol adoption by OpenAI, Microsoft, and Google accelerates agent deployment
  • AI agents require governance layers enterprise can trust
  • First-mover advantage in intelligent orchestration creates defensible position

11

Competitive Landscape
Only Outris Offers AI-Native Intelligent Orchestration
The market offers various approaches to API management, but none combine intelligent routing with complete transparency and continuous learning. Competitors hide their most important decisions—we expose everything.
Our competitors hide vendor choices to protect margins and avoid difficult questions. We expose everything because transparency is our moat. When customers see exactly what they're paying for and which vendors deliver best results, they never leave.

12

Why We Win
Founder-Market Fit. Technical Moat. Perfect Timing.
Founder-Market Fit
Saurabh Sethi, CEO: Built identity products at GSMA, Zumigo, and iProov. Deep expertise in fraud prevention and API architecture.
Rakesh Rohilla, CTO: 20 years building telecommunications APIs at Airtel. Understands carrier infrastructure and network APIs intimately.
Combined 25+ years in fraud detection, identity verification, and API infrastructure. We've been both vendor AND buyer—we understand pain points from every angle.
Technical Moat
LLM-first architecture with sophisticated planner-executor engine distinguishes our approach. Learned routing improves accuracy with every query processed.
Current performance: 94% accuracy from intent mappings versus just 30% for raw LLM routing without our intelligence layer.
Proprietary vendor performance database grows continuously, creating compound competitive advantages. More usage equals better routing equals more usage.
Perfect Timing
MCP protocol adopted by OpenAI, Microsoft, and Google in 2025 creates massive tailwind. AI agent deployment accelerating across enterprise.
Enterprises desperate for governance layer they can trust. Regulatory pressure intensifying globally. First-mover advantage in intelligent orchestration.
The infrastructure for agentic AI needs trust—we're building that foundation precisely when the market demands it.

We're not learning this market—we've lived in it for decades. The moat isn't the protocol or the API integrations. The moat is the intelligence layer that learns and improves, creating a virtuous cycle competitors cannot easily replicate even with more capital.

13

Raising $3M to Prove the Platform
Use of Funds
Investment Breakdown
Product & Tech (50%)
Expand API coverage to 20+ vendors. Enhance AI orchestration algorithms. Build advanced analytics and reporting capabilities.
Sales (40%)
Enterprise customer acquisition in UK/EU markets. Partner programme development. Sales engineering team expansion.
Operations (10%)
Infrastructure scaling. Customer success team. Compliance and security certifications.
Milestones This Round - in next 18 months, with this capital, we will →
UK/EU Market Expansion (5 Enterprise customers)
20+ Vendor APIs Integrated
Prove Workflow Reuse Model. Featured placement in Anthropic MCP directory
$750K ARR Target

14

Let's Talk
There's a Lot More to Show
We'd love to walk you through a live demonstration of the Outris platform in action—showing intelligent routing decisions in real-time, complete cost transparency with per-call attribution, and the seamless MCP integration that transforms 300 tools into one intelligent interface.
You'll see exactly how the planner-executor engine reasons about vendor selection, how automatic failover handles degraded APIs, and how the learning system improves accuracy with every query. The demo typically runs 30 minutes with ample time for technical deep-dive questions.
Saurabh Sethi
Founder & CEO
Schedule Demo

15

Appendix A: The Technical Journey
AI Agent Network
A Master AI Agent. Spawns multiple sub-agents to discover, parse, plan and execute API workflows
Fraud Vendor API Processing
Ability to import new Vendor API's quickly. Already connected with 3 fraud vendors.
API and Algorithms Sequencing
Real-time execution of APIs as well as custom algorithms on customer and new derived data
Workflow Persistence
Custom API endpoint creation for future reuse of the same workflow
Feedback and Memory
Agents learn about API's on the go. Latency, response structures, empty / filled fields - all feedback is used for new queries

16

Appendix B: Unit Economics
$0.01
Average Revenue Per Call
Orchestration fee
$0.0
Average Cost Per Call
Vendor + infrastructure
75%
Gross Margin
Sustainable economics
Customer Economics:
  • Customer LTV: $180K over 3 years
  • Customer CAC: $15K fully loaded
  • LTV/CAC ratio: 12x demonstrating excellent efficiency
  • Payback period: 4 months on average

17

Appendix C: Current API Suite
1
Phone Telco, Social Footprint and Scores (Risk/ Fraud)
Carrier information, porting history, social media presence, reputation scores
2
Phone-to-Address Lookup
Registered address verification, historical addresses, geographic validation
3
Digital Footprint
Messaging app presence, digital identity verification, cross-platform validation, presence across 50+ websites/apps
4
PII / Breach / OSINT Checks
Understand fraud and Collections use cases including Crypto checks.

18