SALAH HASANAIN
AI Founder · CEO @ Repixa
I build systems that move products, and markets - forward.
Founder & CEO of Repixa. Currently focused on agentic AI, real-time product control, and systems that operate inside live businesses.
About
I'm interested in systems that scale human effort, not incrementally, but structurally.
Most software automates tasks. I'm focused on automating outcomes.
That interest has led me to work on agentic AI, real-time control systems, and products that operate inside other products.
If you're a founder, builder, or researcher working on similar problems, I'd love to connect.
Repixa
Autonomous AI That Runs Live Product Demos
SaaS demos are slow, founder-dependent, and inconsistent, which kills conversion and momentum. Repixa is an autonomous AI system that controls the product in real time during live calls.
It navigates the interface, executes workflows, and answers questions dynamically based on user intent. By turning demos into interactive, adaptive experiences, Repixa increases conversion rates, accelerates sales cycles, and reduces operational cost and founder dependency, all through real-time computer control and structured decision loops.
Research & Experiments
Technical exploration and learning — from agent planning to voice and RAG.
Adversarial LLM Evaluation System
A multi-model reasoning framework that reduces hallucinated claims through structured disagreement and hybrid judging logic.
RAG evaluation experiments
Benchmarking retrieval quality and answer faithfulness for product-knowledge RAG systems.
Algorithmic Exposure Simulation Lab
A synthetic audience engine that models how content distribution emerges from interaction dynamics rather than direct prediction.
Projects
Serious engineering work — from product to experiments.
Scientific Claim–Citation Verification Engine
A research-focused system that traces scientific claims back to their cited sources to evaluate alignment between assertion and evidence. It extracts explicit claims, maps them to references, and retrieves relevant source context for comparison. The engine classifies support levels (e.g., supported, misaligned, contradicted) with uncertainty-aware scoring. Designed as a verification assistant rather than a credibility judge, it prioritizes interpretability over black-box prediction. The project explores how AI can strengthen rigor in high-density citation environments such as AI research.
Salah AI
An AI version of my thinking and experience. Ask about AI, startups, or building products.
Ask Salah about AI, startups, or building products.
Contact & Connect
Founders, builders, and researchers — I'd love to hear from you. Reach out for collaboration, feedback, or just to say hi.