Panayiotis Petousis, PhD
Founder · LLMs, agentic AI & healthcare AIAgentic AI · Reinforcement learning · Model deployment experience · Bayesian networks, ML, and scientific LLMs
LinkedIn ↗LLM & agentic AI consulting
Custom AI agents for work that demands accuracy.
RootSquare builds custom LLM applications and agentic AI solutions that reason across data, use tools, automate complex workflows, and keep people in control. Our expertise also includes explainable AI, Bayesian decision support, multimodal AI, and entity resolution.
Custom LLM applicationsAgentic AI
solutionsExplainable AI systems
DATARecords
CONTEXTEvidence
MODELSReasoning
Clear
decision
decision
10+years building applied AI systems
AgenticLLM workflows that use data and tools
Human-ledAI systems designed for oversight
LLM, agentic & custom AI solutions
AI systems that reason, act, and remain accountable.
We create LLM applications and AI agents around specific decisions and workflows, combining trusted data, tool use, clear explanations, safeguards, and human review.
RISKEXPLAINACT
CALTECH LONGEVITY STARTUP LAUNCHPAD 2026
ILAMEE: personalized heart-failure evidence
An AI agent that estimates four-year survival, explains its most important drivers, and returns validated literature findings on factors associated with lower or higher risk.
View ILAMEE project →
MULTIMODAL AI SYSTEMS
AI photo matching for the Sophia Foundation
A custom computer-vision and record-matching workflow with human verification that reduced manual work from days to minutes.
View multimodal AI project →
ENTITY RESOLUTION & DATA DEDUPLICATION
Large-scale duplicate record detection
An entity-resolution system for identifying duplicate people and records across noisy datasets while distinguishing close family relationships.
View data deduplication project →How we work
A clear process from problem to deployment.
We follow CRISP-DM—the Cross-Industry Standard Process for Data Mining—and adapt it for modern AI, LLM, and agentic systems.
Tell us what you are working on →- 01
Business understanding
Define the problem, desired outcome, users, and measures of success.
- 02
Data understanding
Explore the available data, its quality, and what it can support.
- 03
Data preparation
Clean, connect, and structure the data for reliable use.
- 04
Modeling
Build and refine the most appropriate AI or data solution.
- 05
Evaluation
Test performance, safety, explainability, and business value.
- 06
Deployment
Integrate the solution into a practical, monitored workflow.
AI specialists
Technical expertise in LLMs, AI agents, and high-stakes AI.
RootSquare combines research-grade AI knowledge with practical experience designing agentic, explainable systems for clinical, operational, and public-data workflows.
Andrea Loizidou, MS
Data Scientist · Entity resolution & NLPData deduplication and entity resolution · Natural language processing and LLM systems · Applied machine learning and data analysis
LinkedIn ↗
Panagiota Loizidou, PhD candidate
Applied Research ScientistPipeline engineering for noisy, high-dimensional data · Hypothesis-driven experimental design · Long-horizon project management
LinkedIn ↗LLM & agentic AI consulting
Need a custom LLM application or AI agent?
Talk with RootSquare about agentic workflow automation, retrieval-augmented generation, tool-using AI agents, Bayesian decision support, multimodal AI, or entity resolution.