Agentic AI systems that think, plan, and act — automating complex workflows so your team can focus on what matters most.
I specialise in designing and shipping agentic AI systems — software that doesn't just answer questions, but actively reasons through problems, calls tools, and completes multi-step tasks with minimal human intervention.
My builds leverage cutting-edge frameworks like LangGraph, Google Gemini, and LangChain to create production-ready agents with persistent memory, real-time state inspection, and Human-in-the-Loop safeguards.
Production-quality agents demonstrating real-world agentic patterns — from autonomous triage to multi-step research pipelines.
A fully autonomous customer support agent powered by a LangGraph DAG state machine. It categorises incoming queries, dynamically calls the right tools (order DB lookups, RAG knowledge base), and either resolves issues or escalates to a human with a checkpoint interrupt — all in real-time with a live state inspector.
Give it a topic — it does the research. This agent autonomously searches the web, synthesises information from multiple sources, and produces a structured, cited report. Built with an iterative refinement loop that critiques its own output before finalising.
An end-to-end autonomous content engine. Feed it raw data sources — RSS feeds, Google Trends, API data — and it automatically generates tailored social posts, blog drafts, and newsletters, then schedules them across platforms. Zero manual effort.
A Natural Language-to-SQL Analytics agent built with a custom Seaside Cake aesthetic. It safely translates plain English questions into read-only SQLite queries, executes them against an enterprise mock database, and synthesizes a strict under-50-word business summary.
Unlike simple chatbots, agentic systems can plan ahead, use tools, and self-correct — ideal for automating complex, multi-step business processes.
The agent receives a task — from a user, a webhook, or a scheduled trigger — and interprets the intent using an LLM.
It breaks the task into sub-goals and decides which tools to call — APIs, databases, search engines, or other agents.
The agent executes actions, reads results, and updates its internal state graph — iterating until the task is complete.
The final output is delivered — or escalated to a human if the agent determines it needs oversight (HitL).
I build tailored agentic AI systems that integrate with your existing tools and automate your most time-consuming workflows.
Let's TalkHave a complex workflow begging to be automated? An idea for an AI-powered product? Let's talk.