Available for new projects

AI Automation That Delivers Results

We build intelligent systems that reduce operational costs by 40% and process thousands of tasks without human intervention.

Schedule a Call View Case Studies

50,000

Adversarial calls fuzzed

Policy violations that got through

Open

Source you can read and run

Work you can open and check, not logos

MCP Guardrail — open source agent-grader — FastAPI, 34 tests, CI green Exactly Protocol — accepted findings Tare audit, Sherlock — 2 accepted findings LilApe — jetton live on TON

What We Build

Enterprise-Grade AI Solutions

From intelligent automation to blockchain infrastructure, we deliver production-ready systems.

⚡

AI Automation

Intelligent workflows that handle customer service, document processing, and lead qualification with 99% accuracy.

🔗

API Integration

Connect your existing tools — Salesforce, HubSpot, ERPs — into a unified, automated ecosystem.

💬

Intelligent Chatbots

GPT-4 powered assistants that understand context, handle complex queries, and escalate when needed.

📊

Data Processing

Pipelines that handle 50K+ documents monthly with real-time extraction and structured output.

🔐

Blockchain Solutions

Smart contracts, token infrastructure, and Web3 integrations on TON and EVM-compatible chains.

📱

Telegram Mini Apps

Shipped end to end: LilApe runs seven games inside Telegram with a jetton deployed on TON, wallet connection through TON Connect, automated on-chain payouts and a referral system.

Claims you can check in an afternoon

Every number below comes from work that is public, runnable, or independently accepted by a third party. We are a young practice — founded November 2024 — and we would rather show the evidence than round it up.

32,981

Hostile calls thrown at MCP Guardrail — none got past policy

Findings accepted in the Tare audit on Sherlock

Tests green in CI on agent-grader, against a real PostgreSQL

Anthropic Academy certificates, each with a verification link

Violations that got through the guardrail

How We Work

From Discovery to Deployment

The same process on every engagement, from a one-week review to a six-month build.

Discovery Call

We analyze your current operations, identify automation opportunities, and provide an honest assessment of what AI can solve.

Solution Design

Custom architecture tailored to your stack. We document every integration point, data flow, and success metric.

Rapid Build

MVP in 2-3 weeks, not months. We iterate based on real usage data and your team's feedback.

Deploy & Support

Production deployment with monitoring, documentation, and ongoing optimization. We're partners, not vendors.

Hana Kim

Founder & Lead Engineer

Two years building production systems, since November 2024. Not ten. I lead with that because the rest of this page is evidence, and evidence is worth more than a number I would have to defend.

My discipline comes from smart contracts, where a mistake is priced in dollars the same day. That habit carries into AI work: an agent with tools is a program whose control flow is written by a model at runtime, so containment has to be enforced at the call site and proven under adversarial load — not promised in a prompt.

How I work: a written spec in, a working build and a written daily update out. If something I shipped is broken, you hear it from me first.

Nov 2024

Practice founded

Public repositories with CI

Verified Anthropic certificates

Let's Build Something

Tell us about your biggest operational bottleneck. We'll give you an honest assessment — no pitch, just straight answers.

Prefer email? Reach us at contact@automate-ai.live