Back to Portfolio
Case Study

FreedomFund.ai

A fully autonomous AI fund management platform where a multi-agent AI system runs the entire company, from market research and algorithmic trading to social media, analytics, bug fixing, and daily operational reporting.

The Challenge

FreedomFund.ai needed to operate an investment fund with minimal human oversight. Traditional fund management requires large teams for research, trading execution, compliance monitoring, investor communications, and software maintenance, all expensive and slow to scale.

The Solution

We split the work six ways, one agent per domain: research, strategy testing, content, support, engineering and the nightly report. They run unattended and hand work to each other, but nothing reaches production or goes out under the company name without a person agreeing to it. Built with Borys Markiv over about two and a half years.

The AI Agent Team

Six specialized AI agents work in concert to manage every aspect of fund operations autonomously.

Research Agent

Reads the market data, the news and the filings overnight and writes one briefing in the morning. The point was never that it reads faster than a person; it is that it reads the same sources every single day without getting bored.

Trading AI Agent

Runs strategies against fifteen years of historical data and reports which ones actually held up. Most do not. The value of the thing is how quickly it kills an idea that looked good on a chart.

Social Media Agent

Drafts and schedules posts and flags mentions worth a human reply. It drafts; a person still decides what goes out under the company name.

Analytics & Support Agent

Handles the first round of support questions and watches the dashboards. When something breaks at two in the morning it files the report, so the morning starts with a list instead of a mystery.

AI Software Engineer

Triages incoming bugs, reproduces what it can and proposes the patch with tests attached. A person merges it. We were not willing to let software deploy itself to production unattended.

Daily Operations Agent

Emails one summary a night: what each agent did, what it skipped and what it got wrong. This is the agent that matters most. An autonomous system you cannot audit the next morning is not an asset.

Key Outcomes

6

Agents, one per domain

6

AI Agents Deployed

<5 min

Avg Bug Resolution

0

Manual Interventions Required

Technology Stack

Agentic AIRAG SystemsLLM OrchestrationReal-Time Market DataAlgorithmic TradingSupabaseEdge FunctionsReactTypeScriptn8n WorkflowsOpenAIAnthropic Claude

Thinking about agents doing real work?

We have built this and we run it. Happy to tell you which parts were worth it and which parts we would not do again.