We’re Mohammad and Pranav (Moda, W26). We built monitoring & reliability for AI agents.

In production, AI agents fail silently all the time: tool calls that error or time out, agents that claim they “did it” without actually doing it, prompt injections that cause data leakage, and long customer chats that hide the real issue. Most teams only find out after users complain, churn, or something breaks.

https://youtu.be/aKPDhL_AE4E

The problem:

When an agent fails, the “bug” is rarely one line of code. It might be a flaky tool, a prompt regression, a bad retrieval result, or a jailbreak attempt.

But most teams still debug agents by skimming transcripts, grepping logs, and hoping they notice the real issue. Previous tools were built for clicks, not for long, messy conversations with skills, MCP’s and tools.

As agents get more autonomous, this gets worse. Small changes in prompts, tools, or integrations can cause big reliability and security problems at scale.

What we built:

Moda is monitoring & reliability for AI agents. It turns conversations and tool traces into low-noise, actionable signals.

Trace-first drilldowns: click from a metric or alert straight into the exact messages and tool calls that caused it.

Conversation segmentation: automatically splits long chats into topic segments so you can debug the right unit of work.

Failure detection: flags tool-call errors, timeouts, and “claimed it worked” runs.

Quality & safety signals: highlights hallucinations, user frustration, and prompt injection attempts.

Regression tracking: measure the impact of prompt and harness changes, and catch regressions before customers do.

Want to see Moda on your agent in production? Book a quick call: https://usemoda.ai

About Us:

Mohammad built AI agents and infrastructure at Shopify, Notion, and Clio, shipping systems that scaled agents to real users.

Pranav built data infrastructure at banks and insurance firms before diving into building AI agents at Shopify and Cerebral Valley

We both dropped out of University of Waterloo to solve what we believe to be the biggest problem in the next 10 years, AI systems are unreliable, Moda is going to solve that.

Visit us at https://usemoda.ai | pranav@modaflows.com