Harmony Intelligence

Harmony Intelligence develops AI agents that continuously perform penetration testing on customer software, with the agents managed and monitored by human security experts. The product covers web applications, APIs, MCPs and Chrome extensions in any programming language or framework. Onboarding requires the customer to grant code access and point the agents at a staging server, along with business context about what matters most to the organisation.

The audit combines static analysis with dynamic testing across the MITRE Top 25 categories, including authentication and authorisation flaws, injection, business logic issues and SSRF. Findings are prioritised and validated, and delivered with business impact assessments, reproducible proof-of-concept exploit scripts, remediation guidance and code patches; reporting is designed to be flexible enough for compliance needs. Audits run continuously through the year rather than as one-off annual engagements. The company states it does not train on customer data.

Earlier in its history the company positioned itself as a safety evaluator for AI labs and governments, contributing to testing of frontier models in areas such as autonomy and cybersecurity, including work supporting the International Network of AI Safety Institutes and a briefing of the Australian Senate on emerging AI risks. It has since focused on building defensive cybersecurity technology, principally AI-driven automated penetration testing.

Founding story

Harmony Intelligence was founded by Soroush Pour (CEO) and Alex Browne (CTO), who first met at Duke University and later worked together at Plaid, where they were among the first 20 employees. Pour established Plaid's first cybersecurity team, defending against credential stuffing attempts and network breach attempts, and subsequently served as Head of Technology at biotech startup Vow, leading a 35-person product R&D team. Browne was a tech lead at 0x Labs and Edge Analytics, working on security-critical blockchain systems and ML systems for customers including Google X. The founders describe building the company because AI is accelerating both the volume of code shipped and the speed at which attackers exploit vulnerabilities, while practices such as annual penetration tests have not kept pace.

Business model

Harmony sells a fully managed service: customers grant access to code and a staging environment, and Harmony's AI agents, supervised by the company's security experts, run continuous audits and deliver validated findings, POC scripts, remediation guidance and code patches. The company positions its pricing as affordable and predictable relative to traditional human penetration testing engagements, which it characterises as expensive and annual.

The website contrasts Harmony's pricing as "affordable, predictable" against expensive human pentests and hard-to-predict AI tools, indicating a service subscription for continuous audits rather than one-off engagements; specific prices are not published.

Traction

The website publishes customer testimonials from the CEO of Town and the VP of Engineering at 0x, and a customer story on how 0x secures $200B with Harmony's AI AppSec agents. The company describes itself as rapidly growing and remote-first, with team members across San Francisco, Sydney, Melbourne and Tokyo, and reports raising USD $3 million in seed funding in March 2025.

Latest developments

Recent published work includes a customer story on 0x's use of Harmony's AI AppSec agents and security research on denial-of-service against Next.js servers. The careers page describes an offsite in Vancouver and anticipates future openings across engineering, research, sales, marketing, operations and people functions.