G2

Saurabh D.

CTo

Small-Business (50 or fewer emp.)

11/9/2026

"Actionable AI Security Findings with Fast Retests and a True Partnership"

4.5/5

What do you like best about Siemba?

Findings my engineers could act on immediately. Every finding came with evidence, steps to reproduce, and a clear fix. My team fixed most issues within days, without long back-and-forth.

They understood how modern AI products break. They covered prompt injection against our agent, isolation between tenants, role-based access, and business-logic edge cases, which a scanner would never reach. It was clear they had done this before.

Fast and responsive. Retests came back quickly, and Slack notifications meant I always knew what was happening without logging into another portal.

Practical about risk. When some behavior was intentional, like concurrent logins or rich-text support in email, they listened to our reasoning and adjusted the severity, rather than pushing fixes that would have hurt the product.

The team itself. Nithin, Ajaysen and Suraj were easy to work with from kickoff through the final walkthrough. It felt like a partnership, not a transaction. Review collected by and hosted on G2.com.

What do you dislike about Siemba?

Honestly, very little. If anything, put more time into confirming the full API inventory at kickoff, so the scope is locked in from day one. Review collected by and hosted on G2.com.

What problems is Siemba solving and how is that benefiting you?

As a young company selling to healthcare staffing firms, we hit a wall in enterprise deals: customers' security teams wanted proof of an independent penetration test before they'd sign. We didn't have an in-house security team, and our product isn't simple. It's an AI agent that works with sensitive contact data, on a platform where many customer teams share the same system. We needed someone who could test that properly, not just run a scanner.

Siemba solved that end to end. They tested the whole application as each type of user would see it, including how the AI agent behaves under manipulation, and gave us findings my engineers could fix right away. Within a few weeks we had moved from "we think we're secure" to a verified, retested result and an attestation letter. We sent the letter straight to an enterprise customer, and it unblocked the deal.

The bigger benefit is confidence. We now know our tenant isolation holds up, we closed real gaps before anyone else found them, and the team picked up habits, like rate limiting new endpoints by default, that we now use in everything we build. Security questionnaires are faster to answer too, because we have real evidence behind our answers. Review collected by and hosted on G2.com.