About
Built to accelerate your adoption of AI agents.
AvonAI is a platform for AI agents in production that finds issues, fixes them and continuously validates every agent, for banks, insurers, fintechs and other regulated companies that want to scale their agents with confidence.
Today you have one agent. Tomorrow you have five. Problems escalate and change over time. AvonAI reviews every conversation your agents have, whether built in-house or by a vendor, and gives your business team one place to oversee them.
Every conversation is checked against your rules. More than 20 built-in checks catch hallucinations, loops, pushback and policy breaks, ranked by severity and frequency. Your team adds its own rules in plain language, such as a crypto disclaimer on every mention.
Each issue is traced to its cause: the data, the prompt, a guardrail or a tool. Issues are grouped and ranked, so you fix what matters most first, and you tell the platform how the agent should handle it next time.
Every issue becomes a test, built from real conversations and written in plain language. Any update to the prompt, model, knowledge base or vendor agent runs the test set, so you catch problems before your customers do.
An agent vendor's own tools see only that vendor's agents. AvonAI covers every agent, chat or voice, built in-house or by any vendor, and judges all of them by one standard.
Compliance, product, operations and customer-experience teams run AvonAI and write their own rules in plain language, without writing code. Each issue arrives with its cause, so the right team fixes the right thing.
Monitoring and compliance tools flag that something went wrong. AvonAI also traces each issue to its cause, lets your team fix it in the platform, and turns it into a test that runs on every change.
AvonAI runs in the cloud or fully on-prem, wherever your security policy requires. Your data only goes to models you've approved.
AvonAI is built for banks, insurers, fintechs and health providers, where a wrong disclaimer or an unauthorized promise is a regulatory issue, not a bad review.
Companies running AI agents in production that want to scale them with confidence.
How leading regulated companies use AvonAI today.
Nasdaq-listed fintech
Monitors a vendor-built customer agent as adoption grows.
Israel's largest card company
Trains and monitors its card-benefits agent.
Online rental marketplace
Thousands of daily conversations, run by AI quality managers.
Embedded lending
Monitors agents in lending workflows where accuracy is required.
Digital bank
Oversight and testing across its customer-facing agents.
Security and identity
Oversight and testing for agents in security-sensitive conversations.
“We fed it 75 conversations we already knew were broken. It caught almost all of them, including problems our own internal judge had been missing entirely.”
“In about thirty or forty minutes, we found things we'd have had close to no way of ever spotting otherwise.”
“I didn't know the scale or the severity until we saw it. When I saw it, I almost took our bot down.”
“We really did have an issue, resolved within a few hours. Glad you caught it before we did.”
AvonAI was built by leaders who deployed AI agents in production in regulated environments. They learned first-hand that launching an agent is only step one, and built the platform they needed to keep agents right as they scale.
Co-founder and CEO
Chen was Head of AI at Dream Security, and before that worked at Pagaya (Nasdaq: PGY). He deployed generative AI agents in complex on-prem cyber cases, and is an alum of Talpiot, the Israel Defense Forces' elite technology program.
Co-founder and CTO
Amit was Head of Data and AI at ONE ZERO Digital Bank, where he spearheaded Ella, the world's first AI banker, and before that worked at Pagaya (Nasdaq: PGY). He is an alum of Unit 8200, the Israel Defense Forces' intelligence technology unit.
Five steps, split between your team and ours. Once they're done, the platform runs on your conversations and your team knows how to use it.
A fully managed version for teams running agents in production who want to start fast, with Avon hosting the platform and working alongside them from day one.
For banks, insurers and other regulated companies: the full platform runs inside your own environment, on an LLM you approve, so your data never leaves your infrastructure.
AvonAI is a platform built to accelerate your adoption of AI agents. It reviews every conversation your agents have, whether built in-house or by a vendor, and gives your business team one place to find issues, fix them and continuously validate every agent.
Compliance, product, operations and customer-experience teams. Rules and tests are written in plain language, without writing code.
Wherever your security policy requires: in the cloud or fully on-prem. Your data only goes to models you've approved.
Most teams are up and running within a week. The five setup steps are split between your team and ours, and end with a joint calibration session.
Yes. AvonAI works with chat and voice agents built in-house or by any vendor, and judges all of them by one standard in one place.
See what AvonAI can do for your team.
Book a Demo
