For three years, our team didn’t say much in public. No conference demos. No press releases. No product page. Anyone who searched “iReason voice AI” got a holding page and not much else.
Today, that changes.
We’re introducing OpenBrain – a voice AI platform that turns existing call centers into AI-supported units. It runs in production today. It handles real calls in regulated environments. It plugs into the telephony you already have, without replacing any of it.
This post is the first of a series. Over the next twelve weeks, we’ll walk through what OpenBrain actually does, why we built it the way we did, and what changes for the businesses that adopt it. Most posts will include screenshots from the live product. Two will be customer case studies. The final post will be everything you need to scope a pilot, if you want to.
But before any of that, we owe you an explanation for why we waited so long.
Voice AI in 2023 and 2024 was a graveyard of demos.
Every product launch looked great in a controlled environment. Investor decks were full of impressive numbers. Conference stages were full of bots that responded fluently in English to carefully scripted prompts. The marketing was confident.
Then the products met real call centers.
A real call center has a SIP trunk that doesn’t quite follow the spec. Customers who switch between three languages mid-sentence. Regulatory requirements that vary by country, by sector, sometimes by region. Background noise. Older phones. Network jitter. A 3 a.m. failure mode that takes down the whole switchboard. Compliance officers who need an audit trail. CX leads and domain experts who need to correct the bot’s mistakes without filing a Jira ticket.
Most demos didn’t survive that environment. The products that did survive it tended to be expensive, brittle, and locked to a specific vendor’s stack. The cost of getting them deployed often exceeded the operational savings they promised.
We spent three years working to fix that – not by claiming we’d solved it, but by building only what we could prove worked in production. We started with two real deployments in two very different industries: one in healthcare, one in telecom. Every feature we shipped, we shipped because one of them needed it. No general-purpose ambitions. No “AI for everything.” Just a voice AI platform designed to survive real-world deployments.
Today, OpenBrain handles 100% of inbound calls at our healthcare deployment. It resolves roughly 6 out of every 10 calls end-to-end without human handover. The other 4 reach a human faster, with full context already loaded. It runs on existing PBX infrastructure with zero migration cost. The same pattern holds in our telecom deployment, at significantly higher call volumes.
That’s the version of voice AI that’s worth talking about.
What we’ll cover in this series?
The next twelve posts walk through OpenBrain feature by feature, from the architecture that makes the deployment painless to the governance layer that makes it auditable. Each post is self-contained, so if you read them in order, the picture builds.
The hero post arrives in three weeks. It’s the post where we publish the architecture diagram and explain how OpenBrain plugs into existing telephony without replacing anything. If you’ve ever sat in a procurement meeting that killed a contact center modernization project on migration cost grounds, that’s the post written for you.
Between now and then, two posts frame the problem from each side of our customer base — telecom, and healthcare-support industries more broadly. The one published next week looks at why call center modernization fails in procurement. The one after that looks at the operational cost of unanswered customer calls in any service business.
After the hero post, the series shifts into specific capabilities: the visual flow designer, voice quality and brand voice, retrieval-augmented knowledge bases, outbound campaigns, governance and human-in-the-loop, supervision panels, and the deployment architecture for organizations with strict data residency requirements. Two customer case studies anchor the second half of the series.
The final post is for the technical reader – your CTO, your CISO, your DPO. By that point, the rest of the series will have answered the business questions. The last one answers the architecture questions.
What you can do today?
If voice AI in your industry has felt like it wasn’t quite ready for you, we built OpenBrain for exactly that gap. The next twelve weeks of posts are an attempt to show our work, not just the conclusions, but how we got there.
The fastest way to follow is the iReason LinkedIn page, which will mirror the blog series.
If you’d like to scope a pilot directly, the contact page is where to start. The first conversation is with one of us, not a sales rep.
Whichever path you choose, thank you for reading. Three years is a long time to stay quiet, and we’re glad to finally be talking about this.
