About Us
Who We Are
Devlar Technologies is an applied AI research company.
Founded by Alan O'Meara, a Senior Engineering Manager with 9 years at Adobe, Devlar started out building AI products. Building them made one thing clear: the hardest problems were not product problems. They were open research questions about how AI systems fail, remember and spend compute, and how people come to trust them. Devlar now works on those questions directly.
Before founding Devlar, Alan led the team that built a production AI agent deployed inside Adobe that automated Slack war room summarization and Jira ticket updates during live incidents. That experience made the potential of autonomous agents clear, and also showed where they break: silent failures, lost context, unpredictable cost, and graceful handling of failure.
Devlar's research covers four areas: agent reliability, memory and context, token efficiency, and human-AI interaction. Each one is tested against a working system rather than a benchmark alone, because production is where these problems show up.
Core Values
These principles guide what we research, how we test it, and how we report it.
Rigour
We test ideas on working systems under real workloads, not only in demos. Our standards come from maintaining large-scale systems where failure has real consequences.
Honesty
We say plainly what our research has and hasn't shown, and where each project stands. We would rather report a negative result than overclaim a positive one.
Openness
We believe transparency in how AI systems work is essential for trust, and that research is more useful to the wider AI engineering community when it is shared.
Connect With Us
Whether you're a researcher, an engineer facing the same problems, a potential collaborator, or an investor — we'd like to hear from you.
Get in Touch