About
Built on 25 years of
banking, regulatory,
and data work.
Dhari AI is not an AI lab. It is a consultancy built by someone who spent two decades inside the systems we now propose to automate. We know where the manual work hides — because we built the systems that created it.
I'm Manoj. I've spent the last 25+ years building the systems banks actually run on — enterprise data platforms, regulatory reporting engines, market and reference data infrastructure, risk platforms, and delivery management for some of the most complex programmes in financial services.
For the last three years, I've been working with generative and agentic AI inside that same context. What's become clear is that the gap between what an AI demo can do and what a bank can deploy in production is enormous — and it's almost entirely about enterprise plumbing, not models.
Dhari AI is the consultancy I would have wanted to hire when I was on the buy side: senior, opinionated, vendor-neutral, fluent in both the model layer and the regulatory layer, and built to deliver agents that survive an audit on day one.
Principles
How we actually work.
Useful first, novel second
AI that does not move a measurable business metric is a science project. We start every engagement with the number we are trying to move.
Agents are tools, not employees
We never frame AI as a replacement for humans. We frame it as the layer that absorbs the manual friction so humans can do the judgement-heavy work.
Audit the model, not just the output
In banking, every decision must be defensible. Our agents log their reasoning, version their prompts, and surface confidence — by default, not as an add-on.
Open weights where they fit
We mix frontier models with open-weight Llama and Mistral for data-residency-sensitive workloads. The right model for the task, not the fashionable one.
Get in touch
Ready to work together?
Most engagements start with a 30-minute call. We listen, we ask hard questions, and we tell you honestly whether we can help.