Bottl Labs helps founders and teams use AI to change how work gets done inside their companies.
We are a small focused team of data scientists, software engineers, and data engineers. We teamed up to take on the work that usually goes to large consulting teams.
We use AI to deliver projects 2 to 5 times faster. We cut the fluff and help your company change without the bloat.
We take a barbell approach to AI. We use AI heavily where it is most useful: coding, planning, experimentation, iteration, and testing.
We are anti AI where the risk is high. Design, security design, data access, and decisions with real consequences need human ownership. AI can support the work, but people own the judgment.
The work usually starts with one messy workflow.
It might be a reporting loop, a research process, a company knowledge base, a coding agent setup, or a support workflow.
The work has a simple shape.
We find where the company is slow. We decide whether AI should be involved. We design the workflow. We build or guide the first version. We add checks so the system is usable. People stay close to the decisions that need judgment.
Bottl is led by Arun Thiru, a former Staff Data Scientist at Adblock/Eyeo and founder of Helios, an AI company that works at the intersection of data science, horoscope readings, and spirituality.
Our background is in data science, software engineering, data engineering, forecasting, experimentation, and practical delivery. That means we care about the difference between what looks good in a dashboard and what is true.
That matters when you build with agents.
Agents need context, tools, evals, memory, and clear boundaries. Evals are checks that tell you whether the system is doing the work correctly.
If you do not design the loop, agents create noise.
If you design the loop well, agents can remove slow knowledge work.
Bottl is advisory, building, and coaching.
Sometimes we help a founder create a research agent or second brain.
Sometimes we help a team rethink BI, reporting, or internal workflows.
Sometimes we help technical teams work with coding agents without creating messy code.
Sometimes the work is just deciding what not to automate yet.
The goal is to remove slow manual work so people can spend more time making decisions.
That is the work we care about.
If you want to explore what this could look like in your company, bring one workflow that feels more manual than it should.
Reach out and let’s see if we’re a good fit.