Dowellabs: AI agent workspace
Agents that do the operational work.
Dowellabs is building an AI agent workspace: multiple agents, each working in a human-like browser, taking on product planning, research, development, QA, marketing, and sales automation. Every consequential action waits for human approval. The agents propose; you decide.
01The workspace
-
One agent, one browser
Each agent operates a full browser environment: reading pages, filling forms, checking dashboards, gathering data. The work happens where your business already happens, inside the tools you use.
-
Approval gates
Agents prepare and propose; nothing consequential runs without a human saying yes. The line between suggestion and action is explicit, and every decision is traceable.
-
Many agents, one place
Research, development, QA, outreach: separate agents with separate jobs, coordinated in a single workspace instead of scattered across tabs and chat threads.
02Why it exists
Most operational work lives between the tools.
Businesses do not run inside a single app. They run in the gaps: checking one dashboard against another, copying figures into a sheet, following up on what someone forgot. Too small to hire for, too irregular for rigid automation.
That is the work agents with browsers are for: the messy middle that never got software.
03Where it stands
In development, funded by real client work.
Dowellabs is one person: Rahat Murshed, Sylhet. The workspace is being built now, and it is funded the honest way, by building custom systems for education businesses, where the agent ideas get tested against real operational pain.
Scholars Hunt is an English competition in Sylhet. Its platform, registration, three payment paths tracked in one system, and admit cards generated automatically, was built end to end and is run by the organizers' own team.
Talk to me.
If you are thinking about operational agents for your business, or you run an education business buried in manual work, I would like to hear what that looks like.
rahat@dowellabs.com