Sashank Thupukari builds Indent, a shared AI coworker for a company. His work focuses on an agent that uses organizational context across Slack, code and connected tools while keeping track of the work it performs.[1][2][3]
Start with Indent →
X · sthupukari Career
His professional site records earlier software-engineering work at Ramp and Dagster, along with work at Scale.ai, Facebook, Atrium, Gifs.com and Phone2Action. He studied computer science at the University of Maryland, College Park. That background connects product engineering with the current problem of making one agent operate across an organization’s software.[3]
Contributions
Indent accepts requests through Slack and GitHub rather than requiring every employee to set up a separate agent. The product describes retaining shared instructions and context learned through work. A task can involve diagnosing an issue, editing code and opening a pull request, while its session shows the steps and artifacts involved.[2]
Ideas
Thupukari’s public writing emphasizes the distinction between personal and organizational integrations and between private and shared context. An agent serving several employees needs boundaries around what each task can access. This is part of the surrounding harness design, not something that follows automatically from giving a model a larger context window.[1]
Tools, writing & talks
Tools
Start here
Indentindent.comA shared company agent using Slack, GitHub and connected organizational tools.
Follow the illustrated Slack-to-pull-request task and inspect the visible steps. No coding is needed to understand the example; using the product requires connecting a workspace.
Builder[1][2]Original source