Neil Rahilly leads AI Acceleration at Sierra. He writes about deploying agents across a company, including the context, permissions and connections to existing business systems that make their work usable.[1][2]
Start with AI-pilling our company: lessons learned →
X · neilrahilly Career
Before Sierra, Rahilly held leadership roles at Mixpanel, including vice president of engineering and senior vice president of product. His experience spans engineering, product, design and support. He holds a bachelor’s degree from Duke and a joint JD/MBA from the University of Toronto.[1]
Contributions
One interface across department boundaries
Rahilly describes replacing several role-specific internal agents with Pinecone, one agent connected to existing systems of record. A request can cross sales, engineering and support without making the employee choose the right specialist first. Persistent context allows work to continue across sessions. The MCP Gateway applies employee permissions at tool calls, isolates customer data and records an audit trail.[2]
Reveal context when it becomes relevant
His context-engineering essay explains progressive disclosure: an agent receives policies, tools or knowledge when conditions make them relevant. An international-shipping conversation does not need every country’s rules at the start. Learning the destination unlocks the appropriate guidance. This reduces competing information while keeping more complex workflows available when needed.[3]
Tools, writing & talks
Writing, research & talks
Start here
An account of one internal agent spanning company systems and department boundaries.
Read the shift from multiple agents to Pinecone, then the permission gateway. Experience with workplace software helps; the article requires no Sierra setup.
Author[2]An article by Neil Rahilly about adopting AI tools across a company, framed through the lens of the 1968 study on software engineer productivity variance. It explores why AI adoption matters and how it can dramatically amplify individual and team output.
Catalog workOriginal sourceAn explanation of conditional context blocks and progressive disclosure for customer agents.
Author[3]