For small orgs, I think velocity generally comes from focusing on having velocity a year and three years from now. This usually means optimizing team, tools, processes. Consider Slack and Storm as (by)products of optimizing that eclipsed the original vision. And having enough vision to skate to where the puck is. All of the above is made much harder in a typical startup, since management is forced to think of the runway, and has legitimate fires to put out 24/7. Which is why having a great COO helps.
People also talk about doing fewer things well. Sage idea, but I think the vast majority gets it backwards. They act as if doing fewer things will make them do these things well. But having giving your first idea some iteration and polish is not what well really means; I think it’s more about understanding the design space more fully than others, and thinking through the ramifications of a half dozen (or more) solutions. If one instead truly focuses on doing things well in this sense, one winds up doing fewer things.
For orgs that do many things (typically, larger, more mature orgs), attaining speed typically means pushing decision making down to fairly independent units. Berkshire, Alphabet, and Singleton’s Teledyne stand out in this respect.
Anyhow, just my thoughts — YMMV.