Speed is inherently impossible in the corporate structure. There are way too many incentives against it and the entire system works on exactly the opposite of speed. An extreme example of this is the Karachi Shipyard where delivery timelines for projects are in years. And in this case where nothing matters in the short term, it regresses to such an extreme that some offices in it turn off lights after lunch and take short naps on their desks. The problem is with the structure which determines the incentives. People are paid a fixed amount in the short term irrespective if the project is done quickly or slowly. And they also have to put in the same number of hours every day or have the same availability everyday which makes them less likely to do important things in bursts as they become important. Then another problem of corporates is that because they have increased in number of people beyond the ceiling where you can personally know everyone, the value of trust needs to be replaced with something objective. And because trustworthiness is hard to measure, the only thing left to do is to place tests for every task. So you end up with excessive levels of record keeping for everything. This means that there are mandatory steps that everyone needs to follow. This still works when the projects are extremely large in scope because the to and back between people is a small share of the work on the project. But this is deadly when either the scope of the project is small or the ratio of to and back in the project is very high (this happens where the project requires people with different expertise to communicate). There are a lot of checks of these kinds that affect the project's speed. One example is that decisions need to be justifiable at the moment. This works for very big projects in static markets, but in fast changing markets and innovative projects, you can only connect the dots looking backwards. If it was so simple to innovate using logical arguments and data points, all the startups in the world would've been built by Google but as we can see it was easier for OpenAI to build a startup on Google's research on transformers and Google was then just left trying to catch up. These checks are not just because of lack of trust but also because the organisation has to price in stupidity assuming that some people will be stupid or all people will be sometimes stupid. I recently had the chance to discuss this with 2 people in senior positions at a product based company with ~ 1500 employees who were talking about building and retaining exceptional teams and they also thought it was not possible with the standard structure. It was also interesting to have this conversation with people like that because my standard assumption for anyone with a lot of years in management positions is that they've been lost to the dark side regardless of how smart they are their brains are now a mush but it apparently does not apply to everyone. The only possible solution is one small independent team in the organisation that can bypass things. This is a tested format in the sense that there were teams like this in some extremely innovative organizations. One example is the CSL team in Xerox that ended up creating the Graphical User Interface for computers that then was copied by both Microsoft and Apple and which we all use. So many important innovations of today like the mouse, ethernet and laser printers can be attributed to that single team. This format is also how the US DOD is able to stay ahead of everyone in tech because they bring in outsiders and give them clearences ridiculous enough to bypass so much of the structure that even private organizations cannot compete. No wonder the internet came out of a defence organization called ARPA. If you read the book on Oppenheimer (The American Prometheus) you will see the liberties that he and his project was allowed by a military organization that was supposed to be managing it. This is also the format Palantir works on using their Forward Deployed Engineers (Reflections on Palantir)