McKinsey did not commission a six month feasibility study before building Lilli, its internal generative AI platform. A small team built a lean, working prototype in one week, then walked leadership through a five minute demo of that prototype to get the investment approved for full development, according to McKinsey's own account of the project. No steering committee spent a quarter debating architecture. A working thing, built in a week, made the decision for them.
Blue Origin ran the same play on hardware, not software. Its In-Space Systems team used a multi-agent AI system, built on AWS's Strands Agents SDK and Amazon Bedrock, to take TEAREx, a lunar thermal-battery component, from concept to a 3D-printed part in days instead of years, according to AWS's case study on the project. Blue Origin calls it the first AI agent-designed hardware built for the moon.
The shift underneath both examples
Andrej Karpathy named the underlying shift in a February 2025 post on X: "vibe coding," the practice of building software by prompting tools like Cursor Composer in natural language instead of writing and studying code line by line, a term whose origin is well documented in Karpathy's original post. The same logic now reaches past code into prototypes, hardware iteration, and internal platforms. You do not need to master the underlying architecture before you get a working version in front of the person who can approve it.
The catch most operators miss
Here is where most operators get it backwards. Speed does not mean solo. MIT's NANDA initiative found that 95% of enterprise generative AI pilot programs fail to deliver measurable financial return, according to Fortune's reporting on the study. The failure mode is not slowness. It is going it alone.
Buying AI tools from a specialized vendor or building an implementation partnership succeeded roughly 67% of the time. Companies that built their AI systems entirely in-house succeeded only about a third as often, per the same MIT NANDA research reported by Fortune.
McKinsey did not build Lilli on a blank canvas assembled by a team studying transformer architecture from scratch. Blue Origin did not write its own agent orchestration layer, it built on AWS's existing SDK and Bedrock. The one week prototype and the days instead of years hardware cycle were both fast because each team borrowed proven scaffolding and pointed it at one narrow problem, rather than trying to master the full stack first.
Adoption is not the bottleneck anymore
This matters more this year, not less. McKinsey's 2025 Global AI survey found 88% of organizations now report regular AI use in at least one business function, up from 78% a year earlier. Yet nearly two thirds say they have not started scaling AI across the enterprise, according to McKinsey's survey of 1,993 respondents across 105 countries, fielded between June 25 and July 29, 2025. Everyone has tried something. Almost nobody has turned that into momentum.
This is the exact gap AskRajGPT (askrajgpt.com) was built to close. It is a seven day program built around one rule: operators leave with a working system, not a syllabus. The program does not treat a week as a magic number. It treats the deadline as a forcing function, the same one McKinsey and Blue Origin used, that keeps a team pointed at one narrow, shippable thing instead of a study of everything AI could theoretically do.
What to do this week
Pick one process inside your business that is repetitive, well documented, and painful enough that even a mediocre first version would be a win. Do not start by reading about what AI can do for your industry. Start by prompting a working prototype against that one process, using existing tools rather than infrastructure you have to build first. Get it in front of the person who can say yes or no by Friday. Whatever they say, you will know more than a quarter of reading would have taught you, and you will have something the 95% who stalled at the pilot stage still do not: a system that actually runs.
Paul Worrell is a contributor to The Founders Report. He runs Rvysion, a design and growth agency for startups.
Disclosure: The Founders Report's editor is the founder of AskRajGPT, referenced in this piece.