Knowledge Retention & Ownership in AI based development

 

Knowledge Retention & Ownership is no longer a good to have—it's the emergency.


1. Declare a comprehension freeze, not just a metric change


Before any framework can work, leadership must stop the bleeding: no new AI-generated artifacts in the critical path (specs, architecture docs, PRDs) until engineers can explain what already exists. This is the "pull over the car" moment Vembu is describing.


Tactic: Institute a "Explain Before Merge" mandate with teeth—if the author (or the AI) can't defend it in a 5-minute review, it doesn't ship. This directly operationalizes "never cede our understanding to AI."


2. Treat the "forced to ship" mandate as the root cause


The image makes clear this isn't an emergent culture problem—it's an explicit management directive. Solutions must target leadership behavior, not engineer habits:


· Escalate with cost data: AI-generated rework, incident frequency, and attrition.

· Ask leadership a direct question: "When the system breaks, who will fix it—the AI, or the engineers you're burning out?"

· Frame comprehension as business continuity risk, not engineering preference. A company where "nobody knows anything" has no recoverable institutional memory.


3. Make AI adoption slower by design — and defend it


Vembu's warning is the strongest argument for phased rollouts (Pillar 3):


· Pause expansion of AI tooling until the current adoption is understood.

· Audit which artifacts are AI-generated and who can explain them.

· Roll out new AI capabilities only after the comprehension debt from the last wave is cleared.


4. Reframe the "fancy car" as an executive communication tool


Vembu's metaphor is the most quotable line in the image. Use it directly with leadership:


· "We are driving too fast in a fancy new car we barely understand how to drive."

· Pair it with the concrete cost: rework, incidents, attrition, and the fact that the team universally dislikes this—meaning the best engineers will leave first.


Revised priority order


Given the image, the sequence should now be:


1. Knowledge Retention & Ownership — emergency intervention (comprehension freeze + explain-before-merge)

2. Human-Centric Workflows — reposition AI as drafter, retrain engineers to critique it

3. Measured Tool Adoption — slow the rollout until comprehension debt is cleared

4. Sustainable Delivery Metrics — only viable after the first three restore basic understanding



Comments

Post a Comment

Popular posts from this blog

Airbus A320 — caused by a critical software bug

The recent trouble with the Airbus A320 — caused by a critical software bug — shows why rigorous testing is non-negotiable in complex systems. The “A320 bug” — linked to the aircraft’s ELAC (elevator/aileron control) software — was triggered by intense solar radiation that can corrupt flight-control data. As a result, dozens of international carriers were forced to ground or recall thousands of jets: it’s arguably the largest aircraft-fleet recall in aviation history. Had the faulty version of the software been subjected to more exhaustive testing — including scenarios such as solar-radiation-induced data corruption — the vulnerability might have been caught before commercial deployment. A robust test suite would have included “edge cases” (rare but plausible events) to stress-test the code under extreme conditions. The root cause: under periods of “intense solar radiation” (solar flares / charged particles), the electronic data processed by ELAC software can get corrupted — a bit-f...
Read more

Beyond Google: The Best Alternative Search Engines for Academic and Scientific Research

Google may be the most popular search engine, but it isn’t always the best for academic and scientific research. Many valuable scholarly resources are buried under commercial results or hidden behind paywalls. Fortunately, there are specialized search engines designed to help researchers, students, and professionals find high-quality, peer-reviewed content. Here’s a selection of the best alternatives to Google, focusing on science, technology, medicine, and economics. 1. RefSeek ( www.refseek.com ) Best for: General academic research. RefSeek indexes over a billion documents, including research papers, encyclopedias, and books. Unlike Google, it prioritizes educational content and filters out commercial websites. 2. WorldCat ( www.worldcat.org ) Best for: Finding books and research materials in libraries worldwide. WorldCat allows users to search for books, articles, and historical archives in over 20,000 libraries. It’s ideal for tracking down rare or specialized academic ma...
Read more

Relation between T shirt sizing, story points, hours and when to use them #sizing #agile

The T-shirt sizing (XS, S, M, L, XL, XXL) is a popular estimation technique that often gets converted to story points. Here's how they typically relate: Standard T-Shirt to Story Point Conversion Common Mapping: XS (Extra Small): 1 story point S (Small): 2-3 story points M (Medium): 5 story points L (Large): 8 story points XL (Extra Large): 13 story points XXL (Extra Extra Large): 21+ story points This often follows the Fibonacci sequence (1, 2, 3, 5, 8, 13, 21) which reflects increasing uncertainty in larger estimates. [Each number is the addition of preceding 2 numbers] T-Shirt Sizing to Hours (Rough Guidelines) Typical Team Patterns: XS Stories: 1-4 hours Simple bug fixes, minor text changes "I can do this in half a day" S Stories: 4-8 hours Small feature additions, straightforward tasks "This will take me a day or less" M Stories: 1-3 days (8-24 hours) Standard feature development "This is a typical user story" L Stories: ...
Read more