Merged ChatGPT attempts 2 and 3 with manual enhancements

docs/index.md

@@ -18,11 +18,11 @@
 ///
 
 /// html | div.indented_paragraphs
---8<-- "readme.md:10:49"
+--8<-- "readme.md:10:61"
 ///
 
 /// html | div.left_column
---8<-- "readme.md:50:"
+--8<-- "readme.md:62:"
 ///
 
 /// html | div.right_column

docs/readme.md

@@ -13,39 +13,52 @@ Omnipy is a type-driven Python library for:
 **Dataflows, Not Workflows**
 
 Traditional workflows rely on command-line tools and intermediate files, adding complexity to data
-pipelines. Omnipy offers dataflows that operate directly on data in memory or standard formats like
-JSON or CSV. By extending Pydantic models, it simplifies parsing, conversion, and serialization with
-structured, reusable models.
+pipelines. Omnipy offers dataflows that operate directly on data in memory, or on standard formats
+like JSON or CSV if there is a need. It expands the functionality of Pydantic data models to
+simplify parsing, conversion, and serialization of data.
 
 **"It's Static Typing!"… "It's Dynamic!"… "It's Omnipy!"**
 
 Omnipy blends Python’s dynamic typing with runtime type safety. Its models behave like native Python
-structures, ensuring type adherence while allowing flexibility. You get type guarantees without the
-rigidity of static typing, and models are as general or specific as you need.
+structures, but with a bonus: you get type guarantees without the rigidity of static typing! As
+Omnipy models are defined through normal Python code, they can be as general or specific as you
+need.
 
 **Parse, Don’t Validate**
 
-Strict validation often breaks pipelines when dealing with messy data. Inspired by the "Parse, Don’t
-Validate" principle, Omnipy dynamically transforms imperfect input into structured output, following
-the Robustness Principle: "be liberal in what you accept, and conservative in what you send."
+Strict validation often breaks pipelines when dealing with messy data. Inspired by the newer
+slogan "Parse, Don’t Validate", Omnipy data models dynamically transforms imperfect input into
+structured output that stays structured. This approach also lines up the well-known Robustness
+Principle: "be liberal in what you accept, and conservative in what you send."
 
 **Self-Constraining Data Models**
 
 Omnipy models aren’t just one-time validators—they enforce data integrity dynamically. A
-`Model[list[int]]()` behaves like a list, but every modification ensures type safety, rolling back
-to a valid state if needed.
+`Model[list[int]]()` behaves like a list, but one that only contains integers. Every modification
+parses data and ensures type safety, rolling back to a valid state if needed.
 
-**Modular, Reusable, and Scalable**
+**Omnify Your Data Pipelines**
 
-Omnipy encourages modular pipelines built from interoperable components. Dataflows, tasks, and
-models are designed for reuse and extensibility, whether in research or general analytics. With
-seamless integration with tools like Pandas and NumPy, Omnipy simplifies everything from parsing to
-advanced data wrangling.
+Omnipy invites you to “omnify” your pipelines — break them into reusable, universal components. With
+dataflows defined by structured input and output models, Omnipy simplifies code reuse, encourages
+modularity, and promotes a puzzle-piece approach to building robust and extensible pipelines. 
+Omnipy's structure-by-default approach encourages good coding practices such as "design by
+contract" and "separation of concerns", making your code more maintainable as projects grow.
+
+**Catalog of Components for Interoperability**
+
+Omnipy includes ready-to-use components for tasks like asynchronous API requests with built-in rate
+limiting, parsing JSON and tabular file formats, and flattening nested data into relational tables.
+It allows seamless integration with other tools and services, such as Pandas, making it easy to
+tackle data wrangling and interoperability challenges with minimal effort. The catalog will expand
+as the Omnipy community grows.
 
 **Built to Scale**
 
-With built-in support for Prefect, Omnipy scales effortlessly from local experimentation to
-distributed pipelines, making it perfect for projects of any size.
+Omnipy’s hierarchical `Dataset` structure simplifies batch processing. With features like batch
+parsing, serialization, and metadata handling, you can effortlessly scale from small datasets to
+large directory-based pipelines. With built-in support for Prefect, Omnipy scales effortlessly from
+local experimentation to distributed deployment, making it perfect for projects of any size.
 
 ## Generic functionality