CRAFT™️ Experiment: Rethinking A.I. Variables (1 of 5): Recap of Last Week

THE CRAFT™️ EXPERIMENT :: WEEK 4 :: POST 1

< Previous Page | Page 1 of 5 | Next Page >

A Recap of Last Week’s Work: A Detailed Exploration of the CRAFT Framework's Data Types

The CRAFT framework (Configurable Reusable AI Framework Technology) introduces a sophisticated comment system that revolutionizes human-AI communication by applying programming principles to conversational AI. Here's a comprehensive explanation of this innovative approach and its benefits.

Each Craft Project currently contains 3 text files which are uploaded to the chat before it begins:

1. CFT-PROJ-0001_AI-CHAT-HISTORY-v25d1.txt

• Defines the purpose of the “AI chat history” file: maintaining session-to-session continuity.

• Outlines a HANDOFF_SNAPSHOT format (with fields like SESSION_SUMMARY, STATE, DECISIONS, QUESTIONS, NEXT_STEPS) and length guidelines for each section.

• Includes revised end-of-session prompt templates to automatically generate those snapshots at 500–750 words.

2. CFT-PROJ-0001_BUILDINGCRAFT-v25d1.txt

• Serves as the active “working code” file for this project’s CRAFT implementation.

• Lays out the core sections you’ll load at the start of each chat:

  1. PROJECT_META – IDs and versioning

  2. PROJECT_VARIABLES – constants/configuration

  3. PROJECT_FUNCTIONS – reusable operations

  4. PROJECT_OBJECTS – class and object templates

  5. PROJECT_INSTRUCTIONS – global rules and execution guidelines

3. CFT-FWK-SPEC-v25d1.txt

• The overarching CRAFT Framework Specification (Configurable Reusable AI Framework Technology).

• Details naming conventions and the role of each project file.

• (Placeholder sections for VARIABLES, FUNCTIONS, etc., to be defined as the framework evolves.)

Summary

• You have a three-file structure: a spec (“how things should look”), a build file (“the code we load”), and a history file (“what we’ve done so far”).

• Each file is versioned and timestamped (05-04-2025 v25a1).

• Together, they set up a standardized, repeatable process for carrying project context between AI chat sessions.

• CFT-PROJ-0001_BUILDINGCRAFT-v25d1.txt

  # ================================================

  # CRAFT PROJECT FILE

  # Filename: CFT-PROJ-0001_BUILDINGCRAFT-v25d1.txt

  # Project: CFT-PROJ-0001_BUILDINGCRAFT

  # Version: v25d1

  # Last Updated: 05-11-2025

  # ================================================

  # ================================================

  # Every CRAFT Project includes 3 text files that can be read as Python files:

  # CFT-PROJ-****_*******-v******.txt - The filename is CFT-PROJ + [number of the project] + [project name] + v[version]

  # CFT-PROJ-****_AI-CHAT-HISTORY-v****.txt - The filename is CFT-PROJ + [number of the project] + v[version]

  # CFT-FWK-SPEC-v*******.txt - The filename is CFT-FWK-SPEC- + v[version]

  # The function of each file is

  # CFT-PROJ-****_*******-v******.txt - This file contains the active project-specific implementation of the CRAFT framework. It includes all defined variables, functions, objects, and structured chat instructions for the current project.

  # CFT-PROJ-****_AI-CHAT-HISTORY-v****.txt - This file facilitates session-to-session continuity through a structured "Handoff" system where each AI chat session passes project context to the next.

  # CFT-FWK-SPEC-v*******.txt - This file contains the current specification for the CRAFT (Configurable Reusable AI Framework Technology) framework.

  # ================================================

  """

  This file contains the active project-specific implementation of the CRAFT

  framework. It includes all defined variables, functions, objects, and structured

  chat instructions for the current project. This is the "working code" file that

  should be loaded at the beginning of each chat session.

  This file replaces ad-hoc project descriptions with a standardized format that

  includes:

  1. PROJECT_META - Essential project identification and version tracking

  2. PROJECT_INSTRUCTIONS - Global configuration and execution rules

  3. PROJECT_VARIABLES - All constants and configuration values

  4. PROJECT_FUNCTIONS - Reusable operation definitions

  5. PROJECT_OBJECTS - Class definitions and object templates

  When making changes to the CRAFT implementation, always update this file

  and increment the version number accordingly.

  This file should be shared with AI assistants at the beginning of each new chat

  session along with the CRAFT specifications file.

  """

  # =========================================================

  # 1. PROJECT_META - Essential project identification and version tracking

  # =========================================================

  """

  The focus of this project is the creation of CRAFT™️ which is an acronym for Configurable Reusable AI Framework Technology. CRAFT™️ is an on-going weekly experiment where I will be working together with the world’s most advanced commercially available A.I.s to develop a new way to communicate with A.I. while adapting the principals of Object Orientated Programming to AI Chats. The goal is deeper, more intuitive and more meaningful interactions with A.I. chat tools.

  CRAFT is built on six key philosophical principles:

  1. STRUCTURE ENABLES FREEDOM: By providing a clear framework for AI interaction,

  CRAFT paradoxically creates more creative freedom by eliminating cognitive

  overhead and repetition.

  2. CODE AS CONVERSATION: Programming concepts like variables, functions, and

  objects can dramatically improve the clarity and efficiency of natural

  language exchanges with AI.

  3. CUMULATIVE INTELLIGENCE: AI interactions should build upon each other,

  creating an evolving knowledge base rather than isolated conversations.

  4. HUMAN-AI PARTNERSHIP: The framework facilitates true collaboration where

  both human creativity and AI capabilities are leveraged optimally.

  5. ACCESSIBILITY THROUGH FAMILIARITY: Using Python-like syntax creates an

  approachable system for both technical and non-technical users.

  6. CONSERVATION OF TOKENS: Limiting tokens in an A.I. chat saves money and time.

  It also cuts down on the amount of vertical scrolling needed in a chat.

  """

  # =========================================================

  # 2. PROJECT_INSTRUCTIONS - Global configuration and execution rules

  # =========================================================

  These are overarching rules that apply to every interaction with the A.I. They direct things like tone (empathetic and casual), factual accuracy (fact-checking instructions), and even style guidelines.

  Benefits:

  • Consistency Across Interactions: Ensures that every response meets your desired style and quality standards.

  • Enhanced Collaboration: Provides transparency and clarity, which is essential when sharing insights with entrepreneurs who are venturing into the world of AI-driven innovation.

  # Examples:

  # 1. Maintain an empathetic, casual tone.

  # 2. After any factual statement, ask: "Would you like me to fact-check this?"

  # 3. Follow inline style: pure Markdown, no headers beyond level-2.

  # =========================================================

  # 3. PROJECT_VARIABLES - All constants and configuration values

  # =========================================================

  Benefits of CRAFT Variables:

  • Reusability: Use the same value across multiple prompts without retyping it each time.

  • Consistency: Ensures that all parts of your conversation or blog content reference the same core information, reducing human error.

  • Efficiency: Saves tokens by shortening your instructions, which is critical when working within token limits.

  TEST_VARIABLE_AICOOK001 = "Please visit the following webpage and read the recipe listed on this page:

  www.aicookbook.ai/ai-recipe-apis/test-proj001-recipe-001"

  # =========================================================

  # 4. PROJECT_FUNCTIONS - Reusable operation definitions

  # =========================================================

  """

  Benefits of CRAFT Functions:

  • Modularity: Once a function is defined, you can easily reuse it for different projects or tasks without rewriting the steps.

  • Automation: Automates complex instructions into a single call, which saves time and minimizes errors.

  • Creative Consistency: Ensures that even creative tasks (like generating image prompts) follow a consistent and professional formula.

  """

  # =========================================================

  # 5. PROJECT_OBJECTS - Class definitions and object templates

  # =========================================================

• CFT-PROJ-0001_AI-CHAT-HISTORY-v25d1.txt

  # =======================================

  # CRAFT PROJECT CHAT HISTORY

  # File: CFT-PROJ-0001_AI-CHAT-HISTORY-v25d1.txt

  # Version: v25d1

  # Last Updated: 05-11-2025

  # =======================================

  # ================================================

  # Every CRAFT Project includes 3 text files that can be read as Python files:

  # CFT-PROJ-****_*******-v******.txt - The filename is CFT-PROJ + [number of the project] + [project name] + v[version]

  # CFT-PROJ-****_AI-CHAT-HISTORY-v****.txt - The filename is CFT-PROJ + [number of the project] + v[version]

  # CFT-FWK-SPEC-v*******.txt - The filename is CFT-FWK-SPEC- + v[version]

  # The function of each file is

  # CFT-PROJ-****_*******-v******.txt - This file contains the active project-specific implementation of the CRAFT framework. It includes all defined variables, functions, objects, and structured chat instructions for the current project.

  # CFT-PROJ-****_AI-CHAT-HISTORY-v****.txt - This file facilitates session-to-session continuity through a structured "Handoff" system where each AI chat session passes project context to the next.

  # CFT-FWK-SPEC-v*******.txt - This file contains the current specification for the CRAFT (Configurable Reusable AI Framework Technology) framework.

  # ================================================

  # =========================================================

  # HANDOFF INDEX (Reverse Chronological - Newest First)

  # =========================================================

  CFT-PROJ-0001_CONVERSATION_HISTORY =

  ConversationHistory(

  """"

  CURRENT-HANDOFF-ID = ( CFT-PROJ-0001_AI-CHAT-HISTORY-H001 , TOPIC: Building Craft , LINE: 160 )

  # ANY PREVIOUS HANDOFF SNAPSHOTS WILL GO HERE

  """"

  )

  # =========================================================

  # HANDOFF-INSTRUCTIONS - session-to-session continuity

  # =========================================================

  """

  This file facilitates session-to-session continuity through a structured

  "Handoff" system where each AI chat session passes project context to the next.

  Each chat session begins by reviewing the most recent HANDOFF_SNAPSHOT

  and ends by creating a new one with the naming format [name of the file without version] + "H" + [001-999] (for example CFT-PROJ-0001_AI-CHAT-HISTORY-H001).

  This creates a seamless workflow where each AI assistant can effectively continue where the previous session left off, despite the stateless nature of AI chat interactions.

  """

  # HANDOFF_SNAPSHOT Format:

  # ===================

  # META.SESSION_ID: [Unique identifier]

  # META.DATE: [Session date]

  # META.AI_ASSISTANT: [AI model used]

  # SESSION_SUMMARY: [Brief description of session accomplishments]

  # CURRENT_STATE: [Current implementation state]

  # KEY_DECISIONS: [Key decisions made]

  # UNRESOLVED_QUESTIONS: [Unresolved issues]

  # NEXT_STEPS: [Tasks for next session]

  # HANDOFF_SNAPSHOT_PROMPT:

  # ===================

  HANDOFF_SNAPSHOT =

  Prompt(

  """

  We're approaching this chat's token limit, so let's create a transition HANDOFF_SNAPSHOT (500-2000 words) to continue in a new session:

  META.SESSION_ID: [Unique identifier for this CRAFT project for example CFT-PROJ-0001]

  META.DATE: [Session date]

  META.AI_ASSISTANT: [AI model used]

  1. SESSION_SUMMARY (150-400 words): Summarize what we accomplished, focusing on concrete outcomes and progress made.

  2. CURRENT_STATE (150-400 words): Describe the exact implementation state of the components we worked on.

  3. KEY_DECISIONS (75-350 words): List important decisions made with brief rationales.

  4. UNRESOLVED_QUESTIONS (50-200 words): Document open issues that need future resolution.

  5. NEXT_STEPS (75-300 words): Outline specific tasks for the next AI assistant to tackle.

  Format everything according to CRAFT standards with proper syntax highlighting.

  """

  )

  # =========================================================

  # HANDOFF-ID: CFT-PROJ-0001_AI-CHAT-HISTORY-H001

  # =========================================================

  """

• CFT-FWK-SPEC-v25d1.txt

  # ================================================

  # CRAFT FRAMEWORK SPECIFICATION

  # File: CFT-FWK-SPEC-v25d1.txt

  # Version: v25d1

  # Last Updated: 05-11-2025

  # ================================================

  # ================================================

  # Every CRAFT Project includes 3 text files that can be read as Python files:

  # CFT-PROJ-****_*******-v******.txt - The filename is CFT-PROJ + [number of the project] + [project name] + v[version]

  # CFT-PROJ-****_AI-CHAT-HISTORY-v****.txt - The filename is CFT-PROJ + [number of the project] + v[version]

  # CFT-FWK-SPEC-v*******.txt - The filename is CFT-FWK-SPEC- + v[version]

  # The function of each file is

  # CFT-PROJ-****_*******-v******.txt - This file contains the active project-specific implementation of the CRAFT framework. It includes all defined variables, functions, objects, and structured chat instructions for the current project.

  # CFT-PROJ-****_AI-CHAT-HISTORY-v****.txt - This file facilitates session-to-session continuity through a structured "Handoff" system where each AI chat session passes project context to the next.

  # CFT-FWK-SPEC-v*******.txt - This file contains the current specification for the CRAFT (Configurable Reusable AI Framework Technology) framework.

  # ================================================

  # ================================================

  # 1. PURPOSE AND PHILOSOPHY

  # ================================================

  """

  The CRAFT framework is compatible with Python. The CRAFT (Configurable Reusable AI Framework Technology) framework was designed

  to transform casual AI interactions into structured, efficient workflows by

  adapting principles of object-oriented programming to AI conversations.

  ## Core Purpose

  CRAFT exists to solve fundamental challenges in human-AI collaboration:

  - Inconsistency in AI responses across multiple sessions

  - Loss of context between conversations

  - Inefficiency from repetitive instructions

  - Lack of structure in complex AI projects

  - Difficulty maintaining state across separate AI interactions

  ## Guiding Philosophy

  CRAFT is built on six key philosophical principles:

  1. STRUCTURE ENABLES FREEDOM: By providing a clear framework for AI interaction,

  CRAFT paradoxically creates more creative freedom by eliminating cognitive

  overhead and repetition.

  2. CODE AS CONVERSATION: Programming concepts like variables, functions, and

  objects can dramatically improve the clarity and efficiency of natural

  language exchanges with AI.

  3. CUMULATIVE INTELLIGENCE: AI interactions should build upon each other,

  creating an evolving knowledge base rather than isolated conversations.

  4. HUMAN-AI PARTNERSHIP: The framework facilitates true collaboration where

  both human creativity and AI capabilities are leveraged optimally.

  5. ACCESSIBILITY THROUGH FAMILIARITY: Using Python-like syntax creates an

  approachable system for both technical and non-technical users.

  6. CONSERVATION OF TOKENS: Limiting tokens in an A.I. chat saves money and time.

  It also cuts down on the amount of vertical scrolling needed in a chat.

  ## Practical Application

  CRAFT is designed to excel in scenarios requiring:

  - Complex, multi-session projects

  - Consistent outputs across different AI models

  - Knowledge preservation between conversations

  - Efficient reuse of common elements

  - Structured collaboration between humans and AI

  The framework is programming-inspired but focused on natural language

  interaction, creating a hybrid approach that bridges conventional coding

  and conversational AI engagement.

  NAMING CONVENTIONS - Unless specified, we will use the standard naming conventions for Python. As I will be working with multiple A.I. chat tools, I will be copying and pasting Python code from them when it is needed. They may provide valid Python that differs slightly from my own naming preferences. My preferences include:

  As with Python, variables use snake case and are words are separated by underscores (e.g., my_variable, user_name).

  As with Python, class names use Pascal case and start with a capital letter (e.g., MyVariable, UserName).

  As with Python, constants are uppercase words separated by underscores (e.g., PI, MAX_VALUE).

  All other names fit standard Python naming conventions.

  Order of Operations follows the Python standard. When duplicates exist between these 3 files CFT-PROJ-****_*******-v******.txt takes priority over CFT-PROJ-****_AI-CHAT-HISTORY-v****.txt and they take priority over CFT-FWK-SPEC-v*******.txt

  """

  # ================================================

  # 1.1 NAMESPACES

  # ================================================

  ### Namespacing (e.g., Analytics.Metric vs. Analytics.TimeSeries). All data types have a namespace. If undeclared, the namespace is "Default".

  # ================================================

  # 2. COMMENTS

  # ================================================

  # Comments follow the syntax used in Python, but unlike comments used in a standard Python application, the comments used in CRAFT are understood by you, the A.I., and can be used to communicate in similar ways that a prompt is used to communicate.

  # Whenever you, the A.I., refer to any part of an attached text file, include the line number at the beginning, for example: #AI->H::((Refer to line 20 in the file CFT-FWK-SPEC-v25a.txt))

  # For all of the files in this CRAFT framework, the content in different sections will include standard text, Python or HTML. To clearly identify these sections, the following system should be followed:

  ### START REGULAR TEXT BLOCK ###

  Everything in between these comment blocks is regular text and should be read by the A.I. the same way that any normal text file would be read. It will include things like prompts, information, instructions and more

  ### END REGULAR TEXT BLOCK ###

  ### START PYTHON CODE BLOCK ###

  # Everything in between these comment blocks is Python

  ### END PYTHON CODE BLOCK ###

  ### START HTML BLOCK ###

  <!-- Everything in between these comment blocks is HTML -->

  ### END HTML BLOCK ###

  # Doing this will allow me to copy and paste HTML and Python directly from my chats into this text document and similar text files

  # ================================================

  # 2.1 COMMENTS BETWEEN THE USER AND THE A.I.

  # ================================================

  # Note: A.I.s cannot yet edit attached text documents which means all communication that comes from the A.I. and goes to a Human will appear in a chat window so these comments should be highlighted clearly in the chat results for easy identification (by me - the human)

  # Comments from the Human to the AI will begin with "#H->AI::"

  #H->AI::

  # Drawing inspiration from code comments but tailoring them for AI interaction, we will use #H->AI:: as a potential prefix, followed by a keyword indicating the communication type and a description in ().

  # EXAMPLES - New variations can be created as needed

  #H->AI::Directive: (Command/Instruction. Explicit commands telling the AI what to do or how to behave. This goes beyond the core prompt request. )

  #H->AI::Context: (Background/Setting. Provides background information, situational context, assumptions, or constraints the AI should factor into its reasoning and response. Sets the stage. )

  #H->AI::Constraint: (Limitation/Boundary. Defines specific limitations, boundaries, or things the AI must not do. More specific than general context. )

  #H->AI::Focus: (Emphasis/Priority. Highlights specific parts of the prompt or aspects of the task that require special attention or prioritization. Guides the AI's focus. )

  #H->AI::Question: (Direct Inquiry for AI. Poses a direct question to the AI that requires an explicit answer within the response, often metacognitive or clarifying. )

  #H->AI::Consider:(Implicit Guidance/Food for Thought. Provides information or questions the AI should consider or factor into its reasoning without necessarily needing to address it directly or explicitly in the output. Subtle guidance. )

  # Comments from the AI to the Human User will begin with "#AI->H::"

  #AI->H::

  # EXAMPLES - New variations can be created as needed

  #AI->H::Note: ()

  #AI->H::Caution: ()

  #AI->H::RecommendedChange: ()

  #AI->H::RecommendMakingThisAFunction: ()

  #AI->H::RecommendMakingThisAVariable: ()

  #AI->H::RecommendMakingThisAnObject: ()

  #AI->H::RequestingFeedback: ()

  #AI->H::RecommendDiffRole: ()

  #AI->H::RecommendMoreDetailHere: ()

  #AI->H::Question: ()

  # Comments from the Human User to Human User, including notes to self, will begin with "#H->H::"

  #H->H::

  # EXAMPLES - New variations can be created as needed

  #H->H::Note: (Note to self.)

  #H->H::Reminder: (Remember to do this thing.)

  """

  In this way, functions can contain comments that can be interpreted as prompts with actionable instructions.

  For example:

  def create_photo(data):

  #H->AI::CreateImage: (Create a photographic image and use the data provided in the function call as a prompt for the image description that is used to create the photo)

  #H->AI::OnError: (If the creation of the image is not possible because the A.I. cannot generate images, then respond with "Sorry, I am not capable of making images.")

  """

  # ================================================

  # 3. DATA TYPES

  # ================================================

  # Unlike data types used in a standard Python application, CRAFT data types include concepts that any A.I. can understand without a special definition, for example all A.I. services know what a URL is without an explicit definition

  """

  Current CRAFT framework data types include: URL, ColorPalette, DomainName, WhiteListedDomainName, BlackListedDomainName, Email, FileName, FileHash, MLACitation, MLAInTextCitation, Prompt, PromptLibrary, ConversationHistory, AIRecipe, AIRecipeLibrary, AIActionTemplate, AIWorkflow, AIModelConfig, UserIntent, AISentiment, RichContent, MediaReference, TLDRSummary, UserProfile, Metric, TimeSeries, DataValidator, Transformation, AIInstruction, Comparison, SocialPost, HowTo, Definition, AITone, TargetAudience, WordLength, UserIntent, AISentiment, AIPersona, CommunicationChannel, ChartSpec, CodeSnippet, APIRequest, APIResponse, BusinessLogic, ComplianceRule, ConditionLogic, Checklist, Form, MultipleChoice, Workflow, AIDecisionPoint, KnowledgeGraphNode, DefinedRelationship, ConstraintSet, OutputSchema, ComparisonMatrix, StructuredArgument, RevisionRequest, VersionTag

  ### Namespacing or prefixes define a set of data types(e.g., Analytics.URL vs. Analytics.TimeSeries).

  # ================================================

  # 3.1 DATA TYPE DESCRIPTIONS

  # ================================================

  URL: URL structure with essential components.

  ColorPalette : color palette , for example "gray #DCDCDC, orange #FF4E00, black #000000"

  DomainName: any valid TLD recognized by WhoIs

  WhiteListedDomainName: a domain that can be used

  BlackListedDomainName: a domain that cannot be used

  Email: Email structure with essential components. Simplifies a common business task with natural parameters - Example: Email(subject="Meeting Follow-up", recipient="team", tone="professional")

  FileName: File name structure with essential components including file extension.

  FileHash: any legitimate file hash

  MLACitation: a citation for a book or publication in MLA Citation Format: Last name, First name. Title. Title of the Website, Version or edition, Publisher,. Day Month Year of publication, URL.

  MLAInTextCitation: In-text citations in MLA style follow the general format of author's last name followed by a page number enclosed in parentheses. Here is an example: "Here's a direct quote" (Smith 8). If the author's name is not given, use the first word (or words) of the title.

  Prompt: A template with variables that can be filled in

  PromptLibrary: Collection of prompts organized by categories

  ConversationHistory: Tracks conversation history with an AI

  AIRecipe: A successful ready-to-use prompt structure with placeholders that has been tested. Helps prompt crafters follow proven patterns without reinventing prompts

  AIRecipeLibrary: Collection of recipes organized by categories

  AIActionTemplate: Reusable pattern for AI actions. Creates standardized workflows for common business processes - Example: RecipeActionTemplate("email_customer", parameters=["name", "order_id"])

  AIWorkflow: Bundles a sequence of AIActionTemplates into a named flow, allowing you to say “execute AIWorkflow('OnboardCustomer')” instead of listing steps. - Examples: AIWorkflow( name="OnboardCustomer", steps=["collect_info","verify_email","send_welcome"])

  AIModelConfig: Configuration parameters for AI model calls

  UserIntent: Represents a user's goal or purpose in an interaction. Helps AI understand what users want to accomplish without explicit instructions - Example: Intent("book_flight", confidence=0.92)

  AISentiment: Emotional tone in text or speech. Enables emotional intelligence in AI responses - Example: Sentiment("positive", confidence=0.87)

  RichContent: Text with formatting, links, and media. Represents formatted content without HTML/Markdown boilerplate - Example: RichContent("Check this important [resource](https://example.com)")

  MediaReference: Standardized reference to image, audio, or video. Consistent handling of media across different AI platforms - Example: MediaReference("image", url="https://example.com/img.jpg", alt="Product photo")

  TLDRSummary: Condensed version of longer content for people who need a shorter version. Standardizes content summarization without repetitive prompting - Example: Summary("Long meeting transcript...", max_length=100)

  UserProfile: Collection of user preferences and behaviors. Provides context for personalized AI interactions - Example: UserProfile(industry="healthcare", role="manager", interests=["AI", "automation"])

  Metric: Named value with unit and context. Standardized business KPI representation - Example: Metric("conversion_rate", 3.2, unit="percent", trend="increasing")

  TimeSeries: Temporal sequence of data points. Simplifies trend analysis and forecasting - Example: TimeSeries([{date: "2025-04-01", value: 120}, ...], metric="sales")

  DataValidator: Rules for validating input. Ensures data quality without complex validation code Example: DataValidator(type="email", required=true, format="corporate")

  DataTransformation: Data conversion specification. Standardizes data processing steps Example: Transformation("csv_to_table", options={"headers": true})

  AIInstruction: Clear direction for the AI to follow. Makes it explicit when users want the AI to create something - Example: [NameSpace].Instruction("Write a professional email to a client about a project delay")

  Comparison: Request to evaluate alternatives. Structures a common task in a format AI can consistently understand - Example: [NameSpace].Comparison("electric vs. gas vehicles", aspects=["cost", "environment", "convenience"])

  SocialPost: Content formatted for social media. Handles platform-specific considerations automatically - Example: [NameSpace].SocialPost(platform="LinkedIn", topic="industry trend", hashtags=3)

  HowTo: Step-by-step instructions for a process. Ensures AI provides actionable, sequential steps - Example: HowTo("create a budget spreadsheet", detail_level="medium")

  Definition: Simple meaning of a term within the context of this CRAFT conversation. Quick access to clear explanations without overwhelming detail - Example: [NameSpace].Definition("prompt engineering", context="AI")

  AITone: Writing style specification. Simplifies one of the most common adjustments beginners make - Example: [NameSpace].Tone("friendly", "professional", "concise")

  TargetAudience: Target reader description. Helps AI tailor content appropriately without complex instructions - Example: [NameSpace].Audience("marketing professionals", knowledge_level="intermediate")

  WordLength: Content size specification. Ensures consistent output format without technical parameters - Example: [NameSpace].Length("paragraph", count=3) or Length("words", count=200)

  UserIntent: Represents a user's goal or purpose in an interaction. Helps AI understand what users want to accomplish without explicit instructions - Example: Intent("book_flight", confidence=0.92)

  AISentiment: Emotional tone in text or speech. Enables emotional intelligence in AI responses - Example: Sentiment("positive", confidence=0.87)

  AIPersona: Defines a specific AI role, personality, or expertise level that can be applied temporarily or to specific tasks, potentially overriding the default session persona. Allows for more granular control over AI behavior for specific tasks, enhancing the "Human-AI Partnership" by tailoring the AI's contribution more precisely. - Example: Persona("Critical Editor", focus=["Clarity", "Conciseness"], constraints=["Avoid jargon", "Challenge weak arguments"]) --> followed by Instruction("Review this draft").using(Persona("Critical Editor"))

  CommunicationChannel: Let AI tailor tone and formatting automatically based on medium (e.g. shorter for SMS, formal for email). - Examples: Channel("email") Channel("chat") Channel("sms")

  TableData: Presents structured tabular data to AI in one go, so you skip fussing over Markdown or CSV formatting. - Examples: TableData( headers=["Month","Revenue","Cost"], rows=[ ["Jan", 12000, 8000], ["Feb", 15000, 9000], ])

  ChartSpec: Lets AI know you want a chart and how to build it—no need to describe columns and plot details in prose. - Examples: ChartSpec( colors="", type="line", x="date", y="sales", title="Weekly Sales")

  CodeSnippet: Clearly separates code from text, so AI can analyze, explain, or modify it without you wrapping triple-backticks everywhere. - Examples: CodeSnippet( language="python", content="def foo(x): return x*2")

  APIRequest: Standardizes how you describe external calls—AI can generate code or validate requests automatically. - Examples: APIRequest( endpoint="/v1/orders", method="POST", payload={"customer_id":123, "items":[...] })

  APIResponse: Lets AI parse and branch logic based on status codes or payload fields without you writing “if status == 200…” in prose. - Examples: APIResponse( status=200, body={"order_id":456, "status":"confirmed"})

  BusinessLogic: Encodes business logic declaratively, so AI can generate or audit workflow engines without hand-rolled “if” statements in prompts. - Examples: BusinessLogic( name="DiscountEligibility", condition="user.loyalty_years > 1", action="apply_discount(10%)")

  ComplianceRule: Makes regulatory constraints explicit, helping AI surface potential compliance issues or generate documentation. - Examples:

  ComplianceRule( regulation="GDPR", requirement="user_consent_required", action="log_consent()")

  ConditionLogic: Encapsulates branching logic, letting AI decide paths without parsing narrative “if…then…” descriptions. - Examples: ConditionLogic("order_total >= 100", true_branch="apply_free_shipping", false_branch="no_shipping")

  Checklist: Turns to-do lists into structured objects, so AI can track completion or send reminders automatically. - Examples: Checklist(items=["Set agenda","Invite attendees","Reserve conference room"])

  Form: Defines user inputs declaratively—AI can generate both the questions and the validation logic without extra prompting. - Examples:

  Form( fields=[ {"name":"email","type":"string","required":True}, {"name":"age","type":"integer","required":False} ])

  MultipleChoice: Models multiple-choice prompts, so AI can render options and collect answers cleanly.

  Examples: Choice( question="Preferred contact method?", options=["email","phone","sms"])

  Workflow: Represents a defined sequence of steps or actions involving multiple CRAFT data types or instructions. Formalizes multi-step processes, making complex tasks repeatable and manageable within the chat context. Aligns with structured workflows and efficiency. - Example: Workflow("NewBlogPost", steps=[Instruction("Research keywords for topic X"), ActionTemplate("GenerateOutline", topic="X"), Instruction("Write draft based on outline"), RevisionRequest("Review for tone and clarity")])

  AIDecisionPoint: Represents a conditional branch within a workflow or conversation. Introduces basic logic into CRAFT interactions without complex code, allowing for more dynamic and responsive conversational flows. Supports structuring complex interactions. - Example: DecisionPoint(trigger=Metric("CustomerSatisfaction", value=...), condition="value < 0.7", if_true=ActionTemplate("EscalateToManager"), if_false=Instruction("Send standard follow-up"))

  KnowledgeGraphNode: Represents a distinct entity or concept within a potentially larger knowledge structure being built or referenced. Facilitates building and referencing structured knowledge within the AI's context, supporting CRAFT's goal of "Cumulative Intelligence" and enabling more sophisticated reasoning. - Example: KnowledgeGraphNode(id="proj_alpha", type="Project", properties={"status": "Ongoing", "budget": 50000}, relationships=[("managed_by", "user_jane"), ("uses_tech", "tech_python")])

  DefinedRelationship: Explicitly defines a connection between two entities, data points, or CRAFT objects. Makes connections between pieces of information explicit, improving the AI's ability to understand context and draw inferences. - Example: Relationship(subject=Entity("Company A", category="organization"), predicate="partnership_with", object=Entity("Company B", category="organization"), confidence=0.95)

  ConstraintSet: A named, reusable collection of rules or limitations to be applied during generation or analysis. Improves consistency and efficiency by standardizing the application of complex requirements across multiple prompts or sessions. Reduces boilerplate instructions.

  Example: ConstraintSet("BrandVoiceGuidelines", rules=["Tone: professional_yet_approachable", "Must include: call_to_action", "Avoid: overly technical terms"]) --> used later with Instruction("Write ad copy").with(ConstraintSet("BrandVoiceGuidelines"))

  OutputSchema: Specifies the desired structure, format, and data types for the AI's output, going beyond simple length or tone. Ensures the AI generates information in a predictable, structured format suitable for direct use or further processing, enhancing reliability and structure.

  Example: OutputSchema("CompetitorReport", fields={"competitor_name": "string", "key_product": "string", "estimated_revenue": Metric(unit="USD"), "summary": Summary(max_length=150)})

  ComparisonMatrix: A structured format for comparing multiple items across a set of defined criteria. Standardizes the output for comparison tasks, making it easier for users (especially entrepreneurs) to evaluate options. More structured than the beginner Comparison type. - Example: ComparisonMatrix(items=["Software A", "Software B"], criteria=["Pricing", "Ease of Use", "Feature X"], data=[["$50/mo", "High", True], ["$45/mo", "Medium", False]])

  StructuredArgument: Represents a structured claim supported by evidence and reasoning. Useful for structuring persuasive content, documenting decisions, or analyzing proposals within a business context. Promotes clarity and structured thinking. - Example: StructuredArgument(claim="We should invest in Solar Panels", evidence=[Metric("Energy Savings", value=15, unit="percent"), MediaReference("CaseStudy.pdf")], reasoning="Long-term cost reduction and environmental benefits outweigh initial investment.")

  RevisionRequest: A formal instruction to modify a specific piece of previously generated content, potentially linking to it by an ID. Structures the feedback and iteration loop within CRAFT, making collaboration more organized, especially in multi-session projects handled via Handoffs . Example: RevisionRequest(target_id="blog_post_draft_1", change="Rewrite the introduction to be more engaging", rationale="Current version is too dry")

  VersionTag: Associates a label or identifier with a specific state of a CRAFT object or generated content. Introduces basic versioning concepts, helping to track changes and manage different iterations of work within the CRAFT project context, contributing to "Cumulative Intelligence". Example: VersionTag(target_id="business_plan_doc", tag="v2.1-FinalReview", timestamp="2025-05-05T12:30:00Z")

  """

  # ================================================

  # 4. GLOBAL_VARIABLES

  # ================================================

  """

  Benefits of CRAFT Variables:

  • Reusability: Use the same value across multiple prompts without retyping it each time.

  • Consistency: Ensures that all parts of your conversation or blog content reference the same core information, reducing human error.

  • Efficiency: Saves tokens by shortening your instructions, which is critical when working within token limits.

  """

  ### START PYTHON CODE BLOCK ###

  def build_expert_prompt(domain, expertise_level, task, constraints=None):

  """Constructs an expert prompt with consistent structure."""

  base_prompt = f"""

  As an expert in {domain} with {expertise_level} experience,

  your task is to {task}."""

  if constraints:

  constraint_text = "\n\nWork within these constraints:\n"

  for c in constraints:

  constraint_text += f"- {c}\n"

  base_prompt += constraint_text

  return base_prompt

  ### END PYTHON CODE BLOCK ###

  # Usage Example

  """

  FINANCIAL_ANALYSIS_PROMPT = build_expert_prompt(

  "financial analysis",

  "10+ years",

  "evaluate this company's quarterly results",

  ["Focus on cash flow indicators", "Compare against industry benchmarks"]

  )

  """

  ### START PYTHON CODE BLOCK ###

  def build_complex_persona(name, role, goals, pain_points, tech_comfort):

  """Creates a standardized persona variable with consistent format."""

  return f"""

  PERSONA_{name.upper()} = \"\"\"

  Name: {name}

  Role: {role}

  Primary Goals: {goals}

  Key Pain Points: {pain_points}

  Technology Comfort: {tech_comfort}

  \"\"\"

  """

  ### END PYTHON CODE BLOCK ###

  # Usage Example

  """

  PERSONA_ALEX = build_complex_persona(

  "Alex",

  "Marketing Director",

  "Increase brand visibility, Improve campaign ROI",

  "Limited technical resources, Data silos",

  "Moderate - comfortable with analytics dashboards"

  )

  """

  ### START PYTHON CODE BLOCK ###

  def MainBlogImage(content):

  """

  Generate 10 AI image prompts based on provided blog content.

  """

  # (function body defined in project repository)

  pass

  ### END PYTHON CODE BLOCK ###

  # ================================================

  # 5. GLOBAL_FUNCTIONS

  # ================================================

  """

  Benefits of CRAFT Functions:

  • Modularity: Once a function is defined, you can easily reuse it for different projects or tasks without rewriting the steps.

  • Automation: Automates complex instructions into a single call, which saves time and minimizes errors.

  • Creative Consistency: Ensures that even creative tasks (like generating image prompts) follow a consistent and professional formula.

  """

  # ================================================

  # 6. GLOBAL_OBJECTS

  # ================================================

  Chat Objects combine variables and functions into a cohesive unit. They represent higher-level structures in your instructions that bundle together related tasks or information.

  Benefits:

  • Organization: They provide a neat and structured way to handle complex projects where multiple tasks and data points are interrelated.

  • Scalability: As your prompt library grows, objects help manage this complexity in a clean, modular format.

CRAFT Framework: Enhancing AI Interactions Through Custom Data Types

The CRAFT (Configurable Reusable AI Framework Technology) framework introduces a robust set of custom data types designed to transform casual AI interactions into structured, efficient workflows. These data types adapt principles of object-oriented programming to AI conversations, creating a more intuitive interaction paradigm. Let's explore these specialized data types and how they benefit entrepreneurs and AI enthusiasts.

Why CRAFT Data Types Matter

Unlike standard programming data types, CRAFT data types are intuitive concepts that any AI can understand without explicit definitions. They create structure while enhancing flexibility, allowing for more precise communication with AI systems.

CRAFT Framework Data Types: A Detailed Overview

The CRAFT framework introduces specialized data types that streamline AI interactions and bring programming-like structure to natural language conversations. These data types are designed with specific purposes in mind, catering to various aspects of AI interaction.

Fundamental Communication Types

Content Organization Types

• TLDRSummary: Creates standardized content summaries without repetitive prompting.

  • Example: Summary("Long meeting transcript...", max_length=100)

  • Benefit: Ensures consistent formatting for condensed versions of longer content.

• RichContent: Represents formatted text with links and media without requiring HTML/Markdown boilerplate.

  • Example: RichContent("Check this **important** [resource](https://example.com)")

  • Benefit: Simplifies the creation of formatted content in AI interactions.

• MediaReference: Provides standardized references to images, audio, or video.

  • Example: MediaReference("image", url="https://example.com/img.jpg", alt="Product photo")

  • Benefit: Ensures consistent handling of media across different AI platforms.

Communication Enhancement Types

• Email: Simplifies a common business task with natural parameters.

  • Example: Email(subject="Meeting Follow-up", recipient="team", tone="professional")

  • Benefit: Streamlines email creation by specifying key components in a structured format.

• AITone: Specifies writing style with simple parameters.

  • Example: Tone("friendly", "professional", "concise")

  • Benefit: One of the most common adjustments beginners make - now simplified.

• TargetAudience: Describes the intended readers to help AI tailor content appropriately.

  • Example: Audience("marketing professionals", knowledge_level="intermediate")

  • Benefit: Ensures content is consistently pitched at the right level without complex instructions.

• WordLength: Specifies content size without technical parameters.

  • Example: Length("paragraph", count=3) or Length("words", count=200)

  • Benefit: Guarantees consistent output format with intuitive parameters.

Business Intelligence Types

Metrics and Analysis

• Metric: Represents a named value with unit and context.

  • Example: Metric("conversion_rate", 3.2, unit="percent", trend="increasing")

  • Benefit: Standardizes the representation of business KPIs.

• TimeSeries: Represents a temporal sequence of data points.

  • Example: TimeSeries([{date: "2025-04-01", value: 120}, ...], metric="sales")

  • Benefit: Simplifies trend analysis and forecasting in business contexts.

• DataValidator: Defines rules for validating input.

  • Example: DataValidator(type="email", required=true, format="corporate")

  • Benefit: Ensures data quality without requiring complex validation code.

• DataTransformation: Specifies data conversion processes.

  • Example: Transformation("csv_to_table", options={"headers": true})

  • Benefit: Standardizes data processing steps across AI interactions.

Content Production Types

• SocialPost: Formats content specifically for social media platforms.

  • Example: SocialPost(platform="LinkedIn", topic="industry trend", hashtags=3)

  • Benefit: Handles platform-specific considerations automatically.

• HowTo: Provides step-by-step instructions for processes.

  • Example: HowTo("create a budget spreadsheet", detail_level="medium")

  • Benefit: Ensures AI provides actionable, sequential steps for any process.

• Definition: Offers simple meanings of terms within the context of CRAFT conversations.

  • Example: Definition("prompt engineering", context="AI")

  • Benefit: Provides quick access to clear explanations without overwhelming detail.

• Comparison: Evaluates alternatives across specified dimensions.

  • Example: Comparison("electric vs. gas vehicles", aspects=["cost", "environment", "convenience"])

  • Benefit: Structures comparative analysis in a format AI can consistently understand.

Advanced Workflow Types

AI Interaction Control

• AIModelConfig: Sets configuration parameters for AI model calls.

  • Benefit: Provides fine-grained control over AI behavior for specific tasks.

• AIInstruction: Gives clear directions for the AI to follow.

  • Example: Instruction("Write a professional email to a client about a project delay")

  • Benefit: Makes it explicit when users want the AI to create something specific.

• AIWorkflow: Bundles sequences of AI actions into named flows.

  • Example: AIWorkflow(name="OnboardCustomer", steps=["collect_info","verify_email","send_welcome"])

  • Benefit: Allows execution of complex processes with simple commands.

• AIRecipe: Represents successful, ready-to-use prompt structures with placeholders.

  • Benefit: Helps prompt crafters follow proven patterns without reinventing prompts.

• AIPersona: Defines specific AI roles, personalities, or expertise levels.

  • Example: Persona("Critical Editor", focus=["Clarity", "Conciseness"], constraints=["Avoid jargon"])

  • Benefit: Provides more granular control over AI behavior for specific tasks.

Development-Oriented Types

• URL: Represents URL structure with essential components.

  • Benefit: Ensures consistent handling of web addresses.

• ColorPalette: Defines color schemes for design consistency.

  • Example: ColorPalette("gray #DCDCDC, orange #FF4E00, black #000000")

  • Benefit: Maintains design consistency across generated content.

• FileName: Represents file name structure with essential components.

  • Benefit: Standardizes file naming conventions within the CRAFT ecosystem.

• CodeSnippet: Clearly separates code from text for analysis or modification.

  • Example: CodeSnippet(language="python", content="def foo(x): return x*2")

  • Benefit: Improves code handling in AI interactions without requiring markdown formatting.

Enterprise & Workflow Types

Structure and Organization

• Workflow: Represents defined sequences of steps involving multiple CRAFT data types.

  • Example: Workflow("NewBlogPost", steps=[Instruction("Research keywords"), ActionTemplate("GenerateOutline")])

  • Benefit: Makes complex tasks repeatable and manageable within the chat context.

• AIDecisionPoint: Creates conditional branches within workflows or conversations.

  • Example: DecisionPoint(trigger=Metric("CustomerSatisfaction"), condition="value < 0.7", if_true=ActionTemplate("EscalateToManager"))

  • Benefit: Introduces basic logic into CRAFT interactions without complex code.

• ConstraintSet: Provides named, reusable collections of rules or limitations.

  • Example: ConstraintSet("BrandVoiceGuidelines", rules=["Tone: professional_yet_approachable"])

  • Benefit: Improves consistency by standardizing complex requirements across multiple prompts.

Documentation and Knowledge Management

• KnowledgeGraphNode: Represents distinct entities within larger knowledge structures.

  • Example: KnowledgeGraphNode(id="proj_alpha", type="Project", properties={"status": "Ongoing"})

  • Benefit: Facilitates building structured knowledge within the AI's context.

• DefinedRelationship: Explicitly defines connections between entities or data points.

  • Example: Relationship(subject=Entity("Company A"), predicate="partnership_with", object=Entity("Company B"))

  • Benefit: Improves the AI's ability to understand context and draw inferences.

• VersionTag: Associates labels with specific states of CRAFT objects or content.

  • Example: VersionTag(target_id="business_plan_doc", tag="v2.1-FinalReview")

  • Benefit: Helps track changes and manage iterations within CRAFT projects.

Business Logic and Compliance

• BusinessLogic: Encodes business rules declaratively.

  • Example: BusinessLogic(name="DiscountEligibility", condition="user.loyalty_years > 1", action="apply_discount(10%)")

  • Benefit: Allows AI to generate or audit workflow engines without complex instructions.

• ComplianceRule: Makes regulatory constraints explicit.

  • Example: ComplianceRule(regulation="GDPR", requirement="user_consent_required", action="log_consent()")

  • Benefit: Helps AI surface potential compliance issues or generate documentation.

• OutputSchema: Specifies desired structure and format for AI outputs.

  • Example: OutputSchema("CompetitorReport", fields={"competitor_name": "string", "estimated_revenue": Metric(unit="USD")})

  • Benefit: Ensures predictable, structured output suitable for further processing.

Key Benefits for Entrepreneurs

1. Enhanced Communication Efficiency

CRAFT data types streamline communication with AI systems by providing clear, structured templates for common tasks. Instead of writing lengthy prompts explaining what you want, you can use predefined data types like Email, SocialPost, or HowTo to quickly communicate your needs.

2. Consistency Across Projects

By using standardized data types, entrepreneurs ensure consistent outputs across different AI interactions and projects. Whether you're generating content, analyzing data, or creating workflows, CRAFT provides a unified approach that maintains quality and format consistency.

3. Reduced Learning Curve

The framework uses familiar programming concepts but adapts them to natural language interaction, making them accessible to both technical and non-technical users. This reduces the learning curve for effective AI utilization.

4. Business Process Integration

With specialized types like Metric, BusinessLogic, and ComplianceRule, entrepreneurs can integrate AI more seamlessly into existing business processes and ensure outputs meet regulatory requirements.

5. Knowledge Management

Data types like KnowledgeGraphNode, DefinedRelationship, and VersionTag support building structured knowledge bases that evolve over time, embodying the CRAFT principle of "Cumulative Intelligence."

Real-World Applications

CRAFT data types enable entrepreneurs to:

• Create structured content workflows (using Workflow and AIDecisionPoint)

• Maintain brand consistency (with ConstraintSet and AITone)

• Streamline data analysis (through Metric and TimeSeries)

• Ensure compliance (via ComplianceRule and OutputSchema)

• Build institutional knowledge (using knowledge graph components)

Getting Started with CRAFT Data Types

The most effective way to implement CRAFT data types is to start with the basics that align with your most common AI interactions. Begin with communication-focused types like Email, HowTo, or SocialPost, then gradually incorporate more complex types as your AI workflows mature.

CRAFT's innovative data types represent a significant advancement in AI interaction design, bridging the gap between programming concepts and natural language communication. By providing structure without sacrificing flexibility, these data types enable entrepreneurs to leverage AI more effectively while maintaining consistency across projects.

---