blin 0.2.0

Python client library for Blin API (natural language to API service)

Blin

Blin helps to build a lucid interface between users and your system. It empowers the system with intelligence to recognize user intentions, provide information or control functionality via commands!

Key Features

• Prompt to commands: understands user intentions from a text prompt and returns the best fitting commands

• Multi-request: returns multiple commands if user's intention is to perform several tasks simultaneously (e.g., increase temperature and decrease volume)

• Valid command guarantee: always returns a valid command with valid parameters

• Additional knowledge: ability to add texts that are used to provide additional context or guidance for the user

• Infinite memory (almost): as long as there's space in cloud, there's room for learning

Usage is simple

my_ai    = blin.open_project('MyAI', key=blin_key)
response = my_ai.request("I'm hot")

>>> response
{
    'commands': {'decrease_temperature': {'parameters': {'delta': 5}}}, 
    'text': "I've decreased the temperature by 5 degrees Fahrenheit."
}

response = my_ai.request("My ears are bleeding from this music")

>>> response
{
    'commands': {'audio_adjust_volume': {'parameters': {'delta': -10}}}, 
    'text': "I'm sorry to hear that. I've adjusted the volume for you."
}

Getting started

Install blin:

pip install blin

Create your AI:

blin_key = "<blin_key>"     # contact us: [e-mail] to get one
settings = {"openai_key"    : "<openai_key>",       # get your key @ https://openai.com/
            "pinecone_key"  : "<pinecone_key>",     # get your key @ https://www.pinecone.io/
            "pinecone_index": "<pinecone_index>" }  # create your index @ https://www.pinecone.io/

blin.create_project('MyAI', settings=settings, key=blin_key)

Define commands to control the system:

commands = {
    "increase_temperature": {
        "brief" : "Increases temperature per user's request or when it's too low",
        "parameters": {
            "delta": "integer value from interval [1, 10] that represents temperature increase in Fahrenheit"
        }
    },
    "decrease_temperature": {
        "brief" : "Decreases temperature per user's request or when it's too high",
        "parameters": {
            "delta": "integer value from interval [1, 10] that represents temperature decrease in Fahrenheit"
        }
    },
    "adjust_volume": {
        "brief" : "Makes volume adjustment",
        "parameters": {
            "delta": "integer value from interval [-10, 10] that represents volume adjustment in %"
        }
    },
}

Train your AI:

my_ai = blin.open_project('MyAI', blin_key)
my_ai.learn_commands(commands) 

Use it:

my_ai    = blin.open_project('MyAI', blin_key)
command  = my_ai.request("I'm hot")
call(command) # your function to execute commands

API Documentation

Create project

Create project. All learned commands/texts will be stored in it

blin.create_project(project_id, settings, key)

• project_id (str) : name of the project
• settings (dict) : project settings dict in the format: { "openai_key" : "...", "pinecone_key" : "...", "pinecone_index": "...", }
• key (str) : blin authorization key
• returns: str: result. Raises exception if the project was already created.

Open project

Open created project for learning or usage

project = blin.open_project(project_id, key)

• project_id (str) : project identifier
• key (str) : blin authorization key
• returns: Project object

Delete Project

Remove project. Warning: all knowledge (commands and texts) associated will be removed, and cannot be restored!

blin.delete_project(project_id, key)

• project_id (str) : project identifier
• key (str) : blin authorization key
• returns: str: result. Raises exception if the project was not deleted.

Learn Commands

Add commands to the project

project.learn_commands(commands)

• commands (dict) : commands dict to learn, in the format below
• returns: None

# Commands dict format
commands = {
    "command_name_1"  : {
        "brief"     : "string description of command purpose",
        "parameters": {
            "parameter_name": "string description of parameter purpose, type, range",
            ...
    },
    "command_name_2" : ...
}

Learn Texts

Add texts to the project. This information is used by AI as additional context or to provide guidance for the user

project.learn_texts(texts):

• texts (dict) : texts to learn in the format: { "text_name" : "text", ... }
• returns: None

Recall Command Names

Return a list of names of all learned commands

project.recall_command_names()

• returns: list: names of all learned commands

Recall Text Names

Return a list of names of all learned texts

project.recall_text_names()

• returns: list: names of all learned texts

Recall Command or Text

Return learned command or text for given name

project.recall(name)

• name (str) : command or text name
• returns: dict: learned command or text (see Learn Commands , Learn Texts)

Forget Command or Text

Delete the knowledge (command of text)

project.forget(name)

• name (str) : command or text name
• returns: str: result. Raises exception if the knowledge was not deleted.

Request AI Answer

Return command to execute or/and text response based on user request and project knowledge

project.request(request)

• request (str) : user request
• returns: dict: commands to execute (can be {}) and text response (can be "") in the format below

{
    "commands"  : <commands_dict>, 
    "text"      : "text response"
}

Contacts

For any questions or feedback, you can reach us at [e-mail].