Cortex

Cortexes are AI agents that can be easily customized to accomplish specific tasks. Cortexes can be customized with custom instructions about the task it should accomplish, and catalogs of documents that it has access to when generating content and answers.

Create or update a cortex

Cortexes are created and updated through the same configureCortex method available on the client:

const cortex = await client.configureCortex("slack-support", {
  // data sources that this cortex has access to when generating responses
  catalogs: ["docs", "previous-support-cases"],
  friendlyName: "ACME Support",
  instructions: [
    "You will recieve messages from users that are asked via Slack. Respond to their questions to the best of your ability.",
    "When appropriate, provide links to help articles",
    'If you cannot answer a question, or sense that the user becomes frustrated, send a reponse with "@acme-support please engage"',
  ],
  public: true,
  customizations: {
    rules: ["never offer any discounts or information on pricing"],
  },
});

Cortex configuration reference

CortexConfig

The CortexConfig type defines the configuration options for a Cortex instance.

Field	Type	Description
friendlyName	string	The name that this Cortex should refer to itself as (e.g., "Acme Assistant", "Acme AI").
catalogs	string[]	(Optional) The catalogs that should be referenced when generating content and answering questions.
instructions	string[]	A complete description of the goal task and a list of steps this cortex should follow when generating content and answering questions.
public	boolean	Whether or not this Cortex should be available on the internet without authentication. Common for scenarios like publishing blog content and customer support.
customizations	object	(Optional) Fine-tuned control over the verbosity, tone, and rules used to generate content.
chatConfig	object	(Optional) Configuration for the embedded hosted chat UI.
overrides	object	(Optional) Override org-level defaults.

CortexConfig.customizations

Field	Type	Description
rules	string[]	(Optional) A list of "dos and don'ts" that the cortex should follow.
personality	string[]	(Optional) Personality traits or characteristics for the cortex.

CortexConfig.chatConfig

Field	Type	Description
greeting	string	The initial greeting message rendered in chat.
intro	string	An introductory message that explains the purpose of this Cortex and what it can help with.
examples	string[]	A list of questions that will be rendered and suggested to users when they load the chat window.

CortexConfig.overrides

Field	Type	Description
inheritRules	boolean	(Optional) Whether or not global org-level rules should be followed. Defaults to 'true'.
companyName	string	(Optional) The company name.
companyInfo	string	(Optional) A description of your company, industry, and products and services provided.

Get a cortex

const cortex = await client.getCortex("support-engineer");

Chat

The Cortex class exposes convenient methods for starting a streaming or synchornous chat:

Streaming chat

To stream chat responses, set the stream parameter to true:

const cortex = await client.getCortex("support-engineer");
 
// ask questions and stream responses
const streamingChatResponse = await cortex.chat({
  message: "What kind of hammers are best to use on my ACME anvil?",
  stream: true,
});
streamingChatResponse.responseStream.on("data", (data) => {
  process.stdout.write(data.toString());
});
 
const chat = await streamingChatResponse.chat;
 
// respond to chats maintaining memory of previous responses
const respondResponse = await chat.respond({
  message: "Is it safe to use my copper mallet hammer on my anvil?",
  stream: true,
});
respondResponse.responseStream.on("data", (data) => {
  process.stdout.write(data.toString());
});
 
await respondResponse.chat;

Synchronous chat

const cortex = await client.getCortex("support-engineer");
 
const chat = await cortex.chat({
  message: "What kind of hammers are best to use on my ACME anvil?",
});
 
console.log(JSON.stringify(chat.messages));
 
await chat.respond({
  message: "Is it safe to use my copper mallet hammer on my anvil?",
});

Generate content

The Cortex class exposes methods for generating content via streaming and synchronous methods:

Streaming content generation

To stream content as it is generated, set the stream parameter to true:

const title = "Building a Support Slackbot for your Online Community";
const prompt = `Write a blog post about using the Cortex Click SDK to index developer documentation, 
into a catalog, create a cortex that consumes it, and finally auto-repsonds to community questions in slack with answers.
`;
 
let contentStream = await cortex.generateContent({
  title,
  prompt,
  stream: true,
});
 
contentStream.contentStream.on("data", (data) => {
  process.stdout.write(data);
});
 
let content = await contentStream.content;

Synchronous content generation

const title = "Building a Support Slackbot for your Online Community";
const prompt = `Write a blog post about using the Cortex Click SDK to index developer documentation, 
into a catalog, create a cortex that consumes it, and finally auto-repsonds to community questions in slack with answers.
`;
 
let blogPost = await cortex.generateContent({
  title,
  prompt,
});
 
console.log(blogPost.content);