Updated in October 2026. An earlier version of this article walked through the IntelliCloud app, which has since been retired. Everything it did now runs in the open source libraries, Intelli for Python and IntelliNode for Node.js.
Most teams don't start by looking for an AI framework. They start with a job that eats hours every week: answering the same customer questions, reading invoices, writing product copy, or arguing about which of three models is the best.
This article goes through six of those jobs. For each one you'll see what IntelliNode does and where to find working code, so you can pick the one closest to your own problem and start there.
1. Chatbots that know your business
A support chatbot is only useful if it answers from your policies, not from the internet. IntelliNode's Assistant reads your documents, answers with numbered sources, keeps each conversation, and remembers what a customer said last week.
const assistant = new Assistant({ provider: 'openai', apiKey, knowledge, history });
await assistant.addDocuments([{ id: 'refunds.md', text: refundPolicy }]);
const reply = await assistant.chat('How long do refunds take?', { userId: 'u1' });
It runs on OpenAI, Claude, Gemini or a local model, and the same class exists in both libraries. Start with the Node.js Assistant or the Python Assistant.
2. Make documents talk
Invoices, contracts and reports hold data your systems need, locked in PDFs. IntelliNode asks a model for the fields you name and checks the reply against a JSON Schema, so you get clean data instead of a paragraph to parse.
The document extraction example pulls invoice and contract fields, adds a summary and flags risky terms. Gemini can also read audio and video, so the same idea works for call recordings and screen captures, as shown in Gemini and Vertex AI.
3. Search by meaning
Keyword search fails when the customer says "send it back" and your policy says "returns". Semantic search compares meaning instead. IntelliNode stores your text with embeddings and finds the closest passages, even with different words.
You can start with an in-memory store on a laptop and move to Pinecone, Qdrant, pgvector, MongoDB Atlas or Firestore later by changing one line. The vector stores page lists every option, and the Python version works the same way.
4. Content that fits every product
An online shop with five thousand products can't write each description by hand. IntelliNode generates the description, an image and a voice-over for each product, and you choose the model for each step.
See the ecommerce content example for Node.js, or the content platform flow in Python, where text, code and images come from different providers in one run.
5. Test models before you commit
Choosing a model from a leaderboard is a guess. A better way is to ask your own questions to a few models and score each answer against an answer you would accept. Automate LLM evaluation shows the whole script.
Once you know which model fits which job, route each request to the right one, with a fallback when a provider is down.
6. Agents that do the work
Some jobs need more than one answer. They need steps: read a ticket, decide how urgent it is, draft a reply, escalate when needed. In Node.js these steps run as a tool-calling loop. In Python, Intelli builds them as a flow of small steps and draws the flow as a picture you can review.
- Support ticket triage sorts tickets and drafts replies.
- Release smoke checks and supplier portals use computer agents that operate a browser.
- A CI quality gate reviews pull requests and writes tests.
You don't have to write the flows yourself. Give IntelliNode to Claude Code or Codex, describe the job in plain words, and review the picture the agent hands back.
Try one
If you write code, install the library for your language and open the example closest to your problem:
npm i intellinode # Node.js
pip install intelli # Python
If you don't, IntelliChat lets you try the same models in your browser with your own key.