Luzmo IQ
Get instant, concise answers to any data question, backed by easy-to-understand charts. Empower your customers with faster, clearer insights. Integrate Luzmo IQ as an AI-powered assistant, embedded search – or power your own app with our components and APIs.
Luzmo IQ uses AI to supercharge your end-user analytics – making them smarter and more intuitive than ever. Offer your users a more personalized, more accessible, and more digestible form of data analysis, thanks to AI.
Get AI-powered insights to your customers, fast – and make your data accessible to the widest possible audience. Effortlessly generate high-quality answers to every user’s natural language data questions.
View our Developer Docs to see how to integrate Luzmo IQ in your product.
Use flexible UI components and our API to integrate our AI tools faster than you can say Large Language Model. Plug them into your product and workflows however and wherever you like.
Use Luzmo to automate AI indexes and vector embeddings – making data retrieval faster and more accurate. No complex setup means you can focus on an exceptional data experience.
Effortless integration – with all the variety you need. Connect to our API endpoints or use components out-of-the-box.
Get accurate, well-reasoned insights for any data question. Our AI can clarify points upfront and check the relevancy of its responses to ensure high-quality answers – and no data hallucinations.
Answer data questions in text, beautiful interactive visualizations – or both. Refine responses by getting users to rate answers, and connect multiple datasets with no limitation on their size.
Protect user information with our Access Control Layer, so only authorised datasets are queried. None of your data is used to train models – it stays secure & you stay in control.
All your questions about Luzmo IQ answered.
Is my data sent to OpenAI?
Your data is not used by OpenAI for model training.
To create embeddings, row-level data of hierarchy columns is sent to OpenAI.
To answer questions, user input, dataset metadata, and query results (aggregated data or a limited number of rows) are sent to OpenAI.
Do I need to create embeddings?
No, but this will reduce the speed and range of questions that can be answered successfully.
Can I use this to build something other than the chat widget?
Yes, most likely. On top of a set of highly customizable components, you can also use our API directly.
Will this replace the AI Chart Generator?
No. There’s still a good use case for a simpler AI feature that just creates charts, without being aware of the data that goes into the chart. It’s faster, included with Luzmo’s subscriptions (Pro and up), and does only sent metadata to OpenAI.
Which type of question cannot be answered?
- “Why” questions may be difficult to answer depending on the question and breath of relevant data.
- Luzmo IQ can’t do trend predictions, so questions about the future cannot be answered.
How can I be sure the answers are correct?
This is what we do to improve answer quality:
- Ask for clarification in case the LLM is unsure how to answer, greatly reducing hallucinations.
- Run queries with Luzmo’s query engine, which prevents query mistakes and only requires interpretation of aggregated data by the LLM.
- Employ a feedback loop to improve answer results over time.
- Allow users to view and edit how charts are built, giving them control over the end result.
Can I use AI model X, or my own model?
Not right now. We’ve been testing a broad range of models, and experienced it’s not easy to guarantee answer quality across providers.
What are requirements for the underlying data?
Datasets should have descriptive names, column names, and if relevant dataset links set up. The format of these names doesn’t matter (e.g. “shipment_date” or “shipmentDate” are just as good as “shipment date”.).
Luzmo IQ can create formulas to answer questions, but it may help to define much-used formulas on the dataset itself.