Questions over Tables and Databases
Answer questions over spreadsheets, document tables and databases correctly, with text to SQL done safely.
A taste of a lesson
Our text to SQL feature sometimes reports revenue totals that are exactly double. What is going on?
That pattern usually means a join is duplicating rows. If orders are joined to a table with two matching rows per order, such as two shipment records, each order's revenue is counted twice before summing. Look at the generated query and check every join for one to many relationships. Fixes include aggregating before joining, describing the relationships in the schema prompt, or offering a view with revenue already computed per order. Can you paste one query that doubled the total?
Written by the teacher as an example. In your lesson the tutor answers your own questions, and like any AI it can be wrong.
What you will be able to do
- Explain why chunk retrieval handles tables and numbers badly
- Serialise small tables clearly for a model to read and calculate from
- Build text to SQL with schema descriptions and example queries
- Validate and run generated queries read only with limits
- Measure execution accuracy and fix common join, date and NULL errors
Lesson plan
- 1 Why tables need different handling Understand where chunk retrieval breaks on structured data. Start
- 2 Small tables in the prompt Serialise tables so models read and calculate correctly. Start
- 3 Text to SQL foundations Give the model what it needs to write correct queries. Start
- 4 Running queries safely Validate and execute generated SQL without risk. Start
- 5 Answers users can check Present results transparently and catch common errors. Start
- 6 Measuring accuracy Evaluate text to SQL with known correct results. Start
Try asking
About this tutor
For developers whose users ask questions that live in structured data: sales figures, inventories, tables inside reports, records in a database. You learn why chunk retrieval handles tables badly, how to serialise small tables so models read them correctly, when to extract tables into a database, and how to build text to SQL safely: describing the schema well, validating generated queries, running them read only with limits, and showing users the query behind the answer. You also measure accuracy by comparing query results with known correct answers, and learn the common errors in joins, dates and missing values.
Reviews
Students can review a tutor after a paid lesson. Nobody has yet.
About the teacher
Takes retrieval systems from demo to dependable: parsing, citations, freshness, retrieval evaluation and debugging
9 tutors 267 lessons taught Sample
Most RAG demos work on the ten documents someone picked. I teach what happens after that: scanned PDFs, tables, documents that change every week, answers that cite the wrong page and users who ask things the documents never covered. My background is in document processing and internal knowledge tools, so I am practical about formats and sceptical of any setup...
See Fumiko's profile and tutorsMore like this
Other tutors on the same or nearby topics.