Data processing, spreadsheets, databases and analytics
Natural-language database and Excel search (text-to-SQL)
No more dependence on a programmer or the Excel guru. The manager types in Slack, Teams or an internal form: 'Which service made the most profit in the last six months, and who bought at least three times but has been inactive 60 days?' The AI writes the exact SQL or Python in the background, runs it read-only against PostgreSQL, MySQL, Supabase, Airtable or Google Sheets, and returns the numbers, an interactive chart and a three-sentence explanation in seconds.
New to this? Start herespreadsheets, tidying data and reportsThis is what happens to your spreadsheets and lists: a program tidies up and merges the files somebody now copies together by hand, answers questions about your own numbers in plain Hungarian, and writes the weekly summary on its own.
How it runs, in three steps
- 1The files come in as they are: exports from the webshop, the invoicing tool and the bank, plus five to ten supplier price lists, each one in a different shape.
- 2The program repairs the broken accented letters and the three different phone number formats, recognises that “Kovács és Társa Kft.” and “Kovacs es Tarsa Kft” are the same partner, merges everything into one list, and compares the new prices with the old ones.
- 3You get a tidy table, a plain answer to a question you typed in everyday words, and a Monday morning summary written in sentences; every number can be traced back to the file it came from.
A Tuesday morning todayTuesday morning goes on merging four exported files in a spreadsheet, and that the supplier put the price of copper pipe up by fifteen percent only comes out weeks later, once four products have been sold at a loss.
The same morning, automatedTuesday morning the merged table was ready by seven with a warning that copper pipe went up fifteen percent and four products would now sell at a loss, and the question “which customers have not ordered for sixty days” is answered in ten seconds.
Worth it when: It pays off first where the numbers sit in several systems at once, in webshops, wholesalers and anyone juggling many supplier price lists, and where somebody now spends half a day a week in spreadsheets. It only works if the data exists somewhere at all: from figures nobody ever recorded, nothing can be worked out.
Tools
- AI model
- database
- messaging channel
How it works
- 01Question in plain language
- 02Generated, logged, read-only query
- 03Numbers, chart, explanation
Source: AI Automation Market Report and Practical Guide (2025–2026)
Related services
Related terms
Recipes in this category
- Autonomous data cleaning, normalisation and deduplication
- Supplier price-list and catalogue harmoniser
- Cross-system ETL and central data warehouse
- Real-time anomaly detection and business alarm (AI anomaly guard)
- Self-generating weekly and monthly executive reports
- Predictive modelling: LTV and churn prediction
Have a process like this? Tell us about it.
Book a call