Year 7 project · not released yet

Helping big data make more sense.

Sweep AI is a growing family of models designed to read big spreadsheets and databases, analyse patterns and turn the findings into clear reports.

Built by a Year 7 student curious about what computers can help people understand.

library_loans_FINAL_v3 (2).csv
MonthYear levelLoans
aprall120
Aprall120, duplicate row
mayallN/A
mayall150
junMissing year level18O, invalid number
junall180
Example summary

After excluding the highlighted rows, library loans went from 120 in April to 180 in June. That’s a 50% increase, though the numbers don’t tell us why.

Fictional data · example only

  • 01Look at data
  • 02Analyse patterns
  • 03Report clearly
  • 04Improve over time

02 · Why Sweep

Not just another chatbot. A direction for useful models.

The idea behind Sweep is simple: help people work through large databases without making the information feel impossible to use.

Explore the workflow

See the signal

Find useful information inside large, messy collections of data.

Explain the result

Turn analysis into reports that are easier for people to read and act on.

Keep getting better

Each new model should do the work better than the one before it.

03 · How it works

Three jobs, in this order.

The planned workflow: understand the source, look for patterns, and explain the findings.

  1. Look

    Check what each column measures, how much data is present and where information is missing.

  2. Analyse

    Compare groups or time periods, look for changes, and flag unusual values worth checking.

  3. Report

    Summarise the main finding in plain language and explain what the data cannot tell us.

The next page shows what a report might look like using made-up school library data.

04 · Sample report

What could a Sweep report look like?

Using the fictional library data from the home page. This example isn’t generated by a live model.

library_loans_clean.csv

Example data · April to June

School library loans

+50%
View sample data
MonthLoans
April120
May150
June180

Explanation · In plain language

Loans rose each month.

Main finding
June had 60 more loans than April, a 50% increase. The total went up by 30 loans each month.
What we don’t know
These totals show an increase, but don’t tell us what caused it.
Next question
Did borrowing rise across all year levels, or was most of the increase in one group?

05 · The models

Four models, from small to big.

NameWhat it’s forFormatAiming for
sweep-1 The first preview: a starting point for looking at data and explaining what it finds. Preview Building now
Sweep Lite A smaller, quicker version for straightforward analysis and useful summaries. Lightweight
Sweep Latest A more capable model built for deeper analysis and longer reports. Your computer (desktop SaaS)
Sweep Ultra The long-term goal: the best-quality results Sweep can deliver. Your computer (desktop SaaS)

Our rule: aim for better-quality results as the models advance, with clear limits, careful testing and people still in control.

06 · Roadmap

One year, four steps.

These dates are targets. If something needs more testing, the date moves.

  1. Release Sweep Lite

    A faster, lighter model for everyday data questions and quick summaries.

  2. Launch Sweep Latest

    A desktop model that runs on your own computer, with stronger analysis and reporting.

  3. Build Sweep Ultra

    The most advanced planned model, aiming for Sweep’s highest-quality analysis and reports.

07 · Who’s making this

I’m a Year 7 student building something useful.

I’m trying to make models that can help with big databases by looking at data, analysing it and reporting what matters.

Sweep is still being built. This website is a place to share the idea, the roadmap and the progress.

Sweep is an independent learning project. The release dates and features are goals, not currently available products.

See the roadmap