For the non-expert

PrettyWhale.ai: the self-driving car of data code

Picture a self-driving car. You enter your starting point and your destination. The car takes care of everything the driving requires and gets you there. You didn’t have to worry about any of the driving to reach your destination. You spent the trip comfortably settled in your seat, using the time however you liked! Talk about a smooth ride!

  1. A

    Starting point

    A sample of your raw data.

  2. PrettyWhale.ai takes the wheel: analysis, transformations, code, tests, documentation.

  3. B

    Destination

    Complete code, ready for production.

PrettyWhale.ai applies the same principle of comfort to building data pipelines and putting them into production.

PrettyWhale.ai handles everything and delivers the final result to the Data Engineer, who doesn’t have to step in beyond deciding to use PrettyWhale.ai and giving it a sample of the data: the very data that has to be moved intelligently from point A to point B, just like you in your self-driving car.

First of all, what is a data pipeline?

In a company, data comes from several tools, many locations and in different formats. Specific code has to be written to retrieve it, clean it, correct it, make it usable and deliver it safely to where it will be used by an accountant, a salesperson, a division head, a logistics manager, and so on.

This specific code is called ingestion code – for the first phase of processing – and integration code for the second phase.

And it is this code that produces data pipelines, which in turn build the steps of the journey this data must follow to reach its destination and be put to use.

  1. SourcesMany tools, locations and formats
  2. 1 · IngestionRetrieve, clean, correct
  3. 2 · IntegrationMake usable, deliver
  4. UsersAccountants, sales, management, logistics…

Is ingestion and integration code fundamental?

This code is the foundation of every journey data takes.

  • The longer it takes to write, the more the project costs.
  • The poorer its quality, the riskier the project.

Who knows how to write ingestion and integration code?

Data Engineers: engineers who specialise in data AND know how to code.

Their work makes data usable by the teams in charge of analysing it (Data Analysts, Data Scientists).

This fundamental code is difficult, laborious and repetitive to write. It is just as tedious to test and document. Writing it is essential work that requires expertise, yet companies are short on time and available skills. As a result, they often hand this responsibility to junior profiles. Despite their best efforts, the code they write is often mediocre, for lack of expertise. This mediocre code therefore builds fragile, unstable and risky foundations.

A few market metrics

  • ~40%

    Data volumes grow by around 40% a year.

    IDC / Seagate

  • 2 to 3

    Two to three Data Engineers are needed for every Data Scientist. Remember: without the former, the latter cannot make use of the data.

    Jesse Anderson, O’Reilly

  • Shortage

    A growing shortage of Data Engineers since 2024.

    Le Monde Informatique

  • +41%

    A 41% increase in Data Scientist and Data Analyst positions is expected between 2025 and 2030.

    Christian Lauer, Medium

  • 60 to 80%

    60 to 80% of the time in a data project is spent finding, preparing and checking data.

    McKinsey & Company

  • Standardising and automating this work therefore frees up more resources for using data and the value it brings.

The problem

In short, companies are launching more and more data projects, but their capacity to build the foundations remains limited.

These metrics illustrate the problem PrettyWhale.ai solves.

The PrettyWhale.ai answer

PrettyWhale.ai turns the manual, hand-crafted building of pipelines into automated, standardised production.

  1. 01 · Data Engineer

    The Data Engineer gives PrettyWhale.ai a sample of the data.

  2. 02 · PrettyWhale.ai

    PrettyWhale.ai analyses the sample and suggests the transformations to apply to the data (cleaning, correction, enrichment…).

  3. 03 · Data Engineer

    The Data Engineer accepts or rejects PrettyWhale.ai’s suggested transformations in a single click.

  4. 04 · PrettyWhale.ai

    PrettyWhale.ai writes the code, tests it, fixes it and validates it.

  5. 05 · PrettyWhale.ai

    PrettyWhale.ai delivers code to the Data Engineer that can be deployed straight to production to route the data correctly.

  6. 06 · PrettyWhale.ai

    Along with the code, PrettyWhale.ai delivers the tools needed to deploy it to production and all the documentation.

With PrettyWhale.ai
A few minutes to 1 or 2 hours
With manual development
Several days, weeks and months

Depending on the complexity and scope involved, writing the code takes from a few minutes to 1 or 2 hours, where manual development ties up several days, weeks and months.

Engineering AI

We call this approach “Engineering AI”: artificial intelligence specialised in carrying out engineering work.

Just like in the self-driving car analogy, the PrettyWhale.ai user defines the journey expected for the data, and PrettyWhale.ai takes care of making it happen, here in the form of complete code.

Learn more about Engineering AI

Our ambition

Our ambition is to move data pipeline building from largely hand-crafted work to an industrial, automated and reproducible process.

You’re a companyYou’re an IT services firm

Sources: IDC / Seagate, “Rethink Data” (2020) · Jesse Anderson, “Data engineers vs. data scientists”, O’Reilly Radar (2018) · Le Monde Informatique, “La pénurie en profils qualifiés s’accentue dans l’IT en 2025” (2025) · Christian Lauer, “What Skills does a Data Engineer need?”, Medium / CodeX (2022) · McKinsey & Company, “The missing data link: Five practical lessons to scale your data products” (2025)