AI in document search: 70% less search time in an automotive group’s pilot

Featured image: How Much Time AI Saves in Document Search – a Pilot from the Automotive Industry
How much time does searching through documents really cost, and how much of it can AI win back? A pilot at an automotive group with around 450,000 documents in its own document management system gives a clear answer.
↻ Updated on 8 November 2026

An engineer needs the results of an endurance test from three years ago. She knows the report exists. She just does not know what it is called, which folder it is in, or whether the figures in the management presentation match those in the report. So she searches. And searches.

In many companies, the most important knowledge does not sit in databases but in documents: test reports, technical reports and presentations. Anyone who needs an answer often searches longer than the actual work takes. We examined how much time AI can win back in a pilot project with an automotive group.

around 450,000documents in the group’s document management system
a few hundreddocuments in the pilot sample
around 70%less time for searching and compiling

The starting point: knowledge in the company’s own DMS, but hard to find

The group, whose name is anonymised for legal reasons, manages its documentation in an in-house document management system. It holds around 450,000 documents that have grown over years and across departments: PDF, Excel and PowerPoint, without a consistent structure.

A keyword search finds files there, but not answers. Employees open one document after another, compare versions and compile the relevant passages by hand.

The pilot at a glance
Industry
Automotive group (anonymised)
Data
Unstructured documents from an in-house DMS: PDF, Excel, PowerPoint
Total holdings
around 450,000 documents
Pilot scope
Sample of a few hundred documents
Measured result
around 70% less time for searching and compiling test reports and documentation

The pilot: one sample, one clear question

For the pilot, we opened up a sample of a few hundred documents from the DMS with aiuno. The question was simple: how much faster do employees find and compile the relevant information when they can ask the document base instead of searching each document individually?

How searching worked before

  • A keyword search returns a list of files
  • Open and read document after document
  • Compare versions by hand
  • Copy relevant passages into a separate document

How it works with aiuno

  • Ask the whole document base in plain language
  • Relevant sections from several documents in one answer
  • The source in the original for every statement
  • One click opens the document at the right place

The projection: what 70% can mean in day-to-day operations

We measured the saving on the sample. A projection with disclosed assumptions shows what it would mean for the entire document base.

Model calculation. The following figures are based on the pilot results and the stated assumptions. They are not a measured saving in live operation.
Assumption or resultValue
Usage5,000 users per month, each spending an average of one hour searching for documents
Search time today5,000 hours per month
Search time with AI1,500 hours per month, around 70% less
Internal hourly rate€100
Potential per month3,500 hours × €100 = €350,000
Potential per year€350,000 × 12 = €4.2 million

Search time per month in the model calculation

Today
5,000 h
With AI
1,500 h

Even with much more cautious assumptions, one result remains: the time that goes into searching today is one of the biggest hidden cost items in knowledge work.

Why it works: ask the document base, not the document

The difference is not a faster search, but a different way of asking.

01

Context instead of keywords

The AI understands what is meant, even if the term is worded differently in the document.

02

Answers instead of hit lists

Instead of opening files, employees get the relevant sections, compiled across several documents.

03

Evidence for every statement

Every answer points to the passage in the original. Nobody has to trust the AI, everyone can check.

Before an answer appears, aiuno checks that the document exists and that the quote is really there:

Document existsPassage existsQuote is verbatim

Where the data stays

In a corporate environment, where all this happens matters too. The documents are stored encrypted in a data centre in Germany, and AI processing takes place in Germany or the EU. No data is retained at the AI provider, and the content is never used for training. All details and evidence are available in our Trust Center.

What the pilot shows and what it does not

The pilot shows how much working time is tied up in unstructured documents and how much of it AI can win back. It does not show that every company will save exactly 70%. The size of the effect depends on the document base, on the questions employees ask, and on how scattered knowledge is today.

The projection of €4.2 million per year is therefore a model calculation, but it makes the order of magnitude visible. The quickest way to find your own figure is a test with your own data.

Frequently asked questions

Was the pilot run on the entire document base?

No. The pilot ran on a sample of a few hundred of the around 450,000 documents in the DMS. The saving of around 70% was measured there; the figures for overall operations are a projection.

How reliable is the projection of €4.2 million?

It is a model calculation with disclosed assumptions: 5,000 users per month, one hour of search time each, an internal hourly rate of €100 and the time saving measured in the pilot. If the assumptions change, the result changes.

Can this be transferred to other industries?

The pattern is the same wherever knowledge sits in large, unstructured document holdings: in law firms, legal departments, insurance and the pharmaceutical industry. A pilot with your own data shows how large the saving will be.

Which file formats can aiuno process?

In the pilot it was PDF, Excel and PowerPoint. aiuno also processes Word files and scanned PDFs.

Hinweis: Dieser Artikel wurde von einem menschlichen Autor verfasst und mithilfe von KI-Tools sprachlich optimiert.

Note: This article was written by a human author and linguistically refined with the help of AI tools.

Uwaga: Ten artykuł został napisany przez autora-człowieka, a jego warstwa językowa została dopracowana przy pomocy narzędzi AI.

Liked this post? Share with others!

Demo

Schenken Sie uns 20 Minuten.

Bringen Sie Ihren größten Datenraum mit. Wir zeigen Ihnen das Produkt. Gerne mit Ihren Testdaten.

Hosting in Deutschland · § 203 StGB-konform · Zero Data Retention

aiuno Screen

Demo

Give us 20 minutes.

Bring your largest data room. We’ll show you the product – happy to use your test data.

Hosted in Germany · No US transfer · Zero Data Retention

aiuno Screen

Demo

Dajcie nam 20 minut.

Przynieście swój największy data room. Pokażemy produkt – chętnie na Waszych danych testowych.

Hosting w Niemczech · Bez transferu do USA · Zero Data Retention

aiuno Screen
Scroll to Top