Austrian Media Company Saves $8 Million with Temy’s Generative AI Model
100+
millions pages processed
5X
cheaper page processing
4X
faster page processing
SOLUTION
Processing 100M+ pages without increasing costs
Our client, a 500-person media company in Austria, has been in business since 2011. They provide access to newspapers, magazines, and brochures in a clean, readable format on any device — phones, tablets, and laptops.
As the demand for digital content grows, so does the need to process it fast.
The company wanted to process 100M+ A4 files with complex layouts, images, and formatting — without sending costs through the roof. These files also included thousands of product reviews, which our client wanted to organize, analyze, and make searchable.
The goal? Offer curated reviews on consumer products as part of their platform.
Obstacles
That’s when they reached out to Temy
Too expensive to scale
The client found that outsourcing manual processing would cost $0.10 per page, which is financially unsustainable.
Too slow to keep up
A human processes one page for approximately 10 minutes. Processing 100 million pages manually would take…16.6 million hours.
Too complex to automate easily
Pages contained diverse layouts, images, and formatting rules, requiring precise structuring for each device type.
THE CHALLENGE
Temy assembled a dedicated
5-person AI team to handle this project
Our team knows how to make AI work for complex data — both text and images. We showed the client how they could process millions of content pages cost-efficiently with AI.
We kicked off with an evaluation phase
Over two weeks, our business analyst worked with the client to map out their needs, identify key challenges, and create a roadmap for AI integration
We set up a dedicated AI team
We onboarded four AI and data engineers right after the evaluation phase in two days.
We built a PoC
We ran pilot tests, gathered feedback, and fine-tuned the PoC model for accuracy and cost-efficiency. The client’s target price was $0.05/page, but we slashed it to $0.02.
Full deployment in 6 months
In 6 months, the AI was fully operational, processing millions of pages automatically. We continue optimizing as the client scales.
Key project outcomes
80% cost savings
AI-driven automation reduced processing costs from $0.10 to just $0.02 per page, saving $8 million for their target of 100 million pages.
400x faster processing
AI handles 100 pages in 2-3 minutes. A human needs 10 minutes per page.
Easy to search and consume content
AI automatically extracts text, matches images with captions, and formats documents for readability.
We see a hyper-strong difference between what a human and AI can do. We cut processing time by 99% and made correctly formatted content instantly available at scale.”
THE PROCESS
How we automated content processing using multimodal AI or LMM
Processing 100M+ A4 pages with different layouts, images, and metadata required an efficient AI system. It had to extract and format data without driving up costs. Instead of costly AI training, we used multimodal AI to process text and images with “zero-shot prompting,” meaning no pre-training on company data is needed.
What does this mean for our client?
No costly data labeling or long AI training cycles. We ran a quick proof of concept (PoC) showing how AI could extract text, structure metadata, and format content accurately. This gave them a low-risk opportunity to test AI’s impact before scaling. They liked what they saw and decided to continue the project.
Business analyst mapped the client’s needs
Our BA worked with the client to figure out how AI could make their content processing faster and cheaper. They pinpointed the slowest, most error-prone tasks like extracting data from PDFs, making content searchable, and linking images to text.
Our AI team built the model to process text & images at scale
Our four AI engineers built an AI model that processes millions of unstructured pages automatically. It extracts text, detects images, and organizes key metadata (titles, authors, dates). Powered by OpenAI’s GPT-4o, it can handle millions of complex documents.
Temy scaled the team, adding one more AI engineer
As the project progressed, the client needed to speed up implementation. We onboarded a new AI engineer within a week to fine-tune text extraction, metadata tagging, and content structuring.
We achieved unbeatable cost-efficiency
Beyond reducing the per-page processing costs, we optimized the infrastructure for cost-efficiency and scalability:
- Reduced AWS costs by 32x
- Reduced LMM API costs by 30%
- Smart resource allocation helped the client fine-tune AI workloads based on real data, preventing overspending
- AI remains affordable as the solution expands
Our core team continues optimizing the system
Our team of four AI developers and a business analyst deployed the AI solution, automating the processing of millions of pages, which are then fed into the client’s CMS, database, and analytics dashboard. We keep monitoring and improving the system based on client feedback and business needs.
The team’s know-how in how to make it work for you is impressive. We’ve saved tons of money and time by working with Temy
The result
AI automatically formats and structures content, feeding it into a searchable database for instant access. A custom dashboard lets the client search by topic, filter by metadata, and analyze key insights. They don’t need a slow, manual process anymore.
Before
Outsourcing manual processing costs $0.10 per page
Complex layouts, images, and metadata required manual structuring
Limited ability to scale without increasing costs
o real-time insights or analytics
After
AI cut costs to $0.02 per page — 5x cheaper
AI automatically organizes text, images, and metadata for easy searchability
Scalable AI model processes new content instantly without added costs
Custom dashboard for content analysis, instant search, and metadata filtering
Your team’s unfair advantage
Book a free consultation and discover what you can achieve by getting our team to work on your custom software development project.