What we are building right now.
An AI assistant run entirely by text message. Elva keeps track of your calendar, email and reminders, checks in every morning and reminds you in time about what you have promised someone.
Automated forensic analysis of images and documents to detect manipulation and AI-generated content.
When anyone can fake anything, what is actually true?
A few years ago, convincingly manipulating a photo took a graphic designer with months of experience. Today a couple of sentences in a prompt is enough. Modern AI tools can generate photorealistic images from scratch, swap faces, add or remove objects, and create entirely synthetic documents that look genuine down to every detail. What used to be an art form has become the push of a button.
At its core, this is a fantastic technological development. Creativity and the production of visual content have been democratized. But the same capability that lets a small studio create blockbuster effects also lets a fraudster fabricate evidence.
The consequences are already visible. Insurance companies are seeing a sharp rise in damage photos manipulated to inflate claims. Employers and banks receive forged identity documents and payslips. Authorities and the justice system are faced with digital evidence whose authenticity they lack the tools to assess. Journalists and newsrooms encounter synthetic images spread as real news.
The common problem is that the human eye is no longer enough. We are not biologically equipped to see the patterns that reveal modern manipulation. Something that looks right to the eye can be completely fabricated at a level only mathematical analysis can reveal.
Our starting point is that credibility must be verifiable, not just assessed. An answer to whether something is genuine or not is not enough. Whoever receives an analysis needs to understand why, see where the anomalies are and be able to trust that the evidence holds up if it is questioned. Transparency and traceability are as important as the detection capability itself.
We also think about sovereignty. Sensitive material, such as damage photos, identity documents and legal documents, should not have to leave an organization's control to be analyzed. We see that as a given, not an add-on.
We are actively developing a system for automated forensic review of images and documents. The system analyzes incoming material and produces a structured report showing what was examined, how, and what the analysis indicates. The result is readable both for case handlers without a technical background and for experts who want to go deeper.
Right now we are working toward a pilot launch with a first customer in a regulated industry. The goal is a system that can be put into production in a live environment, validated against real cases and then scaled further. We are not looking for volume at this stage. We are looking for the right context to prove the value in.
Do you run an organization where the authenticity of digital material is a real problem? We would be happy to talk.
Contact us