The Amess Guide to Enterprise AI: Why They Spent 5 Years Building Before Launch

Discover why Amess spent five years developing their AI solution before launch, with insights from CEO Fabrice Deprez on building enterprise-grade AI in an era of rapid deployment.

Written By: supervisor

0

The Amess Guide to Enterprise AI: Why They Spent 5 Years Building Before Launch

The Amess Guide to Enterprise AI: Why They Spent 5 Years Building Before Launch

In an era where AI companies launch products in weeks, Amess took a radically different approach. In a recent episode of Category Visionaries, Fabrice Deprez revealed why they spent five years developing their AI solution before bringing it to market – and why that decision is now paying off in enterprise sales.

Understanding Enterprise Stakes

The stakes in financial crime prevention are too high for minimum viable products. As Fabrice explains, “Compliance is getting more and more important. The regulations are a burden one side, but on the other side it’s a saving. We’ve seen it in Europe, but what have happened last few days with the banks in the US and then moving to Switzerland and getting closer.”

The Development Timeline

Rather than rushing to market, Amess invested heavily in development. “On the program that we are bringing on the market, we are already developing for five years. So the first three years it was purely development and then two years of testing, tuning, until the algorithms were delivering the result that we wanted,” Fabrice shares.

Demystifying AI Development

While many companies position AI as mysterious or magical, Fabrice takes a more grounded approach. “When I discuss with my data scientists, my AI specialists, they say, well, what they did is brilliant, but in the end, it’s not complex. They have just used a huge load of manual work. They have used a lot of computer capacity and coding.”

Building for Real Problems

Their development focused on solving real industry problems. Current systems generate overwhelming false positives – “More than 90% of the work that the people are doing is just a false positive, meaning a work which was unnecessary.” Even worse, banks only catch “10-15-20% of detected cases.”

The Value of Long Development

This extended development period allowed them to achieve dramatic improvements. “We are with our solution, we are able to almost multiply this factor by three four or have in total three multiplied by three better performance of up to 10%, ten times,” Fabrice notes.

Key Elements of Their Approach

  1. Start with Clear Metrics: They focused on specific industry problems with measurable outcomes.
  2. Invest in Testing: Two years of testing ensured the solution worked reliably.
  3. Build Complete Systems: Rather than launching features incrementally, they built comprehensive solutions.
  4. Focus on Results: They prioritized performance improvements over speed to market.

The Results

Their patient approach is paying off. “We already have our first three customers outside the group which is already positive. We have an outlook of being financially independent as of mid of next year. So two years after launch being profitable,” Fabrice shares.

Lessons for AI Companies

Amess’s experience offers valuable insights for companies building enterprise AI:

  1. In regulated industries, reliability matters more than speed to market
  2. Focus on clear, measurable improvements to existing processes
  3. Invest in testing and validation before launch
  4. Build complete solutions rather than minimum viable products
  5. Position AI realistically rather than mysteriously

Their success suggests that in enterprise markets, particularly in regulated industries, the “move fast and break things” approach to AI development may be exactly wrong. Sometimes, moving slowly and building things right is the fastest path to market success.

For founders building enterprise AI solutions, the key question isn’t how quickly you can launch, but how thoroughly you can solve real business problems. As Amess demonstrates, enterprises are willing to buy AI solutions that work – even if they take longer to build.

Leave a Reply

Your email address will not be published. Required fields are marked *

Write a comment...