HighPots - digital

KI im eCommerce verändert mehr als nur den Online-Shop.

Mit Agentic Commerce rücken KI-Systeme näher an den eigentlichen Kaufprozess heran. Entscheidungen werden vorbereitet – teilweise übernommen – und der Wettbewerb verschiebt sich stärker in Richtung Daten, Prozesse und Verfügbarkeit.

Für Onlinehändler ist das keine kurzfristige Mode, sondern eine strukturelle Entwicklung, auf die man sich vorbereiten sollte – nüchtern, schrittweise und ohne Hype.

Im Blogartikel ordnen wir ein, was Agentic Commerce und das Universal Commerce Protocol (UCP) für Shopbetreiber bedeuten und warum ein pragmatischer Einstieg über MVPs sinnvoll ist.
https://highpots.com/blog/ki-im-ecommerce-agentic-commerce-ucp/

̈nstlicheintelligenz

KI im eCommerce: Wie Agentic Commerce den Onlinehandel neu ordnet – und was das für Shopbetreiber bedeutet Warum sich der Onlinehandel gerade grundlegend verändert KI im eCommerce wird oft auf Chatbots, Produktempfehlungen oder automatisierte Texte reduziert. Das greift zu kurz. Der eigentliche Umbruch beginnt dort, wo KI nicht mehr nur berät, sondern handelt. Bisher lief Onlinehandel so: Menschen such...

RiskProphet AI: Die HighPots Insurance AI "RiskProphet AI" geht bald als Produkt an den Start. Jeder einzelne Reisende erhält nun einen individuellen Versicherungstarif, der genau die Risiken und Bedürfnisse widerspiegelt. Eine faire Behandlung...
https://highpots.com/case-studies/kuenstliche-intelligenz-versicherungsbranche/

Student job - learning the differences in generative language models: a) building a military strategy AI (see links at the end of the article for code and data) and b) transforming that military strategy into an intelligent strategy enterprise consultancy (training data modification required).

After the researchers from Georgia Institute of Technology, Stanford University, Northeastern University and the Hoover Wargaming and Crisis Simulation Initiative analyzed the escalation risk of using AI in a military cooperation, we recreated the simulation. And then our 2 students Tobias and Michael used "the same" code to build a strategy consultancy for companies. Of course, we changed the underlying data. "We collected all the company data from HighPots that we could find quickly. E.g. from the areas of finance, accounting, data protection, IT and data security, sales, marketing, etc." said Tobias. We then had the AIs compete against each other with the different language models (GPT-4, GPT-3.5, Claude 2.0 and Llama 2).

Of course, the AI is highly simplified, but the results are still good. "Good" means that creative ideas are generated that even experienced strategy consultants would not have thought of. In addition, complex logical steps were generated that build on each other and work towards a long-term goal. Here, too, the Llama model from Facebook adhered much better to laws than the GPT models.

"We will not put Tobias and Michael's bachelor theses in the drawer, but will try to improve the AI and add another language model" says our Head of R&D Rudi. He also says: "Tobias and Michael can then come back to HighPots for their master theses to continue their work and also continue working on it as working students". If that's not a great offer. We'll see what the future holds.

If you are interested in such solutions, please let us know. We have been building expert systems based on neural networks since 1999 and have successfully completed more than 20 AI projects based on generative AI since the ChatGPT AI boom. We are only as far away from you as your phone or computer is from you. 😊

As promised, below are the sources for the military strategy AI based on different language models.

Procedure and preliminary work:
https://lnkd.in/d-ivtiYV

Code:
https://lnkd.in/dHfHF6wc

Data:
https://lnkd.in/drbnUYjf