📈 How are companies using AI?

➜ AI Soothes Emotions: SoftBank Corp. has developed voice-altering technology that can transform the angry voices of customers into calm ones. Who would have thought that AI can protect employees from customer harassment?
Asahi

➜ Gen AI Assistant for Retail: Michael Kors is the first retailer to integrate Mastercard's generative AI assistant into its website. The tool translates customers’ colloquial language into personalized product recommendations. Results? A 15-20% higher conversion rate than traditional search queries, at least during initial tests.

DigitalCommerce360

➜ AI-Powered Recurring Payments: Trustly, a leader in open banking payments, has launched an AI-powered solution to handle and automate repeat transactions. The goal here is to reduce the 50% abandonment rate at checkout due to cumbersome payment setups. 

Finextra

➜ Money and Tax on Autopilot: Megan Caywood Cooper, former CPO at Barclays and Starling, is launching a new AI wealthtech startup named Caywood. The platform promises real-time financial guidance, including smart financial aggregation, cash flow forecasting, automatic portfolio management, and tax optimization. That’s a lot, let’s see whether Caywood can deliver on its promise. 

Finextra

➜ New Hot AI Tool: Dream Machine by LumaLabs is gaining popularity on social media. The AI model creates high-quality, realistic videos from text prompts or images, with the latter option performing better. It's fascinating to see how these models generate physically accurate and consistent shots. Fun to play with – do try.

Lumalabs

➜ Microsoft Delays “Recall”: an AI-powered feature that tracks computer usage, won’t launch next week. Instead, Microsoft will test it on a smaller group. Official reason for delay: concerns of privacy risks.

Reuters

💡 What are the nerds up to?

➜ Amazon Scrapes GitHub: Amazon wanted to feed its AI model on GitHubs’s data and was definitely not happy with GitHub’s data scraping limits (5k request per hour per account). According to a leaked memo, Amazon found a workaround: they encouraged its employees to create multiple GitHub accounts and share their access credentials. When there's a will there's a way, right?

Dataconomy

➜ Superfast PoCs Built by AI Agents in Minutes: Proofs is a Warsaw-based startup that aims to fix one of the biggest pain points faced by API-first companies today – the slow and expensive process of building proof-of-concept apps for prospects. Proofs has just raised €2.4 million in pre-seed financing. Kudos to founders: Bartek Pucek and Zbigniew Sobiecki! 

Proofs

➜ AI Needs Humans: it turns out that enterprises that are using automated customer support bots and have a high volume of support requests, are not particularly happy about the quality of LLMs. They require 95% accuracy, and Automatic Evals often miss nuances that only human expertise can catch. Paramount - an Open Source package - wants to fix this by incorporating human feedback directly into the evaluation process. 

GitHub

➜ AI Ignores Robots.txt: A user caught Perplexity red-handed on not respecting robots.txt or sending the correct user agent so it can be blocked by servers.

Rknight.me

➜ Where Are My Keys? A Home Assistant user hooked up GPT-4 Vision with their security cameras and can find various items in their home. This is next level stuff. It still can’t look inside your wardrobe, though.
Reddit

➜ AI Search: Scaling Laws Broken? Granting foundation models ‘search’ (the ability to think for longer) might upend Scaling Laws and change AI’s trajectory. A longer read, thought-provoking piece. Recent chess history included.

Yellow Apartment

Thanks for reading!

I share these stories because I believe it's important for all of us to keep up with AI. To support my mission, share this newsletter on LinkedIn.