Hi! It’s time for another issue of the Human in the Loop newsletter. This time we’ll cover Amazon’s AI shopping assistant, which reportedly doubled conversion rates over Black Friday. It’s one of the first major, public examples of a chatbot directly moving the needle on revenue at scale.
The conversation is shifting from "does this drive engagement?" to "how much revenue can this generate?" This is the blueprint for any online retailer looking for a new growth lever.
Topics of the day:
Amazon’s AI assistant proves its ROI
DeepSeek’s powerful new open-source models
Runway and Kling push video AI forward
How a reader built an AI college advisor
The Shortlist: Accenture rolls out ChatGPT Enterprise and custom agent programs, Lux comes out of stealth with a screenshot-trained model that outperforms major labs on benchmarks, and Nvidia invests $2B in Synopsys to accelerate chip design with AI.
Amazon Proves AI's Value in E-Commerce
What’s happening: Amazon's AI shopping assistant, Rufus, saw a massive surge over Black Friday, with customer sessions using the tool driving sales at more than double the rate of sessions without it.
In practice:
This is one of the clearest examples yet that conversational AI directly drives revenue in e-commerce, not just engagement.
AI assistants can automate product discovery and comparison, acting as a scalable personal shopper for millions of users at once.
Any business with a large catalog can use a similar AI guide to reduce choice overload and guide customers to the right purchase.
Bottom line: Guiding customers with AI is becoming a clear lever for growth. Expect to see AI shopping assistants become standard for any serious online retailer.
DeepSeek challenges GPT-5 with open-source powerhouse
What’s happening: Chinese AI lab DeepSeek just released two open-source models that rival top models like GPT-5 and Gemini 3 Pro in performance but cost a fraction of the price to run.
In practice:
You can build advanced AI features for data analysis or content creation without the premium API costs of major labs.
The open-source license allows your team to fine-tune the model on your company’s data, creating highly specific and custom AI tools.
This levels the playing field, giving smaller teams access to near-frontier AI to build competitive products and find new growth levers.
Bottom line: High-performance AI is becoming a commodity, and it's happening faster than anyone expected. This is a clear signal to re-evaluate your AI budget and your reliance on expensive, closed-source models.
AI Video gets a major upgrade from Runway and Kling
What’s happening: The AI video space just got a major upgrade with Runway releasing its new Gen-4.5 model, now topping leaderboards, while competitor Kling launched an all-in-one system for both generating and editing video.
In practice:
You can generate high-fidelity video ads or social content in minutes, letting you test more creative concepts without the cost of a full film crew.
Creative teams can use these tools for pre-visualization, quickly mocking up cinematic scenes to get stakeholder buy-in before a big production budget is approved.
Unlike older tools, new models from Kling let you fine-tune video assets after they’re generated, allowing you to swap a character’s outfit or remove background objects with a simple text prompt.
Bottom line: The barrier to entry for professional-grade video is collapsing, giving smaller teams and solo operators powerful new creative leverage. These tools are quickly becoming practical assets in the marketing and content production workflow.
What’s happening: A reader from The Rundown community shared how they built a custom "College Compass" AI to help their son manage the entire college application process, turning a stressful ordeal into a structured, data-driven system.
I love this because it’s a perfect example of applied AI for a real-world, high-stakes project.
In practice:
Think of this as a blueprint for personal project management, using an AI to track deadlines, research, and tasks for any complex goal, from a product launch to a home renovation.
This highlights the power of custom AI agents; instead of a generic app, you can build a tool that understands your specific context, data, and objectives.
The AI does more than just organize, it provides actionable recommendations by calculating admission probabilities and suggesting best-fit scholarships.
Bottom line: This is a fantastic example of moving beyond generic prompts to build a personalized AI operating system. You can create a strategic partner that helps automate workflows and guide decisions for your most important goals.
The Shortlist
Accenture is partnering with OpenAI to provide ChatGPT Enterprise to tens of thousands of its consultants and launch a program to help clients deploy custom AI agents.
Lux emerged from stealth with a model it claims beats OpenAI and Anthropic on key benchmarks by training on screenshots and user actions instead of just text.
Nvidia invested $2B in chip design software maker Synopsys, aiming to use AI and accelerated computing to speed up complex product engineering processes.
This newsletter is where I (Kwadwo) share products, articles, and links that I find useful and interesting, mostly around AI. I focus on tools and solutions that bring real value to people in everyday jobs, not just tech insiders.
Please share any feedback you have either in an answer or through the poll below 🙏🏽
