Log in or create a free Rosenverse account to watch this video.
Log in Create free account100s of community videos are available to free members. Conference talks are generally available to Gold members.
Coffee with Lou #4: Taking a Peek Under the Rosenbot's Hood
Summary
What happens when you cross an eager librarian, a happy puppy, and 800 UX experts? You get the Rosenbot—Rosenfeld's new GPT-4 level chatbot, trained on our books and hundreds of hours of conference and community call recordings. What went into creating the Rosenbot? Lou is joined by SimplyPut's Peter van Dijck, an old friend from the IA community and the chief architect of the Rosenbot. If you're beginning your journey into developing generative AI products, you'll want to join Lou and Peter to learn from their lessons, ask questions, and share your own thoughts on AI's role in making curated content more useful and impactful.
Key Insights
-
•
The Rosen bot uses curated, high-quality content from Rosenfeld Media to avoid common pitfalls of enterprise chatbots operating on messy internal data.
-
•
Significant project resources, around 30-50%, must be devoted to evaluation combining user tests, expert reviews, and repeatable automated metrics.
-
•
The chatbot intentionally avoids summarizing entire books to prevent cannibalizing book sales, instead acting as an entry point to authors’ original work.
-
•
Users and researchers often don’t know how they will use AI tools initially; the use cases evolve through interaction with the technology.
-
•
Building AI conversational agents requires embedding follow-up questions and memory simulation into the interaction design beyond a basic large language model.
-
•
Semantic search and topic analytics behind the scenes enable insights about conversation content trends and authors' influence in real time.
-
•
Balancing fresh, current content with older foundational material is challenging since methodologies and societal norms continually evolve.
-
•
There is a tension between jumping in to build and learn quickly versus cautious, deliberate product development; both parties in the talk appreciate different approaches.
-
•
Interactive UI elements like expandable answers and filters for broadening or narrowing responses are planned to enhance the conversational interface.
-
•
The Rosen bot acts as a creative companion and research assistant, helping accelerate idea exploration and uncover relevant historic knowledge efficiently.
Notable Quotes
"If you don’t jump in and learn, you’ll spend your time reading stuff on it and not really understand the tech."
"People forget the expertise every seven years and have to learn everything again—this bot helps make that expertise accessible."
"The content behind this bot is curated and high quality, unlike the typical confusing, outdated information in corporate repositories."
"We devote 30 to 50 percent of the project time just on evaluation with real users and experts to ensure quality and safety."
"The bot doesn’t summarize books because we’re a publisher and don’t want to cannibalize book sales—it points users to the original sources."
"We’re building in the attribution of ideas to authors and speakers, emphasizing the humans behind the expertise."
"Chatbots aren’t just the language model—they have structure around intent classification, safety checks, and session memory simulation."
"Sometimes you have to build stuff and learn from it rather than wait for the perfect moment or tools."
"AI in design can act as a creative companion, helping explore ideas and analyze past research more efficiently."
"The product is imperfect, probably about 60 percent there, but if we get to 80 percent it will be a compelling tool for UX content."
Or choose a question:
More Videos
"Without these rules and conventions, we wouldn’t be able to improvise."
Jim KalbachJazz Improvisation as a Model for Team Collaboration
November 6, 2017
"People saying you should write a book is nice, but it’s not the real reason you should write one."
Louis RosenfeldCoffee with Lou: Should You Write a (UX) Book?
March 7, 2024
"Propose new IC roles like a design problem: identify the need, gather research, and try it as an experiment with retrospectives."
Catt Small Micah Bennett Brian Carr Jessica HarlleeWhat's Next for ICs: Exploring Staff and Principal Designer Roles
February 22, 2024
"We are really good at finding problems, but the opportunity solution tree helps pick the best ones to focus on."
Marieke McCloskeyUser Science: Product Analytics & User Research
March 11, 2021
"Processing limits and licensing terms currently restrict how much video and audio these AI tools can handle."
Llewyn Paine[Demo] Deploying AI doppelgangers to de-identify user research recordings
June 5, 2024
"If you cannot remove a treatment, it may not be a true cause."
Joshua NobleCasual Inference
October 6, 2023
"If the scale delivers bad news, we jump on and off to check; if it delivers good news, we accept it quickly."
Sara LogelYour Colleagues are Your Users Too
March 29, 2023
"Enterprise is kind of sexy because big, hairy, hard problems of scale and distribution live at the enterprise."
Bria Alexander Louis RosenfeldWelcome
January 8, 2024
"Accessibility isn’t a checkbox you run automated tests for; you must involve people with disabilities to benchmark success."
Sam ProulxOnline Shopping: Designing an Accessible Experience
June 7, 2023