Personalized AI, Creative Work & Knowledge Protocols: Research Brief

Most marketing teams are running AI on generic settings — and the latest research suggests that may be the core problem. A new experiment finds that giving an AI tool a profile of the person it's working with produced measurably more creative and higher-quality marketing campaigns than using a generic AI setup. A second paper proposes a new framework for guiding how AI reasons through expert work — compelling in theory, but not yet tested in practice. In this Research Radar Brief, Dr. Eva Wolf reviews 2 recent AI marketing research papers covering personalized AI collaboration, human-AI creative synergy, and expert-guided AI reasoning frameworks. What you'll learn: - Why a personalized AI collaborator outperformed a generic one in a controlled experiment — and the three mechanisms that explain the gap - How human-plus-personalized-AI teams achieved a synergy effect that beat AI working alone — a result rarer than most assume - What Knowledge Protocol Engineering proposes as a step beyond RAG-based AI — and why it has not yet been validated - How to start loading your AI tools with collaborator context today, before any platform natively supports it Papers covered: 1. Personalized AI Scaffolds Synergistic Multi-Turn Collaboration in Creative Work - Authors: Sean Kelley, David De Cremer, Christoph Riedl (2025) - Type: Preprint (not yet peer-reviewed) - Access: Full text reviewed - Source: https://arxiv.org/abs/2510.27681v2 2. Knowledge Protocol Engineering: A New Paradigm for AI in Domain-Specific Knowledge Work - Authors: Guangwei Zhang (2025) - Type: Preprint (not yet peer-reviewed) - Access: Full text reviewed - Source: https://arxiv.org/abs/2507.02760v1 Full show notes, transcript, and citations: https://bigplans.media/episodes/personalized-ai-creative-work-knowledge-protocols-marketing-2026-06-17 Disclaimer: This is a first-pass research briefing, not a final academic review. Both papers covered are preprints that have not yet been peer-reviewed — findings may change before publication. Dr. Eva Wolf summarizes what the research suggests and where the evidence is limited. Nothing here should be taken as a definitive research conclusion. -- This is a first-pass research briefing, not a final academic review. Read the original papers before making major marketing or business decisions. AI & Marketing Research Radar is produced by BigPlans Media. Subscribe wherever you listen to podcasts.

Thanks for listening to AI & Marketing Research Radar by Big Plans Media.

I’m Dr. Eva Wolf, and I help marketers, educators, consultants, and business owners turn AI marketing research into practical strategy, smarter workflows, and real business opportunities.

More episodes: https://bigplans.media/ai-marketing-research-radar/
Consulting: https://bigplans.media/ai-marketing-consulting/

Big Plans Media — Where Big Ideas Meet Smart Marketing.

Transcript

SPEAKER_00 0:06

You're listening to Evita, an AI-generated research briefing avatar trained on the research framework and methodology of Dr. Eva Wolf, marketing professor, AI researcher, and founder of Big Plans Media. Every day, Avita scans emerging research in AI, marketing, consumer behavior, psychographics, and business strategy to identify the most relevant developments, opportunities, and risks worth watching. These daily radar reports are designed to help busy professionals stay informed without having to read hundreds of research papers themselves. And every Friday, join Dr. Eva Wolfe live for her personally recorded weekly AI Marketing Radar Roundup, where she breaks down the biggest stories, explains what actually matters, and shares practical insights and strategic implications for marketers, educators, entrepreneurs, and business leaders. Now, here's today's radar report. Now, here's today's radar report. Here's the signal I can't ignore today. You open your AI tool, you type a prompt, you get something back that's fine, competent, forgettable, and you think, maybe I need a better prompt, maybe I need a better model, maybe I need to try a different tool. Hmm, what if none of that is the problem? What if the problem is that the AI has no idea who it's talking to? Today's paper points to something that's going to change how I set up every AI session going forward. And I mean every one. We screened 12 papers. One cleared the full text bar and made the radar. Quick caveat: this is a first-pass research briefing, not a final academic review. The paper covered today has full text access. I'll tell you what it suggests, what it doesn't prove, and whether it deserves a deeper read. Okay, let's get into it. Paper one. Here's the business question. If you give your AI a real profile of who you are, your skills, your work style, your personality, does the output actually get better? Or is that just a nice idea that sounds good in a product demo? Most of us assume it helps a little. This study asks whether it helps a lot, and more importantly, why? Here's what the researchers did. Randomized controlled experiment. Three groups, 331 participants. Everyone built a marketing campaign for a fictional startup using an AI assistant. Group one, completely generic AI, no context about the person at all. Group two, partially personalized, some basic context fed in. Group three, fully personalized, built from psychometric surveys and an actual AI-led interview, not a quick form, an interview about how that person works, what they're strong at, how they think. So what happened? The fully personalized group produced better campaigns, more creative, higher quality, better than generic AI alone, and better than partial personalization. But here's the part that stopped me. Huh, it wasn't because the AI gave better individual answers. It was because the whole conversation stayed on track. Without upfront context, multi-turn AI conversations drift. They stop matching what the person actually needs. The AI is guessing the whole way through. With personalization, three specific things improved. Shared memory. The human and AI remembered and reused ideas better across the conversation. Joint attention, they stayed focused on the same goals. And aligned reasoning. They made decisions together for better reasons. In business terms, a generic AI session is you and a contractor who doesn't know your brand, your client, or your standards. A personalized AI session is you and someone who actually read the brief. That is not a minor UX improvement. That is a compounding quality gap, one that grows wider with every turn of the conversation. Yeah, and participants felt it. More confidence, more trust in the AI, rated it more useful overall. So it shows up in the output quality and in the experience of using it. But here's the catch. This was a fictional startup campaign with an online panel, not professional marketers, not real client stakes, not a live agency workflow. The gap in a real professional context might be bigger, might be smaller. We don't know yet. That's the part I keep coming back to. The mechanism, the reason personalization works, is so clearly described here that I believe it even if the exact numbers shift in a real-world replication. Conversations drift without context. That's not a theory. That's what every practitioner already experiences every time they open a generic AI tool and type a cold prompt. Plain English payoff. Spending five minutes telling your AI who you are and how you work will make the whole conversation better, not just the first answer. Okay, here's where this becomes commercially interesting. Money move. Build an AI creative brief for you. A 10-question intake template that any freelancer or agency team member fills out once, then pastes into any AI tool before starting campaign work. Sell it as a template pack. Or go further, build a micro SAS that generates a personalized system prompt from the answers. This paper is your proof of concept. The research is already there. Action step. Before your next campaign session with any AI tool, write three sentences about yourself first. Your skill level on this type of project. What you need help with versus what you'll handle. How you like feedback. Paste that in before your first prompt. That's the minimum viable version of what this study actually tested. Evidence check. This is a preprint, not yet perk reviewed. Sample was a general online panel, not professional marketers. Task was one type of creative work for one fictional client. Treat the direction as reliable. Treat the exact numbers as preliminary. Radar verdicts, read now. The design is genuinely rigorous for this space. Randomized controlled experiment, causal mediation analysis, validated evaluation method. Hold the specific numbers loosely. The mechanism finding is strong enough to act on today. At first glance, these papers look separate, but together they show wait. There's only one paper today. So let me give you the synthesis a different way. This looks like a paper about AI features, better prompting, better tools. But that what it's actually about. It's about conversation architecture. The quality of a multi-turn AI session is determined before you type your first prompt by how much context you've given the AI about who it's working with. Not better prompts, better starting conditions, not more AI capability, more human context fed in up front. That reframes the whole question most teams are asking right now. They're asking which AI tool is best, which model should we use? Which features matter? This paper says you're asking the wrong question. The model matters less than whether the model knows anything about you before the conversation starts. The researchers found three things improved: shared memory, joint attention, aligned reasoning. Hmm. Those aren't AI features, those are collaboration fundamentals. We've spent decades learning how to onboard human collaborators. Brief them properly, tell them who you are, tell them what you need and how you think. We have not done that with AI. We open the tool cold, we type a prompt, and then we wonder why the output feels generic. I'm telling you, the teams that treat AI onboarding as a practice, not just a feature, are going to build a systematic quality advantage over teams that keep starting every session from a blank slate. Not a workflow optimization, a compounding quality gap. And the commercial opportunity hiding in this research is real. Someone is going to build the personalized AI brief layer, the intake, the interview, the system prompt generator that sits in front of every AI creative tool. It could be a template, it could be a SAS product, it could be a premium agency service offering. This paper is the evidence base for that product. It exists now. Use it. Here's the playbook from today. 1. Before your next AI creative session, write a three-sentence brief about yourself, your skill level, what you need help with, how you like to work, and paste it in before your first prompt. That's the minimum viable version of what this study tested. 2. If you run a team or agency, build a short AI onboarding interview into your workflow. Have the AI ask each person about their role and preferences before any campaign work begins. Treat it like a creative brief, but for the human, not the project. 3. If you're evaluating AI tools, stop testing single prompt output quality. Test how the tool performs across a full multi-turn conversation. That's where personalization compounds, and that's the more honest comparison. Evidence check on all of that. Today's paper is a preprint. Strong design, randomized controlled experiment with causal mediation, but not yet peer-reviewed. And the sample was a general online panel, not professional marketers. Right. Use this to decide what to test, not what to blindly build. Links to the paper are in the show notes. Read the original before making major decisions. Want the human expert take? Join Dr. Eva Wolfe every Friday for the AI Marketing Radar Roundup, where she extracts no nonsense, money-making tips, practical strategy, and real business opportunities from the week's research. Subscribe on Apple Podcasts, Spotify, YouTube, and wherever you listen to podcasts. This is Avita for Big Plans Media, and I'll be back in the next radar brief.