In our Creator Highlight collection, TDS editors chat with neighborhood members about profession paths, writing, and sources of inspiration in information science and AI. At present we’re happy to share our dialog with Dr. Zhanna Lipenkova.
Dr. Janna Lipenkova is an AI strategist, entrepreneur, and guide writer. The art of AI product development. With a PhD in computational linguistics, she combines deep technical perception with enterprise technique to assist organizations flip AI into tangible outcomes. Janna has based and led a number of ventures on the intersection of language, information, and intelligence. Anacordtargeted on enterprise AI transformation. EXINTELAn AI platform that helps company sustainability. By means of her thought management and consulting work, Janna frequently shapes and refines complete methodologies for the event and integration of AI.
You name the set of psychological fashions that assist groups align on what they construct and why they construct it the “AI Technique Handbook.” Which fashions present the clearest sound in govt rooms, and why do they resonate?
One of many largest challenges within the govt room is communication. When folks speak about AI blocking execution, they imply various things. I take advantage of three psychological fashions to create a structured widespread floor that permits me to maneuver ahead with out excuses or misunderstandings.
I often begin with AI Opportunity TreeThis helps map the panorama for attainable AI use instances. Executives usually come into the corporate with a combination of curiosity and hype that we have to do one thing with AI, however with out a clear view of the place the actual worth lies. The default path most groups take from there may be constructing chatbots, however these initiatives not often get off the bottom (cf. This article). Alternative Tree breaks this sample by systematically uncovering potential AI use instances and offering a structured and goal foundation for prioritization.
When you’re clear on what you are constructing and why, it is time to take the subsequent step. how and please fill in AI system blueprint. This mannequin helps map the information, mannequin, consumer expertise, and governance constraints of the envisioned AI system. It is particularly highly effective in multi-stakeholder environments the place enterprise, information science, and compliance groups want a shared language. Blueprints make AI complexity concrete and iterative. We will draw it, focus on it, and enhance it collectively.
The very last thing I wish to introduce is AI solution space map. This expands the dialog past at present’s dominant applied sciences (primarily large-scale language fashions and brokers) and helps groups take into account the total spectrum of resolution varieties, from traditional ML to hybrid architectures, search programs, and rule-based or simulation-driven approaches. This broader perspective bases us on offering the proper options, not simply stylish ones.
Collectively, these fashions create a journey that displays how profitable AI merchandise evolve, from alternative discovery to system design to steady exploration. These resonate with executives as a result of they bridge technique and execution.
In your writing, area experience is vital for constructing AI merchandise. The place have you ever seen area information change the form of a whole AI resolution, fairly than simply bettering accuracy on the margins?
One vivid instance of how area experience utterly reshaped an answer was a logistics undertaking that was initially initiated to foretell delivery delays. As soon as area specialists got here on board, we reframed the issue. Delays are usually not random occasions; they’re a symptom of deeper enterprise dangers, resembling provider dependencies, regulatory bottlenecks, and community vulnerabilities. We “AI specialists” couldn’t discover these patterns.
To include information on this space, we prolonged the information layer past transit time to incorporate provider threat alerts and dependency graphs. AI architectures have developed from single predictive fashions to hybrid programs that mix prediction, information graphs, and rule-based reasoning. The consumer expertise has been expanded from reactive delay predictions to threat eventualities with instructed mitigations which can be extra actionable for specialists.
Finally, area information not solely improved accuracy, however redefined the issue, system design, and the worth acquired by the enterprise. This has turned AI fashions into true choice assist instruments. Since that have, I at all times advocate for area specialists to be concerned within the early levels of AI initiatives.
Along with contributing to TDS, he has additionally written the next books: The Art of AI Product Development: Delivering Business Value. What was a very powerful level that modified your personal method to constructing AI merchandise (particularly, one which stunned you or challenged your earlier beliefs)?
Scripting this guide allowed me to mirror on each little bit of theoretical information, sensible expertise, and my very own beliefs and construction them right into a reusable framework. As a result of books want to stay related over time, we additionally needed to differentiate between fundamentals on the one hand and hype on the opposite. Listed here are some issues I discovered myself.
- startingI discovered methods to discover enterprise worth in know-how. Usually we vacillate between two extremes. Both we chase AI for AI’s sake, or we rely solely on user-driven discovery. Within the former case, you aren’t creating any actual worth. Within the second case, nobody is aware of how lengthy we must look ahead to the “excellent” AI downside to look. In actuality, the candy spot lies someplace in between. It is about harnessing the distinctive strengths of know-how to unlock worth that customers can really feel however cannot essentially articulate. We all know it from nice innovators like Steve Jobs and Henry Ford, who created radically new experiences earlier than clients requested for them. However to drag this off requires a magical mixture of technical experience, braveness, and instinct about what the market wants.
- Quantity 2I spotted the worth of consumer expertise within the success of AI. Many AI initiatives fail not as a result of the fashions are weak, however as a result of the intelligence just isn’t clearly communicated, defined, or made usable. The success of an AI product is determined by how intuitive customers can use its options and the way a lot they will belief its outcomes. Whereas scripting this guide, I reread design classics like Don Norman’s The Design of On a regular basis Issues and was consistently asking myself – How does this apply to AI? I believe we’re nonetheless within the early levels of a brand new UX period. Chat is a crucial ingredient, nevertheless it’s positively solely a part of the whole equation. I am actually excited to see the event of recent consumer interface ideas like generative UX.
- thirdAI programs should evolve by way of cycles of suggestions and enchancment, and the method is rarely ending. That is why I take advantage of the Dervish metaphor within the guide. Which means spinning, refining, and frequently studying. Groups that grasp early launch and steady iteration are likely to ship much more worth than groups that look ahead to the “excellent” mannequin. Sadly, we nonetheless discover that many groups wait too lengthy to offer an preliminary baseline and do not spend sufficient time on iterative optimization. These programs could also be put into manufacturing, however they in all probability will not be adopted and can be shelved as one other AI experiment.
For groups releasing AI options within the subsequent quarter, what habits would you advocate to maintain them targeted on delivering actual enterprise worth fairly than chasing hype? And what are the important thing pitfalls to keep away from?
First, grasp the artwork of repetition, as talked about above. We are going to ship early, however we are going to achieve this responsibly. Launch one thing helpful sufficient to earn customers’ belief, after which frequently enhance it. Each interplay brings new information and each suggestions is a brand new coaching sign.
Second, have a broader perspective. It is easy to have a imprecise imaginative and prescient of the most recent LLM or mannequin launch, however actual innovation usually comes from how applied sciences like search, inference, analytics, UX, area logic, and extra come collectively. We design our programs modularly to permit them to scale and repeatedly monitor AI options and developments that may enhance our programs (see additionally What’s coming subsequent) AI radar).
Third, take a look at early and sometimes with actual folks. AI merchandise stay or die relying on how people understand and use them. Inner demos and mock assessments cannot exchange the messy and shocking enter and suggestions you get from actual customers.
Longer texts (books, in-depth content material) keep away from hype and deal with offering worth to the group. What’s your method when selecting subjects and does writing about these subjects enable you to perceive them extra deeply?
Writing has at all times been a means for me to suppose out loud. I take advantage of it to study, course of complicated concepts, and generate new concepts. I often observe my instinct and write about approaches that I really consider in and that I’ve seen work in actual organizations.
However, my firm has somewhat little bit of its personal “secret sauce.” Through the years, we’ve got developed AI-driven programs to observe new developments and improvements. We offer this to some chosen clients in industries resembling aerospace and finance, however after all we additionally use it for our personal functions. By combining information and instinct, you’ll find subjects which can be related now and more likely to be related not simply in just a few months, however in just a few years.
For instance, firstly of 2025, Report on enterprise AI trendsand nearly each theme from it turned out to be very related all year long. My writing is due to this fact intuitive and private, but in addition evidence-based.
To study extra about Janna’s work and get the most recent articles, observe her on TDS. substackor linkedin.
