Nikolina Belic

Writing tickets, running standups, maintaining roadmaps, and translating between development and business are coordination tasks.

• Product work begins earlier, when the problem is vague, the opportunity attractive but unproven, the data conflicting, and no one can yet say what to build.
• The value lies in framing the problem, naming the riskiest assumption, making trade-offs explicit, deciding what not to build, and committing without perfect information.

One post lists tasks commonly mistaken for product management and argues that AI can accelerate artifacts but cannot replace judgment. Another makes the same point, defining the role as reducing ambiguity rather than managing execution.

Read original · Original 2

Building now feels fast and cheap, which can produce more output and less learning.

• AI does not create clarity; it exposes its absence.
• The durable human questions are whether this is the right problem, whether AI is the right mechanism or just the obvious one, what happens when the system behaves unexpectedly, and whether the economics work at scale.

Belic describes product maturity as knowing when to slow down while everything speeds up and when the right answer is “not this, not yet.” In a separate post, they note that AI can now handle much of the translation between technical and business work, but judgment remains harder to access.

Read original · Original 2 · Original 3

Belic’s method is to ask what gets better for the business and the customer, whether that justifies this solution, a different one, or no new build, and whether people will change how they work to use it. A technology request describes what someone wants built; an investment thesis explains why it deserves money, people, and change.

At a watch retailer, a CRM quietly ran six parts of the business: retail, service, inventory, quality control, repairs, and customer operations.

• A customer conversation could concern a delayed order, a repair, or a service need, each involving a different part of the business.
• Belic says taking the request literally could have led to building the right feature in the wrong place.
• The practical test is to write one sentence: what gets better, for whom, and what would make you decide not to build it.

Read original

That person knows which client needs reassurance, which exception to make, and the order in which work really happens—not the order written in a manual.

• Software can carry decisions already made, but it cannot create undocumented judgment.
• Building before mapping that knowledge can automate the wrong thing or just the paperwork around the real decision.

The recurring diagnostic question is what the best person in the business knows that no manual captures. Belic frames this as a business-discovery problem that must be solved before it becomes a technical one.

Read original

It starts with customer conversations, moves to a small, testable prototype, and then asks whether people will pay. The sequence puts evidence of the problem and willingness to pay ahead of a large build.

Belic tells the story of a cousin who inherited an old vulcanizer business with hundreds of customers but no way to track loyalty.

• After talking with other local shop owners and finding the same problem, they used AI to build a tiny clickable prototype and asked owners whether they would pay and how much, requiring payment to secure a spot.
• The framework’s third step, Dollars, is also called Dinars in Serbia.

Read original

A stronger portfolio explains how the problem was framed, which constraints applied, which options were considered and discarded, what was tested and failed, what changed as a result, and which trade-offs were made when there was no right answer. Hiring managers look for judgment rather than polish, and self-explanatory case studies matter more than the portfolio artifact itself.

A portfolio showing only what a team shipped says little about the person; a record of decisions shows how they think. Belic applies the same standard in a course focused on case studies rather than portfolio creation.

Read original

Breaking into product management for the first time is easier than it used to be because AI can handle much of the technical-business translation that once took years to learn.

• Finding a new senior product role is harder because there are fewer openings, more competition, and startups that need product leadership but cannot afford senior salaries.
• In both cases, the differentiator is judgment: business understanding, reading a room, noticing a broken process before the data confirms it, and earning trust.

Belic describes a mentee with 15 years outside tech who is starting from zero on the technical side but already has those judgment skills. The point is not that AI replaces product knowledge; it is that speed has become accessible while judgment has not.

Read original

It answers a newcomer’s question: are they another delivery-process teacher, or do they focus on judgment before build?

Nikolina Belic separates coordination from product leadership.

• The post lists what many people think product managers do — write tickets, run standups, maintain roadmaps, translate between dev and business — and calls that delivery management.
• Product work, they argue, starts when the problem is vague, the opportunity unproven, the data conflicting and no one can say what should be built.
• The practical move is to reduce ambiguity: frame the real problem, identify the riskiest assumption, make trade-offs explicit, decide what not to build and commit without perfect information.
• The post closes by saying AI can accelerate artifacts but cannot replace judgment.

For brands planning a campaign

For a product leadership workshop, this is the clearest statement of the decision-first premise.

• It supports an opening module where teams bring a real ticket or request and rewrite it as a problem statement, riskiest assumption and explicit “not yet” decision.
• It also cautions against positioning the workshop as AI tool training.

A reader considering Nikolina Belic for a workshop can see their view goes beyond a tool demo.

The post argues that AI has absorbed much of the execution work that once signaled product manager value: writing PRDs, running research, synthesizing feedback and translating requirements.

• What remains is judgment.
• Nikolina lists four human questions: Is this the right problem?
• Is AI the right mechanism or just the obvious one?
• What are the failure modes?
• What are the economics at scale?
• The memorable line is that product managers who matter can “slow down when everything is speeding up” and say “not this, not yet.”

For brands planning a campaign

This is direct evidence for an AI-driven discovery and decision-making workshop.

• It gives the session a structure: use AI to speed up evidence gathering, then set decision gates for problem choice, mechanism choice, failure modes and scale economics.
• The line about slowing down can anchor the case for pausing before build.

The recurring argument is that tools expose weak thinking rather than fix it.

The post draws a line between shipping more and learning less.

• Nikolina says AI makes building fast and cheap, so teams build more but can learn less about what not to ship.
• They pose three replacement questions — why build it, can we sell it, should it be used — and put them at the core of product thinking.
• AI does not create clarity, the post argues; it exposes the lack of it.
• It also refers to a five-week program focused on decision-making, not the tool.

For brands planning a campaign

Use this to position the workshop around decision quality rather than AI adoption.

• Teams could bring an AI-assisted prototype and answer the three questions before deciding whether to continue.
• The post also supports a module on stopping rules: its value lies in knowing what not to test or ship.

It is the best starting point for readers seeking a practical discovery example rather than a general opinion.

This is a concrete client-style story.

• A request to add AI to a CRM at a watch retailer turns out to involve a CRM quietly running six parts of the business: retail, service, inventory, quality control, repairs and customer operations.
• A customer conversation can mean waiting for an order, checking a repair or needing service, each depending on a different part of the business.
• Nikolina says taking the request literally could have put the right feature in the wrong place.
• Before scoping, the post asks what gets better for the business and customer, whether that justifies this solution or no new build, and whether people will change how they work.
• It contrasts a technology request with an investment thesis and suggests writing one sentence: what gets better, for whom, and what would make you decide not to build it.

For brands planning a campaign

This is the strongest story for a B2B SaaS workshop: a real request, hidden system complexity and a decision framework before scoping.

• Use the watch-retailer case as a live exercise: map the request, identify the business areas affected and draft a one-sentence investment thesis.
• It also gives the facilitator a tested example of an AI feature request that should not be taken literally.

It shows why Nikolina Belic’s product advice starts with business knowledge, not software features.

Nikolina Belic describes a pattern from working with over 90 companies: the process is documented, but one senior employee is the actual system.

• That person knows which client needs reassurance, which exception to make and the real order of operations, not the manual’s.
• The post says software can only carry what has already been decided.
• Building before mapping what the best people know can automate the wrong thing or automate only the paperwork around the real decision.
• The closing question asks whether a business owner could write down what their best person knows that is not in any manual.

For brands planning a campaign

For an AI discovery workshop, this adds tacit-knowledge mapping before AI build decisions.

• Teams could interview a senior operator, map exceptions, judgment calls and workarounds, then decide which parts are safe to automate or support.
• Its business-owner framing will resonate with senior product and operations leaders.

It explains how they teach the path from idea to paid demand.

The post tells the story of a cousin who inherited an old vulcanizer business with hundreds of customers but no way to track loyalty.

• Nikolina talked to other local shop owners, found the same issue, used AI to build a tiny clickable prototype and asked owners whether they would pay and how much, with payment to secure a spot.
• The framework is Demand, Delivery, Dollars — or Dinars in Serbia.
• It is a compact account of testing demand before building a full product.

For brands planning a campaign

This is useful if the workshop audience includes founders or teams testing new products.

• It demonstrates a fast demand test: customer conversations, a small AI-built prototype and a willingness-to-pay question.
• For a B2B SaaS audience, use it as a contrast case — the same validation principle in a different market — rather than the main example.

Nikolina Belic argues that most product manager portfolios read like product catalogs: screenshots, roadmaps and UI mockups show outputs, not judgment.

• Strong portfolios show how the problem was framed, the constraints, the options considered and discarded, what was tested and failed, and the trade-offs made when there was no right answer.
• Hiring managers skim for judgment, not polish.
• Their course focuses on self-explanatory case studies rather than the portfolio artifact itself.

For brands planning a campaign

This is less directly relevant to a B2B SaaS product leadership workshop, but it shows how Nikolina Belic teaches decision-making as a portable skill. A workshop exercise could ask participants to present a decision record rather than a feature demo, reinforcing the same judgment-first message.

The post distinguishes two market conditions: breaking into product management is easier than it used to be, while finding a new senior role is harder.

• Nikolina points to fewer senior roles, more competition, and startups that need product leadership but cannot pay senior salaries.
• They describe a mentee with 15 years outside tech who is starting from zero on the technical side but already has business understanding, room-reading, pattern recognition for broken processes, and trust-building.
• The post says AI can now handle much of the translation layer, but judgment remains unevenly accessible.

For brands planning a campaign

For a workshop aimed at senior product leaders, the pressure is real, and the value proposition should be sharper judgment rather than another execution skill. The post also supports a workshop segment on how experienced people from outside tech can build product judgment without pretending to have a technical background.

In their view, AI can speed up execution, but the harder questions remain human ones: Is this the right problem?

• Is AI the right solution?
• What could go wrong, and does it make economic sense?
• Belic’s take on product judgment in the age of AI and why tickets are not the job make this distinction concrete.

For brands planning a campaign

A workshop on product judgment fits better than a tool demo: have participants name a risky assumption, weigh trade-offs and explain what they would choose not to build.

Examples: Source 1 · Source 2

In their account, the risk is producing more while learning less.

• AI is most useful when it helps teams prototype, explore blind spots or speed up discovery, while people remain responsible for deciding what is worth testing.
• Belic’s post on AI and product thinking makes the point through a contrast: tools can accelerate work, but judgment includes knowing what not to test.

For brands planning a campaign

For an AI-focused session, use AI to challenge a product bet or surface assumptions, then ask participants to decide which evidence would actually change their minds.

Examples: Source 1 · Source 2

Belic’s point: taking a request literally can put the right feature in the wrong place.

• Before scoping, ask what improves for the business and customer, whether the solution is justified, and whether people will change how they work.
• Their practical test is to state what gets better, for whom, and what would make the team decide not to build.
• The watch-retailer example and thesis test show how to turn a technology request into an investment decision.

For brands planning a campaign

A live workshop can have teams rewrite a feature request as an investment thesis, then identify what evidence would support a build, a different solution or no build.

Examples: Source 1

An experienced employee may know which exception to make, who needs extra reassurance and how work actually flows.

• Software can encode decisions, but it cannot make undocumented judgment appear.
• Their question before building is simple: what does the best person in the business know that no manual captures?
• Belic’s post on knowledge trapped in people frames this as a business-discovery problem, not just a technical one.

For brands planning a campaign

In an automation or transformation workshop, map where work depends on individual expertise before choosing processes to automate.

Examples: Source 1

In their example, local shop owners confirmed a customer-loyalty problem; they then used AI to make a small clickable prototype and asked potential customers whether they would pay.

• The sequence puts confirmation of the problem and willingness to pay ahead of a large build.
• Their career advice carries the same emphasis on judgment: a PM portfolio should explain the problem, constraints, options, tests and trade-offs—not just display polished features.
• See the 3D Framework example and what Belic says a PM portfolio should show.

For brands planning a campaign

For an early-stage product session, participants can interview for demand, sketch a testable prototype and define a payment signal. For a PM-career audience, the work can end with a case study that makes the decisions visible.

Examples: Source 1 · Source 2

View example

Both put willingness to invest at the centre of the decision: Dosapati describes tying AI20 Labs’ return to what it produces for a small number of established businesses, and urges founders to validate ideas with prospective customers before committing.

Dosapati writes as a builder and investor, offering to prototype, validate and market-test ideas on clients’ behalf.

• The practical guidance comes alongside an offer of services.
• Belic stays focused on the decision: what to build, what to leave alone and how a product manager should argue the case inside a team.

Read original · Original 2

Belic makes a parallel argument about technology requests: a request says what someone wants built, while an investment thesis says why it deserves the money, people and change.

Abramson writes about winning professional-services business, covering pitches, client origination and partner-led growth for firms such as law, accounting and consulting practices. Belic writes about product work, discovery and judgment for product teams rather than firm leaders.

Read original · Original 2

Both treat clear decisions and judgment as the real constraint, rather than the toolset, and track how AI is changing where a buying or scoping decision starts.

Buchinger writes for ecommerce and direct-to-consumer operators, with concrete material on segmentation, messaging and platform risk, such as Amazon-only brands as shopping moves into AI chatbots. Belic’s examples focus on product discovery, business process and product careers, with less attention to retail channels or campaign execution.

Read original · Original 2

Dosapati's posts describe a practice that takes an idea from proof of concept to validation with prospective customers and then to a market test before the buyer commits.

• They also list shipped work including inference infrastructure, a quoting and configuration workflow collapsed into a single dashboard for a manufacturer's distributor network, and natural-language analytics that let non-technical staff query their own sales and margin data.
• For a B2B SaaS team stuck in analysis, that build-then-validate approach can be more useful than a half-day on scoping questions.

The emphasis shifts from deciding what deserves to be built to building it.

• Belic's material focuses on stopping work: writing a one-sentence investment thesis, asking whether a request justifies a solution at all, and answering "not this, not yet."
• A build-first session leaves less room for that judgment and may produce more output than insight.
• The two posts reviewed describe services and shipped work rather than teaching, leaving less worked reasoning to turn into exercises.
• Before committing budget, agree on scope, deliverables and what "validation" means.

Examples: Source 1 · Source 2

Abramson's posts describe serving on a City pitch panel where four firms were almost interchangeable and the winning move was thinking harder about the specific client than anyone else in the room.

• The second post argues that leaders should deliberately be the least experienced person in the conversation and hire people better than themselves.
• A firm running a client-origination or partner-development session would get more from that than from product discovery material.
• The underlying theme—that quality of thinking, not credentials or tools, decides outcomes—closely aligns with Belic's argument, though the audience differs.

Belic's posts contain no product management or AI-in-product content, so the discovery-and-decision workshop they teach would give way to positioning, pitching and leadership. Product teams would leave with commercial instincts rather than a method for scoping AI work; the watch-retailer case and four AI questions have no equivalent here.

Examples: Source 1 · Source 2

Buchinger's posts argue that better tools do not raise sales without a named decision owner; traffic and messaging matter far more than cosmetics; segmentation is not optional at scale; and AI shopping inside Gemini and instant checkout elsewhere make single-channel dependence harder to justify. A commerce team working on channel mix and message ownership would find that closer to its actual problem than an AI discovery workshop.

This audience sits outside product leadership and largely outside the AI-decision framing, so the session would not teach product managers how to decide what AI work to fund.

• Belic's four questions and the request-to-thesis exercise described in the posts would give way to commerce execution advice.
• The posts offer broad lessons and market commentary rather than a workshop method; buyers should agree on the session format in advance.

Examples: Source 1 · Source 2