AI in Drug Discovery: Are We Moving Beyond the Hype?

By Sazzad Karim

AI is becoming an increasingly important part of pharmaceutical research. But its real potential goes beyond predicting molecules or accelerating computational workflows.

A recent example is the collaboration between AstraZeneca and insitro, announced on 15 October 2026, to apply AI to identifying disease-driving biological mechanisms and discovering new therapeutic targets.

The collaboration focuses on complex cardiometabolic diseases, including MASH and cardiomyopathies. It combines AI-driven analysis of biological data with pharmaceutical research and development expertise.

The agreement includes $40 million upfront and up to approximately $1.46 billion in potential milestone payments.

My Perspective

The real test of AI in drug discovery is not how many molecules or targets an algorithm can predict.

It is whether those predictions can be translated into experimentally validated discoveries, promising drug candidates and, ultimately, meaningful clinical outcomes.

Scientific validation, differentiated technology and collaboration between AI developers and pharmaceutical researchers will be essential.

AI may accelerate discovery. But biological understanding and experimental evidence remain fundamental.

The key question is not simply what AI can predict, but what science can validate.

Source

AstraZeneca — Strategic collaboration with insitro, 15 October 2026.

3F: Family. Fitness. Freedom.

A Different Perspective on Wealth and Entrepreneurship

By Sazzad Karim

3F: Family. Fitness. Freedom.

What does real wealth mean to you?

In The Millionaire Fastlane, MJ DeMarco introduces an interesting concept: the three F's of true wealth.

Family: Having meaningful time for the people who matter most.

Fitness: Maintaining the physical and mental well-being to enjoy life.

Freedom: Having greater control over how we spend our time, make decisions, and shape our future.

This perspective becomes particularly interesting when we think about careers, business, and entrepreneurship.

A stable corporate career can provide financial security, professional growth, and valuable experience. Entrepreneurship, on the other hand, offers the possibility of building something independently and creating long-term freedom.

But there is another side to that freedom: Responsibility.

Running a business means accepting uncertainty, making difficult decisions, managing financial risks, and taking responsibility for outcomes. In the early stages, it may actually demand more time and effort than a conventional job.

So perhaps the real objective is not simply to escape employment or become an entrepreneur.

It is to build a sustainable professional and financial foundation that eventually allows us to protect what matters most: our relationships, our health, and our freedom.

Financial success is important. But what that success allows us to do with our lives matters just as much.

Inspired by “The Millionaire Fastlane” by MJ DeMarco.

AI in Marketing: Six Key Lessons from Professor Philip Kotler’s Live Session

Why Human Understanding, Customer Trust, and Marketing Fundamentals Still Matter in the Age of AI

By Sazzad Karim

Session date: 10 September 2026

Some moments have a strange way of taking us back to where we started.

Many years ago, when I was studying B.Pharm in Bangladesh, my main world was science: medicines, formulations, molecules and everything that comes with pharmacy.

But in my very first semester, something else found a place on my study table alongside all those science books.

Marketing.

I had several modules in Marketing and Information Systems, and that was when I first came across a name that almost every marketing student knows:

Professor Philip Kotler.

His books, including Marketing Management and Principles of Marketing, have been important texts for marketing students around the world. His work has had a huge influence on how we understand markets, customers and value, as well as the well-known 4Ps of the Marketing Mix.

But marketing did not remain just a few university modules for me.

After my B.Pharm, I spent almost six years working in pharmaceutical marketing.

Life then took me in another direction. I continued with higher studies in pharmaceutical biotechnology and biotechnology, and the next major chapter of my professional life became biopharmaceutical and analytical science.

And now, the journey is taking another turn as I start building my own venture related to pharmaceutical and life-science marketing.

Then, right at this point in my journey, something quite special happened.

On 10 September 2026, at the age of 95, Professor Philip Kotler joined a Marketing Summit in Bangladesh live, and I had the opportunity to listen to him.

The person whose book was once sitting on my B.Pharm study table was now speaking live, while I was listening from a completely different chapter of my life.

For me, it was more than just another webinar.

It was a small reminder that careers do not always move in a straight line.

Sometimes something you learned many years ago, an earlier chapter of your professional life, or even a book that once sat on your study table can come back years later with a completely new meaning.

Six Key Lessons on AI and the Future of Marketing

Beyond the personal significance of the experience, the session and the panel discussion that followed raised important questions about the future of marketing in an increasingly AI-driven world.

The following six insights reflect my key takeaways from the discussions involving Professor Philip Kotler, Professor Waldemar Pfoertsch and the other speakers.

AI Is Making Marketing Faster and More Scalable, but the Human Side Still Matters

AI can support data analysis, customer segmentation, personalization and content creation.

But a customer is more than a data point. Understanding emotions, psychology, trust and context still requires human judgment.

Reaching Customers More Precisely Is Not the Same as Understanding Them Better

AI enables increasingly personalized communication. But the bigger question is whether we truly understand the customer's needs, context and decision-making process.

Technology should help us understand customers more deeply, not simply reach them more often.

Trust May Become an Even Stronger Competitive Advantage

As AI makes content easier to create, we may face an abundance of content.

The real question then becomes: Which brands will people actually trust?

Credibility, expertise, consistency and authentic human connection may become increasingly important.

AI Is Changing Execution, but Marketing Fundamentals Remain

Tools, channels and automation will change. But the basic questions remain:

Who is the customer?

What problem are we solving?

What value are we creating?

Why should customers choose and trust us?

AI can strengthen a good marketing strategy, but it cannot replace customer value or clear positioning.

Global Technology Still Needs Local Context

Customer behaviour, culture, language, trust and local data differ across markets.

Global AI tools therefore need to be adapted to local realities, particularly in rapidly changing markets such as Bangladesh.

AI Is Not Only an Efficiency Tool

Beyond automation and productivity, AI can help us better understand customer behaviour, patterns, intent and unmet needs.

So perhaps the more important question is not only how much faster AI can make marketing, but how much better it can help us understand the customer.

The Human Side of the Future

Overall, the discussion showed that technology can expand what marketing can do, while human understanding, trust and meaningful customer relationships remain essential.

Looking back, that is also what makes this experience personally meaningful.

My journey has taken me from pharmacy classrooms to pharmaceutical marketing, from biotechnology studies to analytical science, and now towards building something of my own.

The tools and technologies around us are changing. But the connection between science, markets, customers and human understanding remains as important as ever.

AI can scale marketing, but it cannot replace the human layer.

Oenlo — Scientific Marketing Services & FoodTrace AI

An overview of Oenlo's scientific marketing services for (bio)pharma and biotech, alongside FoodTrace AI, a practical software tool for food-import quality management teams.

Beyond the Lab Bench: Why Digital Marketing is the New Catalyst for Pharma, Biopharma, CDMOs, and CROs

By Sazzad Karim

Throughout my journey spanning over a decade in pharmaceutical marketing, product management, and scientific operations—from driving a 300% turnover growth in legacy portfolios to translating complex analytical workflows into published case studies—I’ve witnessed a fundamental paradigm shift.

For the longest time, the healthcare and life sciences sectors operated on a strict demarcation line: the laboratory generated the science, and the sales teams pushed the product.

But today, that line is blurring. In a digital-first global economy, scientific excellence alone is no longer enough to capture market share.

Let's look at how this transformation impacts four critical pillars of our industry:

Pharma & Biopharma

Traditional detailing and physical congresses have evolved. Healthcare Professionals (HCPs) and institutional buyers are researching solutions online long before a sales rep ever calls. Biopharma brands must adopt data-driven storytelling and multi-channel engagement to build trust at scale.

CDMOs (Contract Development and Manufacturing Organizations)

With global supply chains fluctuating, pharmaceutical companies look for outsourcing partners who project absolute reliability and technical mastery. For CDMOs, digital marketing and thought leadership act as the digital storefront—showcasing compliance, agility, and capacity long before the RFP (Request for Proposal) stage.

CROs (Clinical Research Organizations)

Trust is the ultimate currency in clinical trials. CROs are increasingly leveraging content marketing, SEO, and targeted digital strategies to connect with biotech sponsors, streamline patient recruitment insights, and establish scientific credibility globally.

The Takeaway

Whether you are managing complex biosimilar pipelines, scaling a CDMO facility, or coordinating clinical data, the bridge between laboratory innovation and commercial dominance is strategic digital communication.

Why strong science stays invisible to AI search

Search is no longer the only route to discovery. When an AI assistant answers a question about a therapeutic area, it draws on sources it can read and attribute — and much scientific content meets neither condition.

A biotech company spends four years and considerable money generating a body of evidence. It publishes. It presents at conferences. It builds a website that describes its platform in careful, accurate language. Then a prospective partner asks an AI assistant which companies work in that space, and the company is not mentioned.

This is not a ranking problem, and treating it as one leads to the wrong fixes. The question is narrower: can a machine read this content, and can it attribute what it reads?

Readability is not about simplicity

Scientific content frequently fails the first condition for reasons that have nothing to do with the quality of the science. Key claims sit inside PDFs rather than on indexable pages. Platform descriptions are written as capability statements rather than as answers to questions anyone asks. Mechanism, indication and differentiator appear in three separate documents, none of which connects them.

A human reader with domain knowledge reconstructs the connection. A retrieval system does not reconstruct; it retrieves what is already assembled.

Attribution is the harder half

The second condition is stricter. An AI system that summarises a therapeutic area needs to associate a claim with a named organisation, and it needs a source it can point to. Content published only on a third-party platform — a conference site, a partner's page, a social network — may be read, but the attribution often lands elsewhere.

This is why the location of content matters as much as its quality. Substance on someone else's domain builds someone else's authority.

What this means in practice

The remedy is unglamorous. Move substantive claims out of PDFs and onto pages. Write the connections explicitly rather than assuming domain knowledge. Publish first on your own domain and distribute from there. Make sure the organisation is named where the claim is made.

None of this requires changing the science. It requires changing how the science is represented — which is a different discipline, and one most scientific organisations have never had a reason to develop.

What your CoA says about you

Certificates of analysis are written for regulators and read by procurement. Between those two audiences sits a third that nobody writes for — and increasingly it is a machine.

I spent years generating the numbers that end up on certificates of analysis. Method development, routine testing, the slow work of establishing that a result means what it appears to mean. What I did not think about then, and think about constantly now, is what happens to that document afterwards.

The document has a second life

A CoA leaves the laboratory as a record of fact. It arrives at a quality manager's desk as a decision input. Somewhere in between it gets forwarded, scanned, redacted, filed, and — increasingly — fed into software that extracts fields from it.

Each of those steps was designed for a different purpose than the one before it. The analyst formats for clarity. The quality manager reads for compliance. The software looks for structure. Most documents serve the first two well and the third badly.

Why this matters more than it used to

For most of the history of quality documentation, the third reader did not exist. A human always sat between the document and the decision, supplying the judgement that filled gaps in the format.

That human is still there, and should be. But the volume has changed. A quality team reviewing fifty supplier documents a week cannot give each the attention a single document once received, and the tooling that closes that gap can only work with what the document actually contains.

A practical consequence

When a document is ambiguous to software, the cost is not a wrong answer — good tooling flags rather than guesses. The cost is that the human has to do the work anyway, and the tooling has added a step rather than removed one.

This is the reason I keep returning to the same design principle: software in a regulated workflow should make the professional's judgement faster to apply, not attempt to replace it. The document was never written for the machine. The machine should accommodate that.

Published as a LinkedIn newsletter. Subscribe to receive future editions.

Being published is not the same as being found

A short observation on why the two are drifting apart, and what it means for companies whose credibility rests on their science.

For a long time these were close enough to treat as the same thing. If the work was published in a credible place, the people who needed it would find it. Indexing was the job of librarians and later of search engines, and it worked well enough that nobody had to think about it.

That assumption is quietly breaking. Not because search has become worse, but because an additional layer has appeared between a question and an answer. A growing share of professional enquiry now begins with a system that reads, summarises and decides what to mention — and it makes that decision from whatever it can reach and attribute.

The consequence is uncomfortable for scientific organisations specifically. They tend to have more substance than their visibility suggests, precisely because they invested in the substance and assumed the rest would follow. Organisations with less to say but better digital representation now surface ahead of them.

This is not an argument for marketing louder. It is an argument for noticing that publication and discoverability have become separate problems, and that only one of them is being actively managed.