Early in my career, I remember sitting in project meetings where everyone celebrated a game-changing milestone.
The compound had shown positive preclinical results.
Months of work had paid off. The data looked promising. Our system worked.
The program expanded the candidate for a new indication.
Many groups that celebrate successful IND submission are celebrating the wrong victory.
Passing a toxicity screen tells us something important. It tells us the therapy is unlikely to harm the patient.
That had already been established with this asset.
What it doesn’t tell us is whether it’s actually going to help them, and for who.
We had found a new disease indication that this drug could be applied for.
The result led to the asset effectively skipping Phase I trials for a new subset of target patients.
That saves time, money, and confidence, all in one experiment.
Let’s back up for context.
For decades, our industry has become remarkably sophisticated at identifying compounds that shouldn’t move forward. We’ve invested billions of dollars refining toxicology models, improving safety assessments, and building regulatory frameworks designed to protect patients.
Those efforts have unquestionably saved lives.
But they haven’t solved the biggest challenge in drug development.
Most therapeutic programs don’t fail because they’re unsafe.
They fail because they simply don’t work.
By the time those failures become obvious, years of research have often passed. Hundreds of millions of dollars have already been invested. Entire teams have dedicated significant portions of their careers to advancing a program that ultimately can’t deliver meaningful clinical benefit.
When a late-stage efficacy failure occurs, the consequences extend far beyond a single balance sheet.
Patients lose potential treatment options.
Scientists lose years of momentum.
Investors become more cautious.
Early-stage innovators find it harder to raise capital.
And promising ideas that might have succeeded under different circumstances never receive the opportunity.
I’ve spent much of my career working at the intersection of biology, technology, and pharmaceutical development. One lesson has remained remarkably consistent.
The scientific question isn’t always the hardest one to answer.
Sometimes the hardest question is whether we have enough confidence to invest in finding the answer.
Investment conviction has quietly become one of the defining challenges in modern biotechnology.
Every financing decision, licensing agreement, partnership, and acquisition ultimately depends on confidence that a therapeutic program has a reasonable chance of succeeding.
Yet those decisions are often made with surprisingly limited translational insight.
We’re asking investors and pharmaceutical companies to place enormous bets while looking through a clouded lens while mapping a course to the moon.
That’s why I believe conversations surrounding the FDA Modernization Act 3.0 represent something much larger than regulatory modernization.
They represent an opportunity to redefine what evidence looks like before therapies ever reach clinical trials.
And sometimes even after they’ve established they are safe.
Much of the discussion has understandably focused on replacing or supplementing traditional toxicology models with more predictive human-relevant systems.
That’s an important step.
But I don’t believe it should be the final destination.
Imagine if advanced biological models could provide meaningful efficacy signals earlier in development.
Imagine if functional human data, computational biology, AI-assisted analysis, and next-generation validation platforms worked together to answer not only, “Is this therapy safe?” but also, “Is this therapy likely to work?”
The implications would extend well beyond science.
Biopharmaceutical companies could allocate capital with greater confidence.
Investors could identify stronger opportunities earlier.
Technology developers would receive clearer feedback about which platforms generate meaningful value.
Most importantly, patients could gain access to effective therapies more quickly because fewer resources would be spent advancing programs that were unlikely to succeed from the beginning.
This is where collaboration becomes essential.
As I discussed in Part 1, no single organization possesses every piece of the puzzle.
Established pharmaceutical companies understand clinical development, regulatory expectations, and have access to decades of historical data.
Emerging biotech platform companies are developing entirely new ways to model human biology.
When those perspectives remain disconnected, progress slows.
When they begin informing one another, entirely new possibilities emerge.
The future of therapeutic development won’t be defined solely by our ability to predict toxicity.
It will be defined by our ability to generate earlier confidence that a therapy can truly change a patient’s life.
That may become the most valuable prediction we ever make, and ultimately, fix the broken parts of an industry that is intended to save lives.



