AI and Robotics Rewrite Drug Discovery

Discover the Process Power your competitors can’t copy

Welcome to Clear Thinking — the go-to newsletter for investors, founders, and senior executives at the intersection of AI, technology, and biology.

Join 950+ industry leaders across the US, Israel, and Europe — including decision-makers from Merck, IQVIA, Medtronic, Flatiron Health, and Debiopharm.

Subscribe for real-world pattern analysis from 45+ founder interviews on what separates a defensible business from a capital-burning science project.

Intro

This week I write about the secrets behind Process Power — how to design a unique factory process that your competitors can’t figure out.

AI and Robotics Rewrite Drug Discovery is Part 5 of a 7 part series on how TechBio companies build moats and achieve exceptional returns.

Previously on Clear Thinking:

• Part 1 - Counter-Positioning Power with Founder-Market misfit at Delve Health

• Part 2 - Cornered-Resource Power at the Israeli brain on a chip company Itay and Beyond

• Part 3 - Switching Costs Power at SAP and Flatiron Health, the oncology evidence company

• Part 4 - Brand Power at Selmer Paris and Caris, the molecular diagnostics company

My podcast guest this week was Yann Gaston-Mathé, founder and CEO of Iktos, the Iktos drug discovery platform that integrates Gen AI with robotics.

I analyze Process Power with 2 examples: the famous TPS - Toyota Production System and Iktos.

Let’s start with a quick review of the 7 powers.

The 7 Powers in Short

In his 2016 book 7 Powers: The Foundations of Business Strategy, Hamilton Helmer looked at companies like Netflix and Pixar and developed a toolkit to create Power and build moats.

Scale Economies — Unit costs decline as production volume increases due to fixed cost spreading and operational efficiencies. Barriers rise when competitors can’t match your volume economics.

Network Economies — Product value increases as more users join the network. Each additional user makes the product more valuable for everyone. See The Network Effects Bible for a detailed treatment of the different kinds of network effects.

Counter-Positioning — A newcomer adopts a superior business model that incumbents can’t copy without damaging their existing business. The incumbent faces a “damned if you do, damned if you don’t” dilemma.

Switching Costs — Customers face high financial, time, or risk costs when changing suppliers, keeping them loyal even when alternatives exist.

Branding — Customers attribute higher value based on reputation and trust, not just product features.

Cornered Resource — Exclusive access to a critical asset (data, talent, IP, relationships, raw materials) that others can’t easily obtain.

Process Power — Organizational capabilities and methods that enable superior operations and are difficult for competitors to replicate — often built through years of learning and refinement.

TPS - The Toyota Production System

GM toured their factories. Toyota hid nothing and shared their manufacturing processes. But GM and Ford were unable to replicate the success Toyota had building cars that ran forever and never needed to visit the shop besides preventive maintenance.

In 1950 Eiji Toyoda, then a managing director of Toyota Motor Company, spent three months in Dearborn, Michigan studying the Ford River Rouge Plant, “the largest integrated factory in the world.”61 His earlier visit to Ford in 1929 had left him profoundly impressed by the Ford revolution in manufacturing. His reaction to the 1950 visit was quite the opposite. The Ford plant maintained deep inventories, which were used to smooth out production irregularities, but this seemed wasteful to Toyoda-san. He was more impressed by the supermarkets he saw around the city: their system of restocking only when shelves had gone empty aligned with the parsimonious nature he had developed over years of war-driven shortages. He thought he could do better than Ford, and so he set to work.
Page 148 The 7 Powers, Helmer

TPS seems to be a straightforward set of interlocking procedures, such as JIT production, kaizen (continuous improvement), kanban (inventory control), andon cords (devices to allow workers to stop production and identify a problem so it can be fixed).

GM executives naturally assumed you could clone TPS by copying these procedures. Their questions were focused on the floor, the assembly plant, on what’s happening on the production line. That’s not the real issue.

Software engineers that adopted agile procedures like this now understand that it isn’t enough to practice rituals with Japanese names.

You need something else.

Whether it’s cars, drug discovery or software engineering - the secret of Process Power is in the patterns and anti-patterns in your organization.

Process Power Is Pattern Power

And pattern power takes decades, not quarters.

Process Power is the most misunderstood of the Seven Powers because people confuse rituals with results. They copy the stand-ups, the kanban boards, the retrospectives — and none of it matters.

What matters is the deep implementation of patterns and anti-patterns in the organization over years. That library becomes culture. And that culture becomes Process Power.

Let’s analyze Process Power in terms of how it creates benefits for the organization and how it creates a barrier to entry for the competition.

Benefit

A Process-Powered company has orchestrated thousands of micro-patterns across engineering, operations, logistics, QA, governance, coordination, and decision-making.

• Which actions reduce rework.

• Which handoffs fail.

• What metrics matter

• What anti-patterns repeat themselves

• What parts in the process break quality and create defects, waste, delays, or mistrust in your teams.

At Toyota (and at other companies with operational excellence like Intel) the patterns and anti-patterns are part of company culture. People leave. New people join. The culture sustains the patterns and anti-patterns.

Barrier to entry

Building the culture and the interlocking support and quality systems inside your company and with your suppliers takes years.

This is why GM could not copy-paste the TPS.

This is how Intel can copy-paste an Intel-standard Fab (especially when a new advanced fab costs $25BN).

Automobile manufacturing is tens of thousands of coupled processes. Drug discovery is even worse. Software engineering at scale isn’t far behind.

You don’t get Process Power with one process improvement or by adopting agile rituals.

The corollary to the time you need to orchestrate patterns and anti-patterns is opacity.

As Michael Porter noted in his landmark paper “What is strategy”, it’s not obvious at all to the competition how you do it.

Let’s segue into Iktos.

Iktos

Iktos built a drug discovery platform that integrates Gen AI with robotics.

I believe that the Iktos moat comes from Process Power, the ability to integrate technologies into a system that compounds value over time.

For the team at Iktos, that moat is AI-driven orchestration — the way their AI doesn’t just design molecules, but actually coordinates and optimizes the entire design-and-make cycle, tailored specifically for robotic synthesis to deliver true process acceleration.

The defensibility is not attached to any single component, as many companies build models or automate chemical synthesis. The hard part, and where the moat is, is in the system-level integration:

AI determines what to design, explicitly taking into account robotic constraints.

AI prioritizes what to synthesize, based on live inventory and chemist-imposed constraints.

Early design cycles are accelerated and more effective, critically shortening timelines to reach candidates.

This orchestration layer is extremely difficult to replicate. And because every iteration improves the system, the advantage compounds over time.

This Week on Life Sciences Today — Iktos

In this episode, I sat down with Yann Gaston-Mathe, co-founder and CEO of Iktos, one of the most technically sophisticated teams in AI-driven drug discovery.
A mathematical engineer trained at École Polytechnique, Jan spent years inside big pharma (Sanofi, Ipsen) and consulting before founding Iktos in 2016 with two partners—a chemist-turned-cheminformatician and an AI scientist.

Iktos tackles the slowest part of drug discovery: the design-make-test-analyze loop. Their integrated platform combines chemistry-aware generative models, virtual synthesis, robotic reaction execution, and automated biological testing—enabling a single chemist to run 100 parallel reactions instead of 2. The goal: cut timelines by a third, reduce costs, and improve success rates on hard targets.

The company runs a partnership-driven model with pharma while advancing in-house programs like MTHFD2. Over the next 12–18 months, they aim to deliver a clinical candidate, close a major pharma deal, and secure a new financing round.

Visit Iktos here

You can see the episode here

About Me

I’m a pharma-tech founder who learned hard lessons the hard way. Over 22 years, I built five companies: four exits, one glorious flop. Customers ranged from Israeli medical device startups to Verily, Amgen and the Fortune 1 company.

I work with a small number of well-funded life science executives using an inversion methodology based on 23 identified anti-design patterns.

Instead of generic startup playbooks, we identify what’s killing your competitive position and systematically remove the blockers to success.

If you’re a Series A+ life science company dealing with competitive positioning challenges, let’s talk.

Book a 15’ intro call.

If you love my writing — share it.

If you want more like this — subscribe