Introduction – The Journey Starts at the Bench
We didn’t set out to build software, we set out to make better medicines, faster. But after years on the bench and the production floor, it became clear that the biggest barrier wasn’t the chemistry. It was the gap between the people doing the work and the systems they were expected to use.
In process development, we obsess over yield, purity, and throughput. But often, success hinges on something more intangible: the intuition of the technician who knows the scent of a clean reaction, the engineer who senses a pressure shift before the sensor does, the scientist who just knows the timing.
These skills are vital, yet hard to teach, harder to scale, and nearly impossible to preserve when someone moves on.
Like many in pharma, I’ve seen robust processes stall not because of bad chemistry, but because the know-how wasn’t transferable. SOPs were too rigid. Batch records too vague. Recipes lived in the margins of notebooks or the memory of an overworked expert. The very systems meant to support us often became friction points.
Even digital tools like ELNs, while a step up from paper, often act as passive repositories, good for recording, but not for guiding. They capture what happened, not how or why it worked.
That realization pushed us to pivot from writing processes to building platforms.
The Problem – Talent Bottlenecks in Modern Pharma
Pharma is filled with brilliant people doing high-stakes work under tight constraints. But across many labs and manufacturing sites, we still rely on tribal knowledge, fragmented systems, and static documentation.
Experimental prep is inconsistent. Procedures are handed down like folklore. Decisions about mixing order, color change, or subtle exotherms often vanish into the ether. When someone leaves, we don’t just lose a pair of hands, we lose judgment, nuance, and experience.
This isn’t just a documentation issue – it’s a scalability problem. A process that only works with one team, in one place, under one specific set of conditions can’t scale. And as the industry shifts toward distributed manufacturing, rapid tech transfer, and leaner teams, this challenge becomes even more acute.
Legacy Manufacturing Execution System (MES) and Quality Management Systems (QMS) tools were supposed to help. But most are rigid, compliance-first systems designed for oversight, not insight. They track but they don’t teach. They preserve steps but not wisdom. They constrain more than they empower.
An ELN can tell you what was done but rarely captures the logic behind critical decisions the “why” that makes or breaks process transfer. These systems document outcomes, not judgment.
What we needed was something different: a platform that captures expertise, guides users of all experience levels, and evolves as our process knowledge grows.
The Pivot – From Paper Recipes to Programmable Chemistry
The turning point came when we asked a simple question: What if chemistry itself could be programmable?
Not just in the automation sense, but in the way we design, document, and scale experimental logic, going beyond ELNs that merely record steps, to systems that encode expert thinking into the workflow itself.
At my organization, we began experimenting with a composable MES — a flexible, modular platform originally built for frontline manufacturing.
Unlike traditional systems, it allows teams to build and adapt digital workflows without writing code. Instead of relying on rigid software or paper SOPs, users can drag and drop elements like prompts, sensor inputs, and decision logic to create step-by-step procedures tailored to their process.
But this shift wasn’t just about going paperless or digitizing forms. It was about building systems that reflect the real decisions chemists make — the kind they often don’t write down.
We built workflows that embedded:
• Photos of color changes and precipitate formation
• Prompts triggered by gas evolution or pH shift
• Video demonstrations of key handling steps
• Inline sensor data with branching logic
• Conditional instructions: “If temperature exceeds X, pause and notify”
These weren’t optional footnotes they were baked into the workflow. A chemist stepping into an unfamiliar process could see what success looks like, how failure manifests, and why certain steps matter.
These aren’t just attached files or free-text notes. Unlike traditional ELNs, every element is operational, prompting action, guiding decisions, allowing user input, and ensuring that critical cues are never overlooked.
Just as importantly, these workflows were editable by the people who actually do the work. You didn’t need to be a software developer. If you could think logically and drag-and-drop, you could adapt the process. It turned our team into builders, not just users.
And while the tool enabled it, the deeper shift was cultural: we began treating our process knowledge as code, something to iterate, test, and scale.
Real-World Impact – From Expert Intuition to Universal Access
This change didn’t happen overnight. But over time, it transformed how we trained, transferred, and executed chemistry.
New hires no longer relied solely on shadowing or deciphering cryptic notes. They followed dynamic workflows, rich with visuals and embedded context, that guided them through unfamiliar procedures and explained why each step mattered.
I’ve seen operators with just a few months of experience safely execute complex fill-finish steps that once required years of tacit knowledge. Not because they memorized a SOP, but because the system made the expertise visible, actionable, and available in real time.
Unlike ELNs, which serve primarily as after-the-fact records, this system supports decision-making as the work happens. It’s not just a journal, it’s a guide.
Senior scientists, meanwhile, became mentors at scale. One of our chemists created a walkthrough of a challenging reaction quench, complete with gas evolution profiles, images of what “too fast” looks like, and side-by-side examples of success vs. failure. That expertise didn’t disappear at the end of the day, it now lives in the workflow, accessible to every team member, on any shift, at any site.
This approach doesn’t replace expertise. It amplifies it, making excellence easier to access and quality easier to reproduce.
Closing – A Call to the Industry
Pharmaceutical manufacturing is built on precision. But too often, that precision relies on individual memory, unspoken judgment, and documentation that can’t keep up with the realities of the floor.
If we want to make medicines faster, safer, and more equitably, especially in a world moving toward distributed, responsive production, we need systems that make excellence easier to scale. That doesn’t mean replacing people. It means designing tools that respect their intuition and help them share it.
Composable platforms aren’t silver bullets. But they represent a shift away from rigid systems that track compliance, and toward flexible frameworks that enable capability. The goal isn’t digital transformation for its own sake. It’s to empower small teams to do sophisticated work, and to allow new contributors to meaningfully participate from day one.
As we confront workforce shortages, increasingly complex chemistries, and growing pressure for faster tech transfer, we must ask hard questions: Are our systems designed to extract expertise or to expand it? Are we building tools for inspection or for insight?
The future of pharma won’t be built on paper. It will be built on platforms. But those platforms must be shaped by the floor, not just the boardroom.
Because at the end of the day, great medicines don’t come from systems alone.
They come from the people those systems enable.
Figure 1. A composable recipe builder interface that embeds expert knowledge—such as visual cues, color changes, and real-time sensor feedback—directly into the procedure. By integrating decision logic and contextual media, the system transforms traditional SOPs.
Figure 2. Example interface screens from a composable MES workflow guiding skid cleaning (top) and reaction monitoring (bottom). These steps integrate manufacturer instruction videos, visual references, and structured user input to ensure procedural clarity and capture critical observations such as color change during reaction initiation. By embedding media and logic directly into the workflow, the system reinforces correct technique and enables real-time documentation of execution conditions.
Table 1. Functional Differences Between Composable MES and ELN.
ABOUT THE AUTHOR
Luke Rogers is Vice President of Innovation at On Demand Pharmaceuticals, where he leads efforts to modernize pharmaceutical manufacturing through modular, distributed technologies. With a background in medicinal chemistry and hands-on experience in process development, he focuses on building systems that make expert intuition scalable. His work bridges chemistry, engineering, and digital platforms to enable more agile, resilient medicine production.
