From Copilot To Autopilot: Why Biopharma Must Stop Renting Its Future
By Romi Singh, Ph.D. and Jared Auclair, Ph.D.
The era of outsourced regulatory work and rented thinking is ending. Agentic AI makes in-house capability inevitable, if leadership has the nerve to build it.
Every biopharma CEO now claims to run an "AI-forward company." Visit the regulatory, clinical, or pharmacovigilance teams, and you find something else. Six-figure SaaS contracts nobody fully uses. Consultants billing to summarize data the in-house team already analyzed. Senior staff are pasting confidential text into public chatbots because IT blocked the tools that would actually help. The gap between the slide deck and reality has stopped being a quirk. It is a strategic liability.
In his March 2026 essay "Services: The New Software," Sequoia partner Julien Bek puts the stakes cleanly: copilots sell the tool, autopilots sell the work. For a decade, SaaS vendors charged by the seat and let the customer figure out what to do next. That model made sense when the models were not yet capable of owning an outcome. It does not anymore. The total addressable market for autopilots, Bek argues, isn't the software budget. It is the entire labor spend, insourced and outsourced combined. In biopharma, the services line dwarfs the software line, and that is where the real AI opportunity sits.
The Regulatory Affairs department could be the cleanest test case. The work is knowledge-heavy and already outsourced to CROs, regulatory consultancies, medical writers, and SaaS vendors. Each line item is a defined scope with a measurable outcome. Replacing any of them with an internal AI capability is a vendor swap, not a reorganization. The budget already exists, and the ROI could land the quarter the move is made.
Take a workflow every sponsor recognizes. A Health Authority sends an information request that touches three regions. The standard playbook routes the work to an expensive regulatory consultant to pull from prior submissions, scan the literature, cross-reference current agency guidance, and draft the response. The actual judgment call could take the in-house lead a day to complete. She knows what to argue, where to push back, and what to concede. The rest is intelligence work. An internal agent built against the company's own dossier, submission history, and house positions can produce the same draft overnight at machine cost, with an audit trail no consultancy will hand over.
Beyond Efficiency Theater
The deeper problem isn't the outsourcing. It is how the industry has framed the AI opportunity. Most "AI initiatives" in biopharma today appear to be exercises in efficiency theater: shaving twenty percent off a literature review, compressing a CSR draft cycle, automating a mail merge. These are real wins, but they are rounding errors against what agentic systems can actually do.
What leadership needs to start naming is that AI is not a productivity tool. It is a partner in judgment. A regulatory strategist working alongside an agentic system isn't drafting faster. She is reasoning across ten jurisdictions in parallel, stress-testing her own assumptions in real time, and exploring options that would have taken a committee six weeks. That is a different kind of work entirely. It cannot be rented from consultants or licensed by the seat.
Yet the prevailing model keeps the thinking outside the four walls. Consultants absorb the learning. SaaS vendors accumulate the patterns of the work. Contractors, some carrying company badges for more than a decade, take institutional knowledge home every night. Meanwhile, the employees closest to work go underground, bypassing blocked tools because the official channels fear what they don't understand. A company cannot credibly call itself AI-forward while its best people are using AI in secret or not to its full potential.
Four Moves That Actually Matter
Reversing this is not a technology problem. It is a governance and culture problem. Four moves separate the companies that will lead from the ones that will only brand themselves as leaders.
1. Put the Chief Procurement Officer in the C-suite, reporting to the CEO. Every consulting engagement, every SaaS contract, every long-tenured "contractor" with a company badge should flow through one accountable executive. Functional silos, where IT buys its own tools, Regulatory hires its own consultants, and Clinical procures its own CROs, are how the industry built the fragmented and externally dependent cost structure it now complains about. A CPO charged with auditing every outsourced workflow against an internal AI alternative becomes the most important seat in the AI transition. Engagements should be time-boxed by default, not renewed by inertia.
2. Run internal hackathons with an uncomfortable prompt: how would you make your own role redundant? Ask every regulatory, pharmacovigilance, clinical, and medical writing team the same question: how would you use agentic AI to automate the work you do today, so you can do harder work tomorrow? Reward the best answers with retention bonuses, not pink slips. An employee who can articulate how to automate her own task is demonstrating the judgment the company will need over the next decade. Punish that honesty, and the response will be silence. Reward it, and the response will be a roadmap.
3. Build the capability inside. The dependence on consultants and SaaS vendors is, at its core, a capability gap. Closing it requires a dedicated internal team, small and senior and permanent, whose mandate is to move the organization from copilot use to autopilot deployment on secure, governed infrastructure. This is not another training program. It is a standing capability that works alongside the business, builds agents against real workflows, and hands ownership back. The tools are there. What is missing is the willingness to insource intelligence rather than rent it.
4. Boards must demand CEOs willing to break the consulting habit. The hardest part of this transition falls on CEOs trained in management-consulting culture, where strategic confidence is outsourced by reflex to a branded deck. The habit produces a quiet dependency: the conviction that no major decision is legitimate until someone wearing a different logo has blessed it. Boards are the only group positioned to break that pattern. They should be asking how much of the company's thinking still lives outside its walls and whether the CEO has the conviction to dismantle a consulting-dependent operation rather than manage around it. The next biopharma CEO cannot be selected for pedigree. The right candidate trusts internal conviction over external validation, and empowers a CPO to challenge the CEO's favorite consultants on the way to bringing that work in-house.
What This Is Actually For
None of this argues for eliminating people. The opposite: use AI to supplement talent, not replace it. Used well, it frees the workforce from repetition and ritual and redirects human ambition toward questions machines cannot yet ask. The point is not what AI does instead of us. It is what becomes possible together: new products, new trial designs, new regulatory strategies, safety signals caught earlier. That is the move from doing the same things faster to doing things that were previously impossible.
Biopharma does not have a technology problem. It has a hubris problem. The tools are ready. The budgets are allocated; they are flowing to the wrong places. People are already using AI; they are doing so in the shadows. Every quarter spent debating the risk of acting is a quarter conceded to companies that have stopped debating.
The biopharma companies that will define the next era of medicine will not be the ones with the best consultants, the most SaaS licenses, or the loudest press releases. They will be the ones who stopped renting their own future and built it.
About The Authors:
Rominder (Romi) Singh, Ph.D., is an accomplished thought leader in global drug development, multiregional clinical trials, and the application of AI/ML in regulatory sciences. Dr. Singh was appointed Professor of Practice at Northeastern University following a distinguished career at leading biopharma companies (Pfizer, Amgen, Merck & GSK). He was on the ICH expert working group that developed the ICH E17 guidelines on Multiregional Clinical Trials. He founded GRA Advisors to advise on global regulatory strategies. Dr. Singh obtained his BSc (Hons) from St. Stephen's College, a doctorate from the University of Kansas, and postdoctoral training from the University of Washington. He completed his certification from MIT in Artificial Intelligence and Business Strategy.
Jared R. Auclair, Ph.D., is the Dean of the College of Professional Studies and Director of Bioinnovation at Northeastern University. In his role as Vice Provost for Research Economic Development, he integrates work-integrated learning with use-inspired research, fostering global partnerships and community co-creation. As Director of Bioinnovation, Dr. Auclair leverages university-wide expertise in biotechnology, regulatory sciences, and AI to advance the expansion of Northeastern’s life sciences programs. Additionally, he serves as a faculty member in the Department of Chemistry and Chemical Biology, collaborating with industry, government, and academia to advance biopharmaceutical development and analysis. His work bridges the gap between educational excellence and impactful scientific innovation.