When I ran bioinformatics groups at Warner Lambert and Sanofi, and later global R&D information systems at Schering Geigy, the hardest part of the job was rarely the technology. It was translation: turning what a biologist needed into something an engineering team could build, and turning what the engineers had built into something leadership could fund with confidence.
That translation problem has not gone away. If anything, AI has made it sharper.
Why the gap is wider now
Twenty-five years ago the computational toolset in a discovery organization was relatively contained: sequence analysis, databases, some modeling. Today a small biotech may be weighing foundation models, cloud data platforms, external AI partners and vendor tools whose claims are hard to verify. Each choice has scientific consequences that the people evaluating the technology are not always positioned to see, and technical consequences the scientists may not anticipate.
What the bridging role actually does
- Frames the scientific question so it can be tested computationally, with clear criteria for success.
- Evaluates tools and vendors against the organization’s real workflows rather than a demo.
- Designs the data foundation (models, standards, validation) so that later AI work has something trustworthy to stand on.
- Tells leadership the truth about what a platform can and cannot do, in terms that support a decision.
Getting it without a full-time executive
Many early- and mid-stage companies need this judgment a few days a month, not five days a week. A fractional arrangement, with a defined scope, regular architecture and strategy reviews, and someone available when a vendor pitch or a partnership term sheet lands, can provide senior perspective at a fraction of the cost of an executive hire, and at the stage when it most changes outcomes.
The best time to bring that perspective in is before the platform decisions are locked, not after the data turns out to be unusable. If you are approaching one of those decisions, I am always glad to talk it through.