It’s one of the big debates in research IT. Which strategy is correct? The all-in-one platform or integrating best-in-class solutions? The answer is, as usual, neither… and both! In this month’s blog we chew over some of the issues.
You know that you need a strategic approach to your research IT. You like the talk of end-to-end solutions, knowledge shared across the organization, traceability, integration. And, of course, rationalizing cost is always a focus. So, the idea of working with as few technologies and vendors as possible – perhaps even just one platform – has a natural appeal. Can you cut through the forest of acronyms – ELN, LIMS, ML, PLM – and have it all in one place, preferably ‘powered by AI’?
That articulates one view of research IT strategy – call it the ‘all-in-one’ approach.
Its counterpoint might be described as ‘best-in-class’ or ‘point-solution’ focused. I should be able to take the best tools to perform each individual research task – a Design of Experiment study, for example – and stitch those together so that individuals in my team always have at their fingertips the capabilities that make their workflows as efficient as possible.
The truth will, as usual, lie somewhere between these two extremes. Most research IT environments will use a small number of central platforms (rarely, in practice, just one) surrounded by an eco-system of point solutions. So, let’s look, briefly, at some of the general factors to consider when assessing your choices.
| Pros | Cons | |
|---|---|---|
| ‘All in one’ platforms | – One (or at least fewer) source(s) of the truth for data – Can reduce barriers to leveraging data across the organization – Fewer vendor relationships to manage and fewer support / training organizations to work with, creating economies of scale | – Often require compromises in usability and in how they fit to individual workflows – Vendor less likely to focus on specific task-centred requirements – Often requires customization, with maintenance costs controlled by vendor – High dependency on one vendor |
| Point solutions | – Better match to specific tasks and workflows – Functionality and support more likely to address detailed requirements and to be more agile – Vendor develops specific expertise and is highly motivated to keep solution ‘best-in-class’ – Spreads risk | – Can result in data silos and friction in sharing data and knowledge – Requires careful thinking and use of flexible, open technologies to connect solutions effectively – Requires more vendors and a partnership / collaboration strategy |
At Intellegens, we are primarily a ‘best-in-class solution’. We have one focus – developing robust machine learning methodology, tools, and expertise for R&D and applying this to recurring tasks in the chemicals, materials, FMCG, and process industries. This means our ML tools are more likely to develop and maintain the functionality that users really need than ML tools competing for attention and investment within larger systems. We’ve written on this blog before about how, even within our own niche, ‘one size does not fit all’ and loading more functionality into one piece of software can cloud the user experience. Our solution has been to develop the Alchemite™ Suite range of task-focused apps.
But we also recognize the power of platforms and the need to integrate and rationalize. That’s why our apps are interoperable with each other and offer an open API to connect to other lab systems and technologies. It certainly makes strategic sense for R&D organizations to implement single platforms in many areas – for example, as the ‘backbone’ for managing their critical data and knowledge. Our aim is to plug into this vital infrastructure – for example, we’ve recently shown how we can interoperate with Revitty Signals ELN. And we want to work alongside other best-in-class expert systems – see, for example, this week’s announcement with materials simulation provider Materials Design. And such integration and connectivity between systems is only going to get easier in the future with the new AI infrastructure of MCP and agents.
In fact, the winning research IT strategy is probably one that will come very naturally to scientists. Recognize the need for a strategic approach and for integration – but be skeptical of solutions that promise to solve all your problems at once. Be prepared to make sensible compromises to increase efficiency and effectiveness across teams – but make sure the researchers at the sharp end are presented with tools they will actually use. Make rational choices and, when you do that, factor-in the future – both maintenance costs and whether your vendors are motivated to keep every tool at the cutting-edge. Avoid dependency, but value partnership, collaboration, and building strong scientific networks.
If our experience of navigating such complexity could help you, get in touch – we’re here to help… working with our partners and always open to new collaborations.
