Why Artificial Intelligence is Tough to Manage, But Can be Worth it for Expert Networks


The following is a Q&A with Graham Mills, co-founder of Cambridge UK start-up techspert.io, an expert network specializing in the healthcare/life sciences sector which uses artificial intelligence to source experts.

SB: There are a growing number of start-up expert networks which claim to use AI to source experts but when you look more closely, you see them hiring lots of entry-level staff like a conventional expert network rather than technology talent.  So, we are wondering if the AI claims are just hype? 

GM:  For expert networks, the industry is an ever-evolving arms race in building competitive advantage and differentiation, both for sustainable growth and for attention-grabbing PR. For newer entrants (and some incumbents), claiming the use of AI to solve a defined business need is a popular route to both, and distinguishing the genuine technological leaders and those who like how it sounds can be challenging.

We’re seeing new entrants arise almost weekly at this stage, and every second firm appears to claim an AI-driven competitive advantage. You’re right that cursory reviews of their employee base on platforms like LinkedIn and Glassdoor suggest they do not currently have the headcount or capabilities to be true technological leaders in our industry.

The reasons why companies might avoid this investment in capabilities is clear though. Investing in building out a high-performing tech team is expensive and has a slower return on investment than rapidly scaling Associate-level teams to process connections manually. At techspert.io, partly by virtue of being embedded in the University of Cambridge’s tech ecosystem, we’re making this investment in building technological leadership, both because we see it as the obvious next stage of evolution for our industry, but also because through this focus we can directly address many of the core challenges expert networks face, including high rates of employee turnover and an almost complete reliance on the LinkedIn search bar for expert identification.

SB: From your experience what are the challenges for an expert network in managing AI?

GM: Firstly, it’s worth underscoring that AI is no magic bullet on it’s own. It’s important that firms approach problems not with a “how do we use AI to solve it?” mindset, but more a “how do we solve this?” mindset, recognising that AI, and the many other technological approaches that we have in our toolbelt, will be more or less appropriate depending on the situation. Sometimes, in fact, tech development may not be the answer – evolving human processes can be more impactful!

Secondly, the development of AI-driven solutions requires the focus of a well structured and curated team of incredibly talented developers, engineers, and project managers, all linked in transparent communications with the commercial side of the business to really understand the crux of the problems we’re trying to solve for our clients. We all know how challenging it can be to build effective communication channels in companies of all sizes, but if this breaks down then we can very quickly end up with projects and investments that don’t yield the net benefit we had hoped for. At techspert.io we have invested significant time and resources into building and structuring this team. Again, we’re fortunate to have a wealth of experienced and intelligent talent to draw from locally.

SB:  You say human processes can be more impactful than AI.  Can you elaborate on this a bit further?

GM: Without a doubt, human input is still an essential component of how expert networks do and should work today. You do see some entrants who are aiming to remove human input entirely, but I do not believe the industry is ready for, or in want of, that.

What’s essential is taking a step back and undertaking a frank analysis of where human input is most valuable and most difficult to replace. We still believe that human input is a major value add for client relationship management, and in particular for working with clients in a consultative capacity to ensure insights we source are as precision matched to client needs as possible.

In contrast, we do not believe that expert identification — all the LinkedIn searching or probing of static databases of experts — is best left to human manual input. Firstly, the amount of data a single human can process is extremely limited. Here we use technology to map out knowledge landscapes and rapidly digest evolving data from different sources, structures, and languages. This approach also means we’re sourcing experts that most other firms may never even know exist, experts who might not even know expert networks exist, and this we believe is a powerful step in sourcing higher quality insights.

Secondly, a major challenge that our clients have faced is varying degrees of output quality depending on the Associate or Project Manager they’ve been assigned. That’s because individuals have wildly varying degrees of literacy in different fields. Some might be better at understanding and finding niche medical experts, some in qualifying legal experts in APAC for example. By employing technology here, we’re able to ensure that regardless of the field, we’re capable of rapidly understanding and recruiting from areas to a higher degree of specificity than any individual, no matter their background, could achieve. This also allows us to invest heavily in developing out own Associates to be as effective project managers as possible.

SB: Can technology be used to more efficiently set up consultations?

GM: Yes, we have an automated, but hyper-personalised, solicitation process to remove manual inefficiencies in the connection process.  Our goal is to keep human time for where it has a genuine value add. This part excites me as it means our cost base and operational structure is phenomenally different to the more human-reliant incumbents which, as we scale, will provide us some powerful strategies with which we can compete and re-shape how companies use primary insight and knowledge.

SB: You mention a lower cost base — is part of your strategy lower fees? There seems to be more price competition in the expert network industry with all the new entrants trying to gain share.

GM: With the rate at which new entrants are joining the market, this can definitely lead to an arms race relating to pricing. This is a slippery slope, as plummeting rates will encourage cutting of corners in identification, depressing expert reimbursement, and scaling back client support, all of which will translate to worsening client experiences, which is already a pain point many firms face.

Ultimately though, pricing is only one of the variables companies can compete on. AlphaSights, for example, are renowned for their customer engagement experience, with well trained Associates providing a high-quality experience to their clients in high pressure industries.

At techspert.io, we have chosen to differentiate on quality of experts, enabled not by throwing more bodies at the problem, but through an overhaul of the fundamental model through which expert identification and recruitment happens.

 SB: We hear that a growing number of experts are opting out of consultations – is this something you encounter? 

GM: Yes, it appears that experts are feeling spammed by multiple expert networks emailing them to participate, all without the networks really knowing what the experts can advise on. This means often experts need to go through burdensome and lengthy screening processes, which typically end up disqualifying them. They can end up investing 15-30 minutes of their time (sometimes the networks even require a call!) only to be declined an opportunity to consult. It’s a no brainer that this would lead to disenfranchisement.

Our approach has been purpose-built to address these key levers, and as such we’re proud of achieving high response rates from experts globally, with a model that fairly compensates all experts for their valuable time.

SB: You raised capital earlier this year with the goal of continuing to extend you technology platform, as well as opening a New York office.  Has the pandemic affected your plans?

GM: Well, as you might suspect it has been a tad challenging to make the transition to the US since the pandemic hit. The majority of our revenue still comes from the US though, so as a region it has never fallen off our radar.

A key reason we wanted to expand in the US was to be close to our current clients, embed in ecosystems, and accelerate our acquisition of new clients. However, the vast majority of our clients are working from home, which means that even if we were there, all our meetings would still be online. Ironically, the pandemic has removed a key competitive disadvantage we had, which was meeting US clients less frequently in person. As such, we’ve been able to massively scale our activity in the US through the pandemic, finding engaging clients more seamless from the UK and elsewhere, but still with the plan in our back pocket of setting up in the US once the business environment allows.

Also, we have been fortunate to have our origins and a strong client base focused on the healthcare and life sciences industry, which has meant we’ve remained well on track to hit all our growth targets this year as the healthcare industry has bubbled away at record levels of activity.


About Author

Sandy Bragg is a principal at Integrity Research Associates. He has over thirty years experience as an investment research professional. Prior to joining Integrity in 2006, he was an Executive Managing Director at Standard & Poors, managing S&P’s equity research business and fund information properties. Sandy has an MBA from New York University and BA from Williams College. Email: Sanford.Bragg@integrity-research.com

Leave A Reply