AI is transforming job references
A startup with less than $20m just automated trust with an AI model that gathers references. Every executive should read this, writes Paul Armstrong
Boardy.ai didn’t announce a new AI model last week. Far more interestingly, the company announced it could perform reference checks. For a startup that has raised less than $20m, that sounds impressive. For everyone else, it should sound like a warning.
An email arrived in my inbox from Boardy’s AI networking agent. Subject line: “I taught myself how to do reference checks.” Boardy explained that before introducing people, it could now contact former bosses, colleagues and other professional referees to build a richer picture of who somebody is and whether they are worth meeting. I had some questions, but Boardy didn’t respond to three separate requests for comment, which means nobody outside the company can speak to what safeguards, verification processes or accuracy standards actually exist inside it. That absence of information is itself informative.
Trust is the real product
Identity, reputation and trust become complicated very quickly once software gets involved. After facilitating more than 100,000 introductions, Boardy identified the same problem recruiters, investors and executives encounter every day: introductions are only as valuable as the trust behind them. Most discussion around artificial intelligence remains focused on productivity, asking which jobs disappear, which businesses become more efficient and which companies win the next platform war. Far less attention is being paid to a more fundamental shift taking place underneath all of that. Software is beginning to move beyond automating tasks and into something much more valuable and much harder to regulate: trust.
Every market has a hidden operating system, and that operating system is trust. Contracts, audits and regulation help manage risk after a decision has been made, yet somebody still needs to decide whether another person is worth backing before any of those safeguards matter although some, including Mark Zuckerberg, are trying new avenues. Investors deploy capital because they trust founders. Boards appoint executives because they trust judgement. Buyers complete acquisitions because they trust management teams. A large part of the professional services economy exists largely to support those decisions, and has existed for a very long time because the underlying problem has never been easy to solve and because the people who could solve it at scale never quite had the tools.
The friction was the point
Before AI, answering that question involved friction. Somebody knew somebody. Another person made the call. Someone else collected the references, interpreted what was said and made a judgement about how much weight to give any of it. Time, effort and cost acted as natural limits because most reputational information remained trapped inside human networks. A founder’s reputation existed in conversations, an executive’s in the boardroom, the investor’s may have been built over decades of being in specific networks. Frustrating limitations that were, in retrospect, protective in ways nobody fully appreciated until software started removing them. AI is changing all these things at a pace that no-one really understands the repercussions of until they smack them in the face.
Boardy’s proposition is that software changes those economics entirely. Questions that once required days of effort can potentially be answered, or perhaps inferred is the better description, in minutes: who has worked with this person, what did they think, who else knows them, how consistent are the answers across different sources. Whether the information gathered is accurate, current or trustworthy is a genuinely open question that Boardy’s silence on the subject doesn’t help answer. Professional reputation is becoming machine-readable regardless of whether the machines reading it are getting it right, and therein lies the rub, and where we currently are. A messy middle or rather the messy beginning if we’re being honest.
Arguing about whether any of this is genuinely new misses the point. Reference checks have existed for decades and so has the professional services economy built around them. What software changes is not the activity itself but the economics underneath it, and history suggests shifts of this kind move faster and cut deeper than anyone expects. Social interaction existed long before social media turned conversations into datasets worth hundreds of billions of dollars, and what changed was not the activity but the concentration of power that followed once someone owned the infrastructure around it.
The deeper pockets have noticed
Boardy matters less because of what it is today than what it reveals about where value is moving next. Professional trust has always been monetised, just slowly and through human intermediaries who at least had reputations of their own to protect. A startup with less than $20m has demonstrated that software can perform parts of that function faster, cheaper and at a scale no human network could match, and the logic of that demonstration doesn’t stay contained to a single small company for very long.
Professional life already happens inside a surprisingly small number of systems. Conversations happen on one platform, work happens on another, identity sits somewhere else and reputation is scattered across all of them with no single owner and no single point of control. Boardy points towards a world where somebody stitches those pieces together and starts selling confidence as a service, and the firms best placed to do that aren’t the ones currently raising less than $20m.
Professional reputation has historically been difficult to collect, difficult to verify and difficult to distribute, and those limitations kept the whole thing from becoming a surveillance problem. Software is changing that without much noise and without the scrutiny the shift warrants. Most executives won’t notice until the infrastructure is already in place and the economics have moved on without them.
Once credibility becomes infrastructure, whoever controls that infrastructure holds a new kind of power over professional life. None of this is happening in a vacuum either. An unstable administration in Washington is spending, and doing, whatever it takes to ensure no rival gets a technological edge, which means the money flowing into these systems isn’t necessarily being guided by questions of societal value. Products get built for paydays, and safeguards get treated as someone else’s problem. The future is about to get considerably messier than it already is, and anyone who thinks we’re close to understanding what we’ve started hasn’t been paying attention.
Boardy sent an email and demonstrated something most organisations aren’t ready for. Whether that’s smart or not has yet to be seen, but what follows isn’t likely to have a cute subject line.
Paul Armstrong is the founder of emerging tech advisory, TBD Group, and its intelligence community, TBD+.