Why Backing Founders Deepened My Conviction About People About Culture

Wiki Article

AI Is Only As Good As The Culture It's Constructed Into
The conversation about artificial intelligence in business has a problem, and the problem isn't a technical one. The capabilities of modern AI and machine learning are incredible, moving at a rate that makes most predictions on the state they'll be in the next 18 months obsolete by the time these 18 months have expired. The problem is the gap between the what AI is able to do under controlled conditions - in a properly-funded research environment, with crystal clear data, a clear definition of the issue, and engineers with the option of continually testing until the system operates as it should - and the actual results when implemented in genuine organizations with actual cultures that are governed by real organisational structures and real people who have their own established views on the validity of a brand new system as something to take seriously or something to route around while still maintaining the appearance compliance. I've been building products using AI since prior to when the wave of AI enthusiasm became fashionable for all businesses to claim fluency in the space. When I founded 1Touch, AI-driven matching and recommendation systems were not an additional feature that we incorporated to make the product more attractive to investors. They formed part of the product architecture, it was the basis on which the platform produced value and the thing that had to be reliable and work at level for the business to be viable. So I have direct, personal experience of what happens when you try to construct something truly intelligent within a organization and product at the same time and the main thing that I am always returning to, across every context in the past I've faced this issue, is the technology isn't always the only factor that is limiting. It is always the culture.
What I refer to as specific and practical, not abstract. AI systems require data to perform - accurate, clean and well-structured data that shows the actual phenomenon that the system is trying understand and make predictions about. Businesses with strong data culture create that kind of data in the natural course of their existing processes. They are clear and have consistently applied definitions of what they're measuring and the reason for that. They've agreed on a set of standards for the way data is recorded, collected, and stored. They have accountability frameworks that provide data quality as an explicit task rather than the general intentions. Companies that lack strong data cultures produce something that appears like data. It's there in systems that are able to be examined and utilized to create charts but does not have a consistent definition as well as in its quality and full of issues with structure and not mapped out that any AI system built on over it will increase and magnify the problem rather than obtaining genuine signals from it. The companies in that segment often don't realise this until they're already well into an AI deployment and the results do not correspond to the vendor's promises, at which point it is tempting to blame the technology when the actual problem is the operational and culture that the technology was built on.

The second element of culture which determines AI outcomes is openness within the organisation in the sense that employees in the company are genuinely willing to let the system influence or alter how they work instead of interpreting it as threat to their professional know-how, their institutional authority, or their job security. This is a socio-cultural and leadership problem as opposed to a technical one that needs to be addressed. It is a problem that begins at the high levels. If leaders of senior positions engage with AI outputs in a way that is selective - accepting those that validate the previous beliefs and disadvantaging those that do not – their behavior sends an indication to anyone who is watching that the commitment of the organisation to data-driven decision-making is conditional rather than genuine and that conditionality will propagate through the organisation faster than any other training program or change management plan can neutralize. If senior management models an authentic, consistent approach to AI outputs that include the ability to modify their decision-making when evidence suggests that they need to, then the company's capability to utilize AI effectively is significantly improved and remarkably quickly.

This isn't an abstract statement about how organisations ought to behave in the context of theory. It is a description of the pattern I've witnessed play out over and over again in organizations that had a significant amount of budgets, a genuine strategic commitment to AI adoption, as well as leadership teams that were truly excited about the possibilities of the technology. This pattern is so common that I've decided to treat practices for data governance as a fundamental diagnostic factor when I am evaluating any company's AI preparedness. Before I ask what the current technology stack is, before I inquire about the specific uses cases that the organization is developing, I want to know about data governance. What is the definition of its primary metrics? Who's responsible if data quality is not high enough? What happens when two different organizations have different information on similar business facts, and how is that conflict resolved? The answers to those questions can tell me more about chances of AI succeed in comparison to any discussion regarding algorithms, platforms or timelines for implementation.

I believe that the organizations that will reap the most lasting value out of AI in the coming decade will not be those which adopt the latest technology first, nor those who invest the most extensively in AI talent and infrastructure over the next few years. They are the ones that construct the cultural and operational foundations for using that technology well - the data management processes that result in trustworthy inputs, decision-making frameworks that offer evidence to influence outcomes and leadership behaviors which signal to all people in the company that the commitment to an operation that is driven by data is real and not just a flimsy performance. The technology itself will become more commoditized and accessible. The culture to use it well will remain scarce, because it requires constant effort and genuine dedication from an executive over time rather than an individual strategic decision or a technology investment. This lack of resources is where the actual competitive advantage will lie and it's an advantage that, once it is built will grow in a manner other advantages purely technological ever. View James Deller for more info including what working across industries confirmed what i suspected about lasting impact.



The Data Infrastructure Problem Nobody Wants To Discuss
Each company I've been closely with during the last 10 years and a half - whether as an investor, founder or as an operational adviser I have been told, at some point in our collaboration, that data is a critical element of how they take decisions. Certain of them are truly committed to this in a way that is apparent in the way the organisation actually operates. Most believe they're doing it right, but the concept they're proposing is more of an aspiration than actual operational reality - that is, a vision of the type of organisation they're aiming for rather than the one they currently live in. The gap that exists between genuine information-driven decision-making and performance of data-driven decisions – the careful maintenance of the public appearance of an evidence-based processes without the infrastructure that can make it real - is one the most important gaps present in contemporary business. It is also one of the most frequently ignored ones due to the fact that the infrastructure problem that causes it is difficult to talk about, difficult present to external stakeholders, and enormously difficult to classify against the more prominent strategic and commercial jobs that vie for the same attention from leadership and organizational resources.
When businesses talk about Data strategy, they generally tend to focus on what they are planning to develop on top of their data - tools for analysis, machine learning applications for operational dashboards, and real-time data that provide the kind of predictive information that sounds truly compelling in the context of a board conference or an investor update. What they usually talk about less often, and with considerably less energy and enthusiasm, are the core infrastructure which determines whether all of those capabilities are actually working as they are advertised: the data governance frameworks which define distinct and consistent definitions of what's being measured and for what purpose what is being measured; the collection and retention methods that establish the accuracy and comparability of the information to be gathered; the checks that find the errors and correct them before they propagate through the system, and cause harm to the outputs that everyone depends on; the structure of the organization and accountability mechanisms that make data quality an ongoing and explicit obligation as opposed to everyone's vague ineffective plan. The plumbing, or the. The plumbing isn't glamorous. It's difficult to photograph for a annual report. It has no outputs that can be demonstrated in a compelling way. It is, from my experience across a substantial amount of organizations across different sectors and in different stages of development, much worse than the company believes it is.

The problem becomes more serious over time in ways that are becoming difficult and expensive to fix. A company that has been operating with inconsistencies or inadequately defined terminology for data across different functions for the past three years has three years of historical information that cannot be easily compared or aggregated and compared. This is not due to the fact that the data is not there, but because the same language has been used to denote different elements in different parts of the enterprise, and the differences are buried in the data itself rather than appearing on the surface. An organization whose quality assurance has been the responsibility of a primary responsibility, and not having a properly resourced and dedicated function has data that's reliability is varying in ways that are not systematically documented and therefore cannot be systematically accounted for when the data is used for making decisions. A company that has allowed multiple operational systems to accumulate redundant and partially conflicting data on the same customers, products or transactions may have a data-related landscape that's really difficult to fix without operational disruption significant enough to create a risk.

The reason that this problem continues to exist over a large number of organizations that are actually smart in the field of strategy and totally dedicated to a data-driven approach to business is that fixing it requires sustained investment in work that does not provide visible quick-term results of the sort the resource allocation systems of an organisation are designed to reward. Analytics platforms that are new produce visual outputs: dashboards and reports that can be shown and reports that can be shared with the board of directors, and information which can be used to create press releases regarding digital transformation. A data governance system creates invisible infrastructure, which is cleaner in its definitions along with more standardized collection processes as well as more reliable inputs to existing systems in place. The first is relatively easy to justify in budget discussions because you are able to demonstrate what they will gain. It's the second, and requires enough organisational credibility and perseverance to prove that the infrastructure investment will, over time, bring better results from every technology that is built on top it. It's an effective argument in abstract, but difficult to make in a competitive environment with initiatives whose benefits are more immediate and more apparent.

I've presented this argument in enough different organisational contexts and observed it either succeed or fail due to certain reasons, to gain a fairly clear view of the factors that determine whether the company finally solves its data infrastructure issue or simply defers it. The primary factor is a leader - a specific one with enough organizational credibility and an conviction about why the infrastructure is important, and the perseverance to continually make your case till the infrastructure becomes an absolute priority, rather than something that is a constant item on the list of items that everyone agrees on but which never climb to the top. The leader must be able to pay for the short-term cost of the infrastructure investment - the delay or disruption to existing processes, or the absence in the immediate production of results with the conviction that the long-term capabilities it generates will justify the price several times over. What this requires, ultimately, is a culture in where the long-term investments in infrastructure are highly valued and recognized at the upper levels of management, not simply mentioned in strategic documents, but later discarded when the quarterly discussion on resource allocation occurs. To create that kind of culture is, itself considered a long-term investment. In my opinion, one those investments with the highest returns an organization with a commitment to data-driven operation can make.}