Abstract
Manual sales prospecting in emerging Connected-Commerce (C-Commerce) markets creates a fundamental misalignment between individual representative judgment and organization-level optimization. Representatives at Zbooni, a UAE-based C-Commerce platform, allocated approximately 30% of working hours to manual prospecting while achieving only 1% conversion rates. This paper presents LeadFlow, a full-stack AI-driven prospecting system developed using formal systems engineering methodology. The central challenge was deploying machine-learning-based qualification without historical performance data—a zero-baseline environment. We contribute (1) a five-phase zero-baseline validation framework that bootstraps qualification capability from stakeholder-derived heuristics and progressively transfers trust to empirical evidence, and (2) a requirements traceability methodology that resolves principal-agent misalignment through objective-tree decomposition. This paper's primary contribution is methodological: the zero-baseline framework and requirements traceability approach are designed to be potentially transferable to similar data-scarce deployment contexts independent of domain. Pilot deployment has confirmed discovery pipeline throughput and scoring coverage across 12,720 businesses and 2,496 active leads; projected impact on effort reduction and conversion rate improvement are design targets dependent on outreach deployment, reported here as evaluation hypotheses rather than empirical outcomes. We predict that the framework generalizes beyond C-Commerce to any organization deploying AI-assisted judgment in data-scarce environments.
My STS research investigates how social media algorithms restructure human subjectivity and autonomy. Specifically, I explore how engagement-driven platforms transform Kantian subjects of rational moral reasoning into Freudian subjects governed by desire. Drawing on Kant’s moral philosophy and Freudian psychoanalytic theory, I argue that recommendation systems optimized for engagement replace rational autonomy with algorithmically engineered drives for power, dominance, and belonging.
The manosphere, a loosely networked constellation of online communities organized around gendered grievance, serves as the primary case study. While the pervasive psycho-social effects of social media are a population-wide phenomenon, the manosphere offers a particularly legible and intensified expression of these dynamics. These communities are not a cultural anomaly but a revealing window into the underlying mechanisms of platform-mediated transformation.
Through discourse analysis of Reddit manosphere content and close reading of platform design incentives, I trace how algorithmic amplification produces the digital unconscious: a collective psychic structure shaped by continuous interaction with systems that condition users toward affective response over reflective judgement. In these environments, polarization is not incidental, its structural.
I make two central claims. First, the Kantian framework of autonomous moral reasoning is systematically undermined by engagement optimization, which rewards emotional outrage over reflective judgement. Second, that Freudian concepts of the pleasure principle and drives provide a more accurate model of algorithmically-mediated subjectivity than rational choice frameworks. From this lens, the manosphere is not a failure of individual reason, but a predictable product of platform design. More broadly, this research reframes contemporary political polarization as a structurally induced phenomenon rather than a purely individual or ideological failing.
Although building LeadFlow and writing this thesis appear unrelated, they ultimately engage with the same underlying phenomenon from opposite perspectives. My STS research examines how engagement-optimized systems shape human behavior by conditioning users toward affective responses and exploiting psychological drives such as belonging and dominance. LeadFlow operates through structurally similar mechanisms. Each outreach message is designed as a targeted interaction intended to elicit a response, and the system iteratively refines itself based on which signals predict conversion. What is framed in practice as personalization can, from the perspective of this research, also be understood as a form of behavioral shaping.
What unsettles me most is the scale. LeadFlow is a small system with a limited dataset. Social media platforms deploy the same logic at billions of users, with vastly more data and fewer constraints. If my thesis is right, platform design is one of the most consequential, yet least governed, forms of power in our lives today. Recommendation systems are deeply embedded in everyday life, influencing not only what people see, but how they think, feel, and relate to others. Understanding these systems is therefore a necessary first step toward addressing their broader social and political consequences.
