The Southampton friends building an AI future that 'works for humans' - In-Common

TEXT START: When Fin Jones and Alfred Sommarstroem were pupils at Wildern School in Southampton, neither could have imagined they would one day be building a technology company together.

The Dissection

The article is a launch-story wrapped in humane language. It converts an AI labor displacement problem into a platform opportunity: AI agents supposedly create work by hiring humans for tasks they cannot yet perform in the physical world or tasks marketed as “innately human.” The founders’ friendship, trust, sacrifice, and Southampton origins provide emotional legitimacy, while the economic mechanism receives almost no scrutiny.

TaskPool is not preserving the old employment system. It is building an intermediary layer between autonomous software and residual human capabilities. That is transition intermediation, with elements of verification arbitrage and the Hyena’s Gambit: profit from the gaps between what AI can currently do and what physical reality still demands.

The Core Fallacy

The central error is confusing AI’s current limitations with a durable human economic role.

The platform assumes that because AI agents cannot yet photograph, verify, run errands, or produce certain forms of creative work, humans will remain economically essential. Under the Discontinuity Thesis, that is a temporary technical gap, not a permanent moat. Once robotics, computer vision, synthetic media, remote actuation, trusted sensors, and automated verification improve, the same tasks become targets for further automation.

TaskPool may increase human demand during the lag. It does not defeat P1, P2, or P3. It merely organizes a shrinking residue of work until the residue becomes cheaper, faster, and more scalable to automate. The claim that AI will “create more opportunities for people” mistakes induced demand for restored productive participation. AI agents hiring humans is still AI-controlled production. The human is a callable endpoint, not a restored economic sovereign.

Hidden Assumptions

• That physical-world tasks will remain human-exclusive rather than becoming robotics and sensor problems.
• That “uniquely human touch” will retain sufficient economic value after synthetic content becomes abundant and cheap.
• That AI agents will route work through a public human marketplace instead of vertically integrating trusted contractors, robotics, or proprietary execution networks.
• That task volume will grow faster than automation erodes task prices.
• That workers will receive meaningful bargaining power rather than competing in a race to the bottom for atomized assignments.
• That platform activity, user accounts, and applications indicate durable demand rather than novelty, subsidized labor, or temporary curiosity.
• That ethical intent can override competitive pressure. It cannot. If automation is cheaper and adequate, firms will use it.
• That keeping humans “in the loop” preserves productive participation. It may instead formalize humans as low-margin maintenance, verification, and exception-handling labor.

Social Function

Primary classification: transition management and ideological anesthetic, with a partial truth.

The partial truth is real: AI deployment initially generates demand for physical execution, edge cases, authentication, local knowledge, and human trust. Platforms can capture that temporary demand and may produce income for some workers.

The anesthetic is the implication that this makes humans central to the long-run AI economy. It does not. The article supplies a morally attractive narrative—technology creating opportunity—at precisely the moment when the wage-to-consumption circuit is being severed. It turns displacement into a feel-good marketplace story and treats founders’ sincerity as evidence of structural viability. Sincerity is not a defense against cost curves.

The founders are not villains. They are building a plausible instrument for the transition. But the article mistakes an ambulance service for a cure.

The Verdict

TaskPool is a potentially useful transitional niche, not a solution to AI job losses. Its strongest position is as a temporary human-execution and verification layer while AI remains digitally dominant but physically incomplete. Its eventual exposure is obvious: the platform’s own success makes the remaining human tasks legible, standardized, and easier to automate.

Under DT logic, TaskPool can create pockets of income, possibly a viable company, and perhaps a profitable position for its owners. It cannot preserve mass productive participation. The likely endpoint is not “humans remain at the centre,” but humans become exception handlers around an expanding machine core—until the exceptions are automated too.