Foresight

Every conversation about AI and jobs collapses to the same question: is this role safe? That’s the wrong grain to plan at. A role is not one thing that either survives or doesn’t; it’s a bundle of tasks, and AI capability moves through those tasks unevenly — some shift a great deal within a year, some barely move at all. Treating the whole role as a single verdict throws away exactly the information a workforce plan needs, and it leaves planning to follow headlines instead of evidence. That framing is easy to state and hard to act on: an organisation cannot restructure a job title, only the tasks that make it up, and without task-level detail, planning has nothing concrete to hold onto.

Foresight breaks a role down into its component skills and tasks rather than stopping at the job title, and shows where the underlying capability is moving: what’s being absorbed by AI tools already, what’s shifting rather than disappearing, and what’s becoming more valuable as a result. The distinction matters in practice — a task being absorbed needs a different response from a task that is simply changing shape, and collapsing both into a single verdict about the role erases that difference. This skill-level breakdown replaced an earlier role-summary view for the same reason the whole product exists — a summary at the level of the role hides the movement that actually matters.

A comparison view sets two points in time against each other directly, so a change in the underlying analysis shows up as a visible difference rather than something you have to notice by memory. The task-level analysis, originally built around a narrower set of roles, has since been extended to cover operations roles as well. None of this is delivered as a one-off report: the analysis is refreshed as the underlying AI capability changes, because a snapshot of a fast-moving thing goes stale within months, not years.

Foresight suits organisations that need to plan around AI’s effect on work but don’t have a dedicated workforce-planning or people-analytics function to do that analysis in-house — often smaller or mid-sized organisations where the person thinking about this is doing it alongside another job: an operations lead, an HR generalist, a founder setting a hiring plan. It suits planning that has to start somewhere concrete rather than from a general sense that things are changing, and it suits anyone preparing a business case, a training plan or a hiring decision who wants the evidence behind it to be checkable rather than taken on trust.

It does not suit an organisation that already runs its own workforce-analytics function with bespoke modelling — Foresight is a structured starting point, not a replacement for an internal team building organisation-specific models on proprietary data. It also does not suit anyone who wants a single fixed verdict delivered once and filed away: the analysis is deliberately kept current rather than frozen, which means the picture it shows this quarter is not a permanent artefact, and a buyer who wants a static report to cite indefinitely will find that assumption doesn’t hold. It does not suit anyone expecting individual, named-employee predictions either: the analysis works at the level of roles and tasks, not people, and does not attempt to say whose job is safe. And because the product is labelled Evolving rather than a finished, stable release, it does not suit a use case where the questions being asked of it cannot tolerate the tool itself changing shape underneath them.

Foresight is free to use, with no separate paid tier attached to it. To start, get in touch through the contact page — DarkHorseOne’s engineering team reads every enquiry directly and usually replies within two working days. There is no self-serve sign-up published for it: the way in is a conversation with the people who build the product, the same people who read the message.