Prompt It!

Promptlets as Lenses for Niche Knowledge

Most workplace text is underspecified by default. A bug report, a spreadsheet, a requirements document, or an internal email rarely explains itself. Its meaning depends on how insiders read it: what they notice first, what they ignore, and what they treat as a signal.

That interpretive layer is rarely written down. It lives in experience, habit, and shared understanding. Two people can read the same text and reach different conclusions—not because the text changed, but because the lens applied to it did.

This is what we mean by niche knowledge. It is not general intelligence, and it is not public domain expertise. It is the local, internal way a specific team or organisation decides what matters.

Niche Knowledge Is a Lens, Not a Dataset

Niche knowledge is often mistaken for data. In practice, it behaves more like a set of interpretive rules:

• What counts as high priority
• Which omissions are suspicious
• Which metrics matter, and which are noise
• What language signals risk or escalation
• What “good enough” actually means in context

This kind of knowledge is procedural, not declarative. It does not sit comfortably in a knowledge base or a memory store. It only reveals itself when applied to new material.

Why Conversational AI Struggles Here

In a conversational interface, niche knowledge tends to dissolve into implication. Assumptions are absorbed implicitly, reinforced by tone, and carried forward as narrative momentum.

Over time, it becomes difficult to answer basic questions: Which assumptions were applied? Why was this judged important? Would we reach the same conclusion tomorrow?

The knowledge is still there—but it is no longer explicit, inspectable, or reusable without contamination.

Promptlets as Explicit Lenses

Prompt It takes a different approach. Instead of treating niche knowledge as something the system should remember, it treats it as something the user should apply.

A promptlet can encode:

• Assumptions the model is allowed to make
• Priorities specific to a team or domain
• Heuristics for evaluation or critique
• Questions that should always be asked
• Failure modes worth checking by default

Running a promptlet is equivalent to saying: “Apply our way of thinking to this text.”

Reuse Without Drift

Because a promptlet is explicit, it can be reused without accumulating hidden state. The same lens can be applied to different inputs, on different days, by different people.

There is no memory to decay, no personality to drift, and no narrative to inherit accidentally. The knowledge stays where it belongs: in the lens itself.

Organisational Consistency Without Bureaucracy

This approach allows organisations to achieve consistency without centralised control or rigid process.

Promptlets do not enforce conclusions. They enforce how conclusions are reached. Disagreement remains possible—but it happens in the open, against shared criteria.

In One Line

Expertise lives in the lens, not in memory.

Prompt It does not try to internalise niche knowledge. It gives you a way to make it explicit, reusable, and safe to apply to new text.