Source provenance answers one critical question: where did this information come from? Freshness answers another: is the information still current enough to support the decision being made?

Opportunity intelligence needs both.

Correct information can become misleading over time

A procurement record can be accurately captured and still become less useful if the underlying notice changes, a deadline moves, an amendment appears, a status changes, or the buyer publishes new information. The original extraction may remain historically correct while no longer representing the latest state.

That is why freshness should not be treated as a cosmetic timestamp. It is part of the evidence model.

There are several clocks in one record

A useful system may need to distinguish among multiple dates:

  • When the buyer published the information.
  • When the source was last changed, if that is available.
  • When Kellette collected or refreshed the record.
  • When the reasoning was last evaluated.
  • When a deadline or decision point occurs.

Those dates answer different questions. Collapsing them into a single “updated” label can hide important context.

Freshness affects reasoning

If an opportunity was evaluated against yesterday’s source and the source changed today, the previous reasoning should not automatically be treated as final. Material changes may need to trigger re-extraction, reclassification, or a fresh review against the company’s Opportunity DNA.

This is especially important for rejection intelligence. A record that was correctly rejected based on one version may deserve another look if a material requirement changes.

Evidence has a source, a version, and a moment in time.

What a freshness signal should communicate

Kellette’s intended design is not to reduce freshness to a mysterious score. The interface should make the state understandable. Examples of useful states include:

  • Recently collected and unchanged.
  • Source changed since last evaluation.
  • Material fields require re-check.
  • Deadline approaching.
  • Source unavailable or freshness uncertain.

The exact production states will be validated when live source integrations are connected. The principle is already clear: users should be able to see when recency affects confidence in the recommendation.

Freshness belongs beside provenance

Provenance without freshness can tell you who said something but not whether the statement still represents the active opportunity. Freshness without provenance can tell you something changed but not which evidence supports the change.

Together they create a stronger audit path:

  1. Preserve the source.
  2. Track the source version or collection state.
  3. Record the facts extracted from that state.
  4. Attach the reasoning to the evidence used.
  5. Re-evaluate when material evidence changes.

Why this matters for a small cyber firm

A small team should not have to remember which of dozens of notices changed overnight. That is exactly the kind of operational burden software should absorb. But automation only helps when it is honest about what it knows and when it last checked.

Kellette is being designed so freshness is visible as part of the evidence chain. The goal is simple: a pursuit decision should not look current merely because the interface is current.

Next decision

Make one live pursuit decision visible.

Explore Decision Sprint for a source-linked written recommendation on one named opportunity.

Review Decision Sprint