gishkimResearch concept · 16 November 2026
Exploring change-review evidence

Make change reviews
easier to check.

When a rule, calculation or model changes, teams must decide what to recheck and explain what they left out. We are exploring whether a structured review can make that work faster and clearer.

Discuss a recent change

Starting with software teams handling financial rules and calculations. No production integration required for an initial conversation.

01 / THE QUESTION

What needs checking again?

A useful review needs more than a list of changed files. It needs a clear boundary, a record of the dependencies considered, the checks performed and the questions that remain open.

What changed?

Record the change, the before and after states, and the connections supplied by the team.

What supports the conclusion?

Separate observations from declarations. Make assumptions visible to the reviewer.

What is still unknown?

Keep unresolved items visible. Missing evidence should never silently become a positive result.

02 / AN ILLUSTRATION

Different evidence. Different conclusions.

This fictional example shows the distinctions under investigation. It does not execute CHANGE, discover dependencies or assess your software.

Synthetic example only

A lending calculation changes.

The team wants to leave one connected report out of its repeat checks. What evidence supports that decision?

Observed unchanged

The recorded output did not change in this example. That observation supports a bounded conclusion, not proof that every possible behaviour is unchanged.

Any conclusion depends on the supplied model, recorded observations and agreed review boundary. A wrong dependency map can leave effects unseen.

03 / A FIRST EVALUATION

Start with one past change.

  1. Understand the existing review. What happened, how did the team check it, and what did the work cost?
  2. Agree a fair comparison. Use the same historical case, recorded expectations and the team's current tools.
  3. Count the whole effort. Include preparing and maintaining the dependency map, reviewing findings and handling mistakes.
  4. Let the evidence decide. Continue only if the result is useful enough for the customer to pay for.

This is a proposed evaluation process. Scope, commercial permission and technical feasibility must be agreed before delivery. No savings or assessment outcome is guaranteed.

04 / WHERE THINGS STAND

A working research tool.
A business still to prove.

Gishkim is a provisional brand exploring a commercial use for Luiz F. Nunes da Silva's CHANGE research software. The software currently consists of command-line instruments. A hosted service and customer integrations are not available through this page.

In the inventor's controlled comparisons, conventional alternatives matched the software on the properties tested. Those results do not establish functional exclusivity. The commercial question is whether the assembled tools and review evidence save enough total effort to justify adoption.

These findings are reported by the inventor and have not been independently reproduced by Gishkim. The software and any patent rights are separate assets. This page makes no claim of granted patent protection, unique functionality, regulatory approval or proven customer savings.

Basis for this update: the inventor's 16 November 2026 email, Diligence Digest Part I, and What Changed Since the Earlier November Brief. No code, confidential technical material or patent licensing offer is included. Partnership, ownership and commercial rights remain unagreed.

How did your last change review work?

If your team spends significant time deciding what to recheck, we'd like to understand the process and the gaps your existing tools leave.

Speak to Jerome

This opens your email app. There is no account, upload service or payment collection on this page. Please do not send confidential code or customer data.