Documents
Evidence Timeline Reports
For a bounded source set: organise events, accounts, supporting passages and visible gaps into a readable review package.
Brochure / FAQ
A clearer record is a better starting point for a difficult decision.
Legenda is a small evidence practice. We help teams turn bounded material and recorded system behaviour into source-backed reports that another person can inspect.
What we are
We make important claims answerable to the record.
We turn a real question into a bounded investigation: what belongs in scope, what can check the result, and what evidence a reviewer needs to see.
What we are not
Not another confident answer from an AI.
Tools can help prepare work. They are not the conclusion. The conclusion has to stay connected to source material, observed behaviour and visible limits.
Who we are
People who would rather show the working than hide it behind a smooth answer.
We build and review evidence trails for situations where a confident summary is not enough. The useful question is not only “what does this say?” but also “where does that come from, what was left out, and what would change the conclusion?”
What we do
Agent systems
Behaviour and change checks
For a recorded rule-governed system: show a concrete input, its observed result, and a defined alternative.
Delivery
Inspectable artefacts
We package the result so a reviewer can follow the source trail, rather than asking them to trust a black-box summary.
Ongoing
Repeat measurements
A dated answer to "is this still true?" — for a promise, guarantee or known defect that matters to you, re-checked on the releases you care about. "Still holds" is a full, reportable outcome.
How we work
A useful result is more than a tool and more than a summary.
01 · Frame
Name the question and the boundary.
We agree what claim or decision needs support, which material or behaviour belongs in the work, and what is outside it.
02 · Check
Use evidence the system or record can answer.
We test difficult but allowed situations, or trace a statement back to its support, instead of treating a plausible answer as proof.
03 · Explain
Show the result and its limits.
A reviewer sees what was examined, what was observed, where it varies and what still needs human judgement.
What we can do now
Start bounded. Stay honest about the limit.
We are currently set up for scoped, human-reviewed work. You bring a defined question and a bounded record. We agree what is included, what a useful result looks like, and what cannot be inferred from the available material.
We do not offer an instant upload tool, an automatic legal conclusion, or a replacement for accountable professional judgment.
Questions people ask
What is an Evidence Timeline Report?
A concise, source-backed package for a bounded set of material. It can organise events, claims, supporting passages, differing accounts and known gaps so a reviewer can inspect the record without starting from a blank document pile.
Can I upload anything and get an answer immediately?
No. The first step is a scope conversation. We currently work as a concierge service because the boundary of the source material and the question both matter to the reliability of the report.
Could I not do this with an AI or a formal tool myself?
You can use those tools, and an independent rerun is valuable. The tool is not the service. Legenda provides a bounded question, checks that do not depend on a fluent answer, a preserved trail of what was examined, and a conclusion that stays within what the record or the system actually shows.
How is this more than a script that logs inputs and outputs?
A log records what happened once. A bounded verification also fixes the question, controls and stop rules in advance; distinguishes a clean result from an unexecuted check; repeats where variation matters; and preserves enough material for another reviewer to inspect or rerun it.
That is the standard our current technical examples are built toward. They remain self-measurements: no public example has yet passed an independent customer-side audit, so we do not claim that an outside expert has accepted them. Every result carries a version, date and stated limit because a new release reopens the question.
What if the answer isn't a simple yes or no?
Some questions have one honest answer: a claim holds, or it does not. Others are really a question about a range of cases — for example, "under which of these configurations does the same behaviour still hold, and where does it change?" For that kind of question, the useful result is a map: which cases held, which did not, and which we were unable to test. A mixed map is not a failed check or a missing answer. It is the answer you asked for, and we will tell you which parts of it you can rely on.
Do you decide who is right?
No. Legenda does not decide truth, fault, motive, cause, credibility or legal liability. It makes support, disagreement and limits easier to inspect.
Can you work with complex software?
For a concrete, recorded situation we can compare what happened before and after a defined change, and keep the evidence trail intact. The technical details live separately; the public result remains a readable account with clear limits, not a general certification.
How do we start?
Send a short note about the material, its size, the question you need answered and your timeline. We will say whether it is a fit before discussing a paid scope.