Why Telli 5 scores its own answers

When an assistant writes every answer in the same calm, confident voice, you can't tell which answers stand on solid ground. So on every Evidence answer, Telli 5 shows a Reliability Factor: a score from 0 to 100 that tells you how well the sources support what the answer says.

How the score is built

Two things go into it. First, how strong the sources are: who published them, whether they are independent, and whether they are fact or opinion. Second, whether two separate AI checkers found each cited sentence supported by its source. A claim that isn't supported pulls the score down, and you can see which one it was.

What the bands mean

Strong (85–100): strong, independent sources agree.
Supported (70–84): good sources back the main claims.
Qualified (50–69): parts hold up, parts are thinner or disputed.
Limited (below 50): weak or little support. Treat it with care.
Inconclusive: too little evidence to score. Telli 5 says so instead of guessing.

What the score is not

It is not a probability that the answer is true. A high score means strong, independent sources agree; it doesn't mean the answer can't be wrong. A low score doesn't mean the answer is false; it means the evidence found doesn't carry it far.

Three promises about the score

Shown with its version. Every score names the version of the rules that produced it.
Never changed quietly. Any change to how scores are made gets a new version name.
Never for sale. No partner, advertiser or vendor can raise or lower a score or a source grade.

Why it matters

Most questions don't need a score. But when an answer will shape a decision about money, school work, the news or your family, knowing how much weight it can bear is the most useful thing an assistant can tell you.