On this page
01Written by AI, or computed by Othor AI
Ask an AI model where your revenue is heading and it will write you a confident paragraph. Othor AI computes the forecast instead, and the difference shows in every part of it.
| A forecast written by AI | A forecast computed by Othor AI | |
|---|---|---|
| Where the number comes from | Words that sound right | Your own numbers, computed by code each night |
| Tested on your past | No | Yes: each number’s way of forecasting is the one that did best on its own history |
| How sure it is | It sounds sure either way | A range, and one word: Reliable, Fair or Rough, earned from its track record |
| When it can’t tell | It still answers | It says nothing, and says why |
| The same answer twice | Not guaranteed | Yes: the same data gives the same forecast, to the byte |
| Gets checked | No | Every forecast is scored against what really happened |
02Tested on your own past
For each number, Othor AI tries several ways of forecasting on that number’s own history: it pretends to stand at each of the last twelve periods, forecasts the next one, and checks how close it came. The way that did best is used. It has to beat the simplest guess of all, repeating the last period (or the same period last year), or Othor AI uses that simple guess instead.
The forecast is a range, not a single number: where the number will most likely land. The range is set from how far off the past forecasts were, so it is as wide as your own numbers make it, and no narrower.
03How sure, in one word
- Reliable: the number has landed inside its range at least three times in four, and the typical miss is small.
- Fair: usually inside its range, with some misses.
- Rough: often outside its range or far off. Read it as a direction, not a plan.
The word comes from the number’s own track record: its last forecasts against what really happened. Until a number has six of those, the word comes from the test on its past, and Othor AI says so.
04When it says nothing
Some nights a forecast would mislead you, so Othor AI says why instead of giving one:
- Not enough history yet to forecast it.
- Waiting for a source to update.
- Too uncertain to call right now: the range would be too wide to plan with.
- Its definition changed recently, so its past doesn’t describe it any more.
05The record, on the sample company
This is the track record of Indigo Thread Co., the invented company Othor AI is built and tested on, after 100 nights of runs. It is computed by the run itself; nothing here is typed by hand.
| Number | Inside its range | Typical miss | How sure | Scored on |
|---|---|---|---|---|
| DSO (monthly) | 9 of the last 11 | 4.0% | Reliable | its past, tested |
| DSO (weekly) | 9 of the last 12 | 10.6% | Fair | its own forecasts |
| Gross margin per metre (weekly) | 6 of the last 12 | 8.4% | Rough | its own forecasts |
| Metres dispatched (weekly) | 9 of the last 12 | 17.3% | Fair | its own forecasts |
| Net revenue (monthly) | 10 of the last 11 | 3.3% | Reliable | its past, tested |
| Net revenue (weekly) | 10 of the last 12 | 8.6% | Reliable | its own forecasts |
| Receivables over 60 days (weekly) | 10 of the last 12 | 5.4% | Reliable | its own forecasts |
06Where you see it
What’s ahead, in every Morning Brief, shows where each headline number is heading, how sure Othor AI is, and why; “How this forecast was made” opens the method, the range, the last result and what drives it. Big Picture, on Home, puts the forecasts in plain words beside last night’s findings. In both, the words may be written with AI’s help, but every figure is computed. More on how every number is made.