The Inputless Manifesto
Why we refuse to make operators invent the question before they can see the problem.
The bottleneck is not usually missing data. It is the ritual: someone has to know what to ask, write it, wait, and interpret. By then the useful window has often closed. We built Inputless Analytics so that step is optional.
Talorik, on Inputless Analytics
Four constraints
What the product has to do, or it is just another BI tool.
No blank query box
If you have to invent the question, you will miss the event. The software watches the feeds and points at what changed.
Runs on the stream
The model updates as records arrive. You are not waiting for last night’s batch or Monday’s pack.
Your machines
The model is generated and stored on infrastructure you operate. Cloud is optional.
A call, with a trail
Output is a ranked Course of Action and the records behind it. Charts exist if you need them.
The argument
Inputless Analytics is a product choice. Classic stacks ask people to translate a hunch into a query, the query into a job, the job into a chart. Each hop costs time and drops context.
We skip the hunch. The software keeps a model of how things connect in your operation. When a pattern breaks, it raises it, with the subgraph attached. Someone still decides. They just start from a shortlist instead of a blank page.
We do not replace judgment. We cut the hours spent hunting for which feed would have shown the problem if anyone had known to look.
Six working rules
How the stack is supposed to behave in production.
Do not wait for a question
Classic BI assumes an analyst who already knows what to look for. That person writes SQL, waits, interprets, and repeats. Inputless Analytics inverts the workload: ingest first, then flag. Operators still approve the action. They just stop spending the morning guessing which query would have mattered yesterday.
Keep a working model
The Cognitive Substrate is the live map of how your organization classifies events and what it usually does next. An ontology with history. When two desks disagree about what an “incident” is, that disagreement is the bug. The substrate is how you make the vocabulary explicit.
Store the links
A row in a table is a fact. A relationship is why that fact matters. The Kinetic Graph keeps both: twenty-two node types, thirty-plus relationship types, in Inputless DB. You can see that a delay and a price move are connected, not merely that both numbers changed.
Keep it on the premises
If the data cannot leave, the software has to run where the data is. On-premise and private cloud are first-class. We do not need a copy in our cloud to “think.” That is a deployment constraint.
Update as the world updates
Scheduled reports freeze a moment that has already passed. The substrate refreshes as feeds arrive. A recommendation is only as current as the last ingest. If the line is down, the model is honest about being stale.
Answer “what now”
A dashboard answers “what happened, if you already knew where to look.” We ship ranked options: what to check, what to pause, what to escalate. The past is in the evidence trail. The product is the next step, signed by a person who can still say no.
Read the rest, or talk to us
If this matches how your desks actually work, we can walk through a feed and a decision trail.
Longer notes live at inputless.org. A working install lives on your network.
Longer notes at inputless.org