From trivial pursuits to
non-trivial answers.
We turn data into evidence, and evidence into decisions.
Two things, held with conviction.
- We give meaning to data that sits unused and underutilized in the world. Most of it is never read. We read its shape.
- We give builders a platform to deploy their own data-driven solutions. Fast, on their own terms, and in the open.
Four pipelines, one geometry.
- Build systems. Sites, applications, and the machinery behind them.
- Read data. Structured until it answers, and reproducible enough to defend.
- Score anomalies. We learn your normal and rank what breaks it.
- Plan what follows. Evidence in, a plan out. Being built now.
- Arm the builders. The tools underneath are open. Deploy on your own terms.
Because most results cannot be repeated.
- Reproducibility is rare. Floating-point drift, hidden state, and unversioned models make an answer hard to trust or run again.
- So it was made the foundation, not an add-on. Fixed-point arithmetic removes the drift; stateless services over a shared backend remove the hidden state.
- The payoff is a decision you can defend. Same inputs, same answer, on any machine, today or a year from now.
Without friction there is no shine.
We are not a spreadsheet consultancy. We are brokers of meaning and insight, agents of will and action.
We do not hoard IP. Like the industrial giants we admire, we open our inventions once they are matured and battle-tested by the pioneers we set out to gather. We are confident we can put our concepts to meaningful use, and we do not fear competition. The impetus to innovate was never comfort. We do things because they are hard, and we seek the few who appreciate that effort.
Open-source warranty: the tools ship as they are, with no liability for damages from their use. A channel is always open, and we will help where we can, but the responsibility for what you build is yours.
Gradient descent.
Big steps first, then smaller ones to refine what works. Three chapters to start.
-
1
Manifold Destiny
Released. The open-source launchpad for builders. Enter -
2
Return on Invariance
The next open-source chapter. Not scoped definitively yet. -
3
Inference
On the horizon.
Anonymous. No email needed. A minute of your read shapes what ships.
The launchpad. Here is what is already on the rails:
In the open.
Nierto Solutions, a Dutch one-person company (ZZP). KVK 82716099.
Founded and built by Niels Erik Toren. Two and a half years of building. Founding phase, pre-revenue.
Every request is read personally by someone who wrote the code.
Discovery engagements.
Most engagements run one to two weeks. Want a figure for your specific data? Tell us the shape of it and we send a quote within 24 hours. Regulated or sensitive data runs on-premises, in your own environment. Everything else can run on managed infrastructure.
- A scoping call. We look at the shape of your data and whether the method fits.
- One full run. The engine ranks what stands out across your dataset.
- A written read. What surfaced, what it means, and what is worth a second look.
The first call is free.