Plain-english assistant
Say what the signal should do. The assistant translates it into a precise, editable configuration — not a black box.
Describe it · don't hand-code it
TSERIZ turns a plain-english description into realistic, physics-grounded time series — reproducible, documented, and safe to generate. No scripts, no black box.
Simulate a motor at 1800 RPM with light bearing wear over 2 hours
No credit card to start · 100 free generations / month · CSV, Excel, JSON & InfluxDB export
01 · How it works
Every series is a composition of physically meaningful parts — you steer it in plain english or dial it in by hand, and the result is always reproducible from a seed.
Say what the signal should do. The assistant translates it into a precise, editable configuration — not a black box.
Every parameter is exposed — waveforms, transients, correlations, degradation, gaps and outliers — for exact control when you need it.
Aliasing guardrails and Nyquist checks run before generation, so the data you get is physically valid and has a known answer.
02 · Multimodal
Reach for whichever input is fastest. The assistant reads a written brief, a reference chart, or a spoken description and turns any of them into the same precise configuration.
Describe the scenario in a sentence and refine it in a conversation.
Drop in a chart or screenshot and the assistant matches its shape.
Talk through what you need and let the engine build it as you speak.
03 · Under the hood
The engine models the phenomena that make real sensor data hard — so the series you generate behave like the ones you'll meet in the field.
Second-order impulses — impacts, valve slams, pump trips — with realistic ringing and decay.
Cholesky-correlated channels from −1 to +1 that share one physical process.
Linear trends, drift, contiguous gaps and injected outliers — the messiness of real logs.
Nyquist checks catch under-sampling before you generate, not after.
04 · Use cases
Because you defined the signal, you know the ground truth — which makes it ideal for building, benchmarking and stress-testing.
Motors at a set RPM with harmonics and bearing wear — for fault detectors.
Multi-phase arrivals with realistic ringing and decay for event detection.
Correlated channels sharing one process, from −1 to +1.
Contiguous outages and outlier sweeps for gap-filling and detection.
Reproducible fixtures with a known ground truth to score pipelines against.
Generate labelled series at volume to pre-train and augment models.
05 · Pricing
Credits meter the AI assistant — the one real cost. Generation, channels and exports stay generous on every plan, and Free never charges you a cent. $1 = 30 credits.
Prices exclude VAT / sales tax, which is added at checkout based on your country. Businesses with a valid VAT ID aren't charged VAT.
$0
The adoption on-ramp — API access included.
$99$990
Unlimited generation for busy teams and pipelines.
Custom
For larger teams with custom volume and invoicing needs.
| Plan | Free | Team |
|---|---|---|
| Price | $0 | $99/mo |
| Credits / month | 100 | 6,000 |
| Assistant model | Haiku 4.5 | Haiku · Sonnet · Opus · Fable |
| Generations / month | 500 | Unlimited |
| Data points / job | 1M | 5M |
| Data points / month | 10M | 10B |
| API keys · rate | 1 · 60rpm | 5 · 1000rpm |
| InfluxDB export | — | ✓ |
| Over allowance | Hard stop | Metered · $1 = 30 credits |
| Overage limit | — | $200/mo · adjustable |
| Support | Community | Priority |
Credits meter the AI assistant — the deep-model work behind translating your description into a configuration. Generation, channels and exports never cost credits. $1 = 30 credits, and one Opus turn is roughly 4 credits.
No. Only assistant credits meter your bill — generating series, adding channels and exporting are covered by your plan's allowances.
Free hard-stops at its 100 credits a month, so you never get a surprise bill. Team includes 6,000 credits a month and keeps working past them: the extra is billed as overage at $1 per 30 credits, up to a monthly overage limit of $200 that you can raise, lower or set to zero in your billing settings. We email you at 80% of the limit and pause the assistant at 100% — generating series is never affected. Overage is invoiced with your next bill, or as soon as it reaches $200 if that comes first.
Yes. Every plan includes API access, and InfluxDB export is included with Team, alongside CSV, Excel and JSON on all plans.
Most synthetic predictive maintenance data relies on unrealistic statistical noise. This post explores why naive noise fails and demonstrates how to generate modeled bearing faults, offering a rigorous benchmark for industrial fault detection.
18 min read
Every data-driven project I've worked on has started the same way, a hunt for data.
8 min read
Describe the signal you need and generate it in seconds — reproducible, documented, and yours to export.