Amplytics: between your instruments and your decisions

01 · Speed

Optimized results, with >50% fewer experiments.

One cycling run blocks a channel for 1–2 weeks. Amplytics proposes the next experiment with the highest information gain, so every week of channel time moves you forward.

02 · Understanding

Gets smarter with every cell you measure.

An active learning model that learns continuously from your data, surfacing which formulation and process parameters truly drive performance, without needing a data scientist.

03 · Memory

Your data & knowledge, never lost.

Structured so your data is AI-ready by default, queryable in plain language by Claude or your own tools, so what your team learned years ago is never more than a question away.

What it is

The data platform accelerating battery R&D.

Structured data

One database, every cycler. Whatever format your data comes in, BioLogic, Neware, Arbin, and more, it's unified into a single structured system.

Automated quality control

Every run checked automatically. Anomalies and drift flagged as data comes in, not discovered months later.

Active learning

Suggests your next experiment. A continuously updating model recommends the next highest-information experiment, no data science degree required.

AI-ready datasets

Every project stored in a format queryable by Claude, OpenAI, or your own scripts, from day one.

app.amplytics.org / C-F01-C03 / upload

Amplytics upload: the file is recognised, measurement type determined, missing data calculated, filters applied and written to the database

0.4s Format, protocol and channels detected automatically. Extracts, compares and matches metadata autonomously.

app.amplytics.org / C-F03-C03 / QA & measurement details

Amplytics cell detail: voltage vs capacity chart, cell info, QC insights with flags, and the measurements table

QC Quality-checked on upload, every flag tied to the exact cycle that caused it.

app.amplytics.org / Project datasets

Amplytics project dashboard: cells table with capacity, CE and cycle counts across the NMC622 cathode screening campaign

Structured Every sample in the workspace, comparable on shared axes: clean, queryable, model-ready.

app.amplytics.org / NMC622 / C-0147 / insights

Amplytics decision support: the insights panel proposes rate performance on the NMC622 series as the next measurement, with expected information gain

Next run The campaign model proposes the highest-information next measurement, reasoning attached.

How it works

Watch a measurement become intelligence.

One raw file is all it takes. Scroll to follow it through the loop.

01UploadDrag in a raw file: BioLogic .mpr, Neware .xlsx or .csv. No vendor software, no export step.
02Raw signalEwe vs capacity, parsed straight off the instrument file.
03ValidatedCycles screened for anomalies, CE outliers flagged, exclusions under your control.
04StructuredNormalised into one comparable, queryable dataset, every value linked to its raw file.
05InsightA live model of your KPI across every parameter, not just the cells you've run. Confidence intervals, KPI maxima and minima, and a proposed next experiment, all read directly off that model.

SiGr-anode_C03.mpr UPLOAD

Drop a raw measurement to begin

.mpr · .xlsx · .csv · straight from the instrument

SiGr-anode_C03.mpr

Insight ready. Best result so far: 4.2% binder · 180 MPa calendering · 1.8 wt% additive (fade 0.3%/100 cyc). Next run: 4.5% binder · 165 MPa · 2.0 wt% additive.

scroll to transform

Ask your data

And once it's structured, it doesn't stop there. Ask anything, anytime.

Your workspace answers in plain language, in Claude or your own scripts, with every number linked to the raw file behind it.

Claude · Connected to Amplytics.org

Claude answering a question about hard carbon porosity in a sodium-ion anode project, with the answer citing the correlation across 14 cells and provenance chips linking the project, cells, and campaign

The problem

Right now, your best data dies in a folder.

The bottleneck isn't running experiments: it's trusting, comparing and acting on what comes out.

Without Amplytics

Every group exports a different format. Parsing scripts rot and break with each firmware update.
Quality control is manual and inconsistent. Bad cycles slip into conclusions and nobody notices for weeks.
Experimental drift is caught late. By the time you see it, you've burned weeks of channel time.
Every tool creates another silo. Cycler software, ELNs and spreadsheets each hold a fragment, and none of them talk to each other.

With Amplytics

One ingestion path, maintained by us. Data from any instrument: mapped and transformed to the same standard schema.
Every number comes with its uncertainty. Confidence, significance and anomaly flags: tied to the exact cycle that caused them.
The next experiment proposed from data. The data from your cells recommends what to measure next, avoiding wasted channel time.
Deep integration into YOUR system. Adapted to your IT, security, software and governance needs. No yet another data silo.

Founding lab

"Deciding what to measure next always felt ambiguous — we lost weeks on samples that added no real insight. Seeing our cells as one mapped parameter space changes how we plan experiments."

Christian Prehal

Co-founder & Principal Investigator

>0%

Faster optimization of material parameters

Of your lab's data in one database, every result traceable to its raw source

0min

From question to project report, on data already in your workspace, in Claude or your own scripts.

Battery R&D has two silent costs: channel time that teaches nothing new, and expertise that walks out the door.

01Every major materials company will plug AI into their R&D data. The labs whose data is actually usable will benefit first.
02Data you can't learn from is channel time already lost. If a result can't be reused or compared later, you've paid for it twice.
03Every campaign becomes a permanent asset, or it's gone. Structured, it compounds. Unstructured, it's a one-time result filed away with the report.
04Integration debt keeps growing. Every campaign that lands in yet another spreadsheet makes the eventual cleanup bigger.

Capabilities

What makes it powerful.

Underneath the loop sits the depth a battery R&D team actually relies on.

Protocol recognition

GCPL, OCV, EIS and more, detected straight from the instrument file. No manual tagging. Protocol, steps and metadata are read the way the cycler wrote them, so every measurement lands correctly classified from the first upload.

Agent-ready by design

A live MCP server and public API: your AI agents query the same provenance-backed numbers you see. Ask a question in Claude, get an answer grounded in your lab's actual measurements, not a copy-paste export.

Bayesian optimization at the core

Every recommendation comes from a Gaussian process fitted to your measurements, the same principled approach used in materials discovery and drug design. Reliable at 10–15 experiments, not hundreds, so it works from your very first campaign.

Every correlation, with confidence

See how any two parameters relate, with a clear sense of how certain the answer is. Statistically checked, with per-cell drill-down: real uncertainty, not a guess.

Shared workspace

Projects, cells and measurements your whole team reads from. One shared source of truth that doesn't expire: a question about that cell someone measured 5 years ago? Just one question away.

Audit trail

Every computed value links back to the raw file, measurement and analysis step it came from. When a number looks surprising, you trace it to its source in seconds, and cite it in a report with confidence.

Secure by design

Role-based access, invite-only workspaces and your data isolated per workspace. Sign-in and permissions run on hardened, industry-standard infrastructure: your results are visible to your team and no one else.

Your custom infrastructure

Amplytics adapts to your hosting, database, storage and backup requirements: on-premise on your servers, managed hosting, or private cloud. Our setup service integrates it with the systems you already run.

Who it's for

Built for the people who live in the data.

From the bench to the boardroom, everyone works from the same source of truth.

Two engineers reviewing a pouch cell beside voltage curves

Battery labs

Drop a file, get quality-checked charts, auto generate KPIs, talk to your data with AI.

Researcher preparing coin cells at a glovebox in an academic battery lab

R&D teams

One workspace, several researchers, multiple campaigns, and connected data.

Data lead at a dual-monitor workstation beside instrument racks

CIOs & data leads

One governed, AI-ready platform: your needs, your infrastructure, your security. One source of truth.

Battery scientist loading a coin cell into a multi-channel cycler

Academic labs

Rigorous, citable analysis: every number traceable from raw file to claim.

How teams use Amplytics.

Optimization

Materials & process optimization

Rank which formulation and process variables actually move performance, energy density, rate performance and more, then let the model propose the next batch.

Lifetime

Lifetime & degradation studies

Track capacity fade, cycles-to-80% and CE over hundreds of cycles, with model-fitted fade laws, not eyeballed trends.

Formation

Formation & protocol optimisation

Compare formation recipes side by side and converge on the protocol that holds capacity best.

Better cells, fewer cycles: faster results.