01
Hours lost on repetitive processing
Format conversion, peak fitting, plotting, literature comparison — done by hand, every time.
AI-native workspace
for materials & physics
researchers.
SciPhys turns raw XRD, Raman, SEM and spectroscopy data from materials and physics labs into publication-ready figures, scientific interpretations, and the next experiment to run.
Instrument coverage
XRD · Raman · SEM
Raw data → analysis
5 min
Figures & reports
Publication-ready
Lab research memory
Private
The Problem / 02
Experimental data is the largest unstructured asset in the AI era.
Materials, semiconductor and physics labs generate massive volumes of instrument data every day. Today it is scattered across instrument PCs, Excel files, Origin projects, local folders and paper drafts — invisible to AI, unusable as memory.
02
No reusable lab memory
Failed runs, parameters and instrument settings are forgotten the moment a student graduates.
03
Trial-and-error is expensive
Every experiment burns materials, instrument time and human hours — without a system that learns.
04
Generic LLMs cannot read your data
ChatGPT can read papers. It cannot read your XRD, Raman, SEM or electrical measurement files.
The bottleneck of scientific discovery is not the lack of papers — it is the messy, invisible state of real experimental data.
Product / 03A
Drop in raw instrument data. Get a scientific report.
Upload an XRD, Raman, SEM or spectroscopy file — or sync directly from the instrument PC. SciPhys runs QC, fitting, phase identification and literature comparison in minutes, not hours.
- Format-agnostic ingestion — XRD, Raman, SEM, FTIR, UV-Vis and more.
- Auto QC, peak detection, background subtraction and curve fitting.
- Phase, structure and parameter extraction with cited reasoning.
Product / 03B
A workspace that remembers every experiment.
Every sample, condition, parameter, figure and decision is committed to a private Research Memory — your lab's second brain, growing with every run.
- Structured timeline of samples, runs, parameters and outcomes.
- Failed experiments preserved — with the lessons attached.
- Searchable across people, projects and years of work.
Workflow / 04
From raw signal to scientific decision — in one continuous loop.
Step 01
Ingest
Drop in raw XRD, Raman, SEM, spectroscopy or electrical files. Or sync directly from the instrument PC.
Step 02
Analyze
Auto QC, peak detection, background subtraction, fitting, phase identification and parameter extraction.
Step 03
Visualize
Publication-ready figures in Nature, ACS, APS or IEEE styles — exported in one click.
Step 04
Compare
Match against literature, prior runs and your lab's private knowledge base, automatically.
Step 05
Recommend
An agent proposes the next experiment — which variable to change, why, and what to expect.
Step 06
Remember
Every analysis is committed to your lab's private Research Memory, growing with every run.
Live Demo / 05
Raw data in.
Scientific insight out.
Watch a real XRD file run through the pipeline. No slides, no marketing — the same output a researcher gets in their workspace.
sciphys / xrd-agent
run_id: 0x9f2c · 2026-04-28Input file
sample_B17_xrd.raw· 2.4 MB · Bruker D8
2θ range
10°–90°
Step
0.02°
Time/step
0.5 s
Source
Cu Kα
Status
Idle
Click Analyze to run the pipeline
01 · Data quality
98 / 100
Clean baseline, low noise. Ready for analysis.
02 · Detected peaks
14
Auto-detected after background subtraction.
03 · Phase
Perovskite-like
ABO₃ structure, cubic Pm-3m.
04 · Peak shift
+0.12°
Lattice expansion vs. literature reference.
05 · Literature match
8 papers
Matched to known perovskite phases.
06 · Next step
Lower-temp annealing
Reduce strain — try 380°C for 30 min.
Demo data shown for illustration. Your files stay private to your workspace and are never used to train shared models.
Capabilities / 06
Built for the way real labs actually run.
A focused workspace for materials and physics teams. Six surfaces, one continuous research loop.
Project Workspace
One workspace per project — samples, runs, instruments, figures and decisions, all versioned together.
Perovskites, 2D materials, battery cathodes — every project gets a structured home, not a folder of files.
AI Analysis
Peaks, phases, parameters — explained.
Auto QC, fitting and phase identification with cited reasoning.
Figure & Report
Publication-ready, in one click.
Nature, ACS, APS and IEEE-style figures, methods and results drafts.
Research Memory
Your lab's private second brain.
Every sample, condition, parameter and failed run becomes structured, searchable, reusable knowledge.
Next Experiment Agent
The variable to change next — and why.
Helps you propose, screen and validate hypotheses faster — never replaces the scientist.
Solutions / 07
One product.
Three ways in.
From a single researcher to a regulated R&D org — the same workspace, with controls that scale to the buyer.
01 · For Researchers
Free
Your personal instrument copilot.
For PhDs, postdocs and individual scientists. Drop in files, get publication-ready figures, build your own private memory.
- Personal workspace, unlimited samples
- Up to 50 analyses per month
- Nature, ACS, APS & IEEE figure styles
- Community templates & exports
02 · For Labs
Subscription
One workspace for the whole group.
For PI-led research groups and university labs. Sync directly from instrument PCs, share Research Memory across the team.
- Team workspace, unlimited members
- Unlimited analyses & instruments
- Shared Research Memory & search
- Priority support, weekly model updates
03 · For R&D Teams
Enterprise
Production-grade, inside your network.
For semiconductor, battery and advanced-materials R&D. On-prem or VPC deployment with the controls enterprise data demands.
- On-prem / VPC / data residency
- SSO, RBAC, audit trail, SLA
- Custom integrations & instrument SDK
- Dedicated solutions engineer
Security & Trust / 08
Materials data is sensitive.
We treat it that way.
Built for semiconductor fabs, advanced-materials R&D and regulated labs — with the controls security and IP teams expect.
01
Private workspace
Lab data is isolated by default. Never shared, never used to train shared or third-party models.
02
Role-based access
Granular permissions per project, dataset and instrument. Scientist, reviewer, viewer, admin.
03
Encryption end-to-end
AES-256 at rest, TLS 1.3 in transit. Customer-managed keys (BYOK) available for enterprise.
04
On-prem & VPC
Deploy inside your network when sensitive material or fab data cannot leave. EU & US data residency.
05
Human-in-the-loop
Every interpretation is auditable, editable and rejectable by a scientist. Models propose; people decide.
06
Full traceability
Every figure and conclusion traces back to its raw file, parameters and model version. Reproducible by design.
Compliance roadmap
SOC 2 Type II
ISO 27001
GDPR
HIPAA-ready
Single sign-on
Why Now / 09
A narrow window.
A wide opening.
01
AI for Science is leaving papers behind.
Capital is moving from chat to discovery. Lila Sciences has raised about $550M to build AI Science Factories — the new frontier is real experimental data, not text.
02
Scientific data platforms are validated.
Citrine proves enterprise materials R&D SaaS. TetraScience proves the AI-native scientific data layer. Both markets are real, and growing.
03
Frontier models can now run lab workflows.
Multimodal reasoning, code generation and tool use are finally good enough to read instrument data and reason like a graduate student.
04
No one owns the instrument data layer — yet.
There is no Cursor, no Notion, no Figma for materials and physics labs. SciPhys starts where the data is born: at the instrument.
Roadmap / 10
Software-first. Always.
We do not build instruments. We integrate with the ones the world already trusts — and become the intelligence layer above them.
- Phase 01Today
XRD Agent for materials labs
A focused wedge: raw XRD file in, publication-ready figure, phase interpretation and next-experiment suggestion out.
- Phase 02Tomorrow
Multi-instrument lab workspace
Raman, SEM, AFM, FTIR, UV-Vis, electrical and thermal data — one project workspace, one team, one memory.
- Phase 03Enterprise
Private Research Memory layer
On-prem and edge deployment for semiconductor, battery and advanced-materials R&D — with audit, RBAC and data residency.
- Phase 04Ecosystem
Instrument vendor SDK & API
Embedded inside XRD, Raman, SEM and spectroscopy hardware — turning instruments from data sources into conclusion sources.
- Phase 05Endgame
AI-native scientific discovery loop
Closed-loop: model proposes → automated lab executes → data flows back → memory updates → hypothesis sharpens.
Vision / 11
Not an instrument company.
The intelligence layer above every instrument.
We are looking for materials, physics and semiconductor labs to join our early access program. Bring your hardest experiments. We will bring the copilot.
We'll never share your data. Lab pilots are private by default.