AI-Driven Automated Wet-Lab Research: Current State of the Art
What is the state of the art in automatic AI-based research with wet labs in the loop?
Read the report (PDF)Ask any question to receive a structured, fully cited research report. Our engine searches the web, analyzes sources, and backs up key claims with direct citations. Hosted entirely on European infrastructure, your queries and findings remain within the EU.
On search and extraction, open-weight models are now on par with the big US models. The difference: your research stays strictly under EU jurisdiction and is never used to train AI models.
🇪🇺 All EU data processors · 🇩🇪 Hosted in Germany · 🚫 Never used for AI training
Reports from real runs: the question, a structured answer with citations, and the sources behind it.
What is the state of the art in automatic AI-based research with wet labs in the loop?
Read the report (PDF)What is “Deep Research”, what are the legal and practical reasons to run it on EU infrastructure, and where do those reasons not apply?
Read the report (PDF)Examine the current legal framework governing asteroid mining under international space law and conflicting interpretations of the Outer Space Treaty.
Read the report (PDF)Führe eine umfassende, evidenzbasierte Meta-Analyse zur Quantifizierung der tatsächlichen Effizienzgewinne (in Prozent) durch den Einsatz von Large Language Models …
Read the report (PDF)Regulated industries, public institutions, healthcare providers, legal teams, and companies with works councils often cannot send sensitive questions through US-hosted AI services. Deep Research is built specifically to address these requirements.
CIATA GmbH in Berlin is the sole data controller. Reports are deleted automatically after 30 days, or with your account whenever you delete it. You retain full rights of access, rectification, and erasure under Art. 15–17 GDPR.
Every report is designed to meet the Art. 50(2) transparency obligation for AI-generated content: a visible "AI-generated" label plus an invisible, machine-detectable watermark (VSRMark) containing no personal data.
No third-party cloud storage. Everything is stored on dedicated servers in Germany (Hetzner) — never in a US-owned cloud. Connections are TLS-encrypted and backups remain within the EU.
Every provider we use — for AI inference, search, hosting, email, payments, and analytics — is based in the EU. Data processing agreements are currently being finalised. By default, no US service touches your research.
| Partner | Purpose | Location | Certifications & Trust Stamps |
|---|---|---|---|
| AKI.io | AI inference (LLM) | 🇩🇪 Frankfurt | Suitable for professional secrecy holders (§ 203 StGB), Zero-Retention Policy, ISO 27001 / TÜViT TSI / DIN EN 50600 |
| Staan | Web search index | 🇪🇺 EU | GDPR "Zero-Data" Design, European Sovereign Search Index |
| Hetzner | Infrastructure hosting | 🇩🇪 Germany | DIN ISO/IEC 27001:2022, BSI C5, KRITIS-V / NIS-2, PCI DSS |
| Mollie B.V. | Payment processing | 🇳🇱 Amsterdam | SOC 2 Type 2, PCI-DSS Level 1, ISAE 3402 Type 2, Regulated Institution (DNB/FCA) |
| Billie | Business invoicing (B2B) | 🇩🇪 Berlin | ISO 27001, BaFin Regulated |
| Qonto | Invoicing | 🇫🇷 Paris | ISO 27001, SecNumCloud Support, ACPR Regulated, PCI DSS / DORA |
| Scaleway | Transactional emails | 🇫🇷 Paris | ISO/IEC 27001:2022 (cloud infrastructure), HDS |
| Plausible | Cookie-free analytics | 🇪🇺 EU (self-hosted) | 100% Cookie-free / PECR Compliant, Open Source, Self-Hosted Data Sovereignty |
A structured research workflow — decompose, search, evaluate, verify, and cite — conducted entirely within the EU.
Your question is processed by AKI.io in Frankfurt using a German-hosted language model held in volatile memory only, with no logging and no training on user prompts (data processing agreement currently being finalised).
The engine derives targeted search queries and sends only those queries — never your raw question — to Staan, an EU search index based on Ecosia and Qwant. Queried websites never receive your identity or IP address.
Our server in Germany fetches each source directly, extracts key findings, and generates a cited report. All data is stored locally in Germany and deleted automatically after 30 days.
One credit runs one full research report. Pay by card — processed by Mollie (Amsterdam, NL) — or, for businesses, by invoice with Net-30 terms via Billie (Berlin, DE). Every payment provider is EU-based.
Volume pricing — credit packages offer a lower cost per report.
No offshore holding structures. Full documentation on data processing and sub-processors is available openly.
CIATA GmbH — Arnimallee 7, 14195 Berlin — a spin-off from Freie Universität Berlin. Full Impressum and contact details are always available.
Your questions, retrieved sources, and finished reports are never used to train models — neither by us nor by our AI provider. This is a contractual no-training commitment.
Research data is deleted automatically after 30 days. You can delete your entire account at any time, and single reports through the API.
We use cookie-free, self-hosted EU analytics (Plausible). No Google Analytics, no advertising pixels, and no cross-site tracking.
Documentation and legal details: Privacy Policy · Terms · Impressum · Compliance pack (ZIP)
No. Your questions, retrieved sources, and final reports are never used for model training — neither by CIATA nor by our AI provider AKI.io, which processes prompts in volatile memory only on a contractual no-training basis (data processing agreement pending execution).
All stored data lives on servers in Germany (Hetzner). There is no third-party cloud storage and no US-hosted database anywhere in the data path. Connections are TLS-encrypted and backups are encrypted and held within the EU.
A structured report with an executive summary, sections that answer your question, citations to the sources behind its statements and the full source list. Download it as PDF or Markdown, or share it with a public link.
Usually 10 to 20 minutes. You can close the page in the meantime: we email you when the report is ready (you can turn this off in your account).
You can ask in English or German. The report is written in the language of your question.
Before delivery, the report's cited paragraphs are checked against what their sources say, and statements that do not hold up are corrected or removed. The reports are still AI-generated and can contain errors, so check critical facts in the linked sources.
Publicly available web pages, found through Staan's European search index by default. Our server in Germany fetches each page itself. It cannot read content behind paywalls or logins, or your internal documents.
Yes. Every new account comes with three free reports.
You get the credit back automatically.
Reports are deleted automatically after 30 days. You can delete your account, with all its reports, at any time; through the API you can also delete single reports.
Questions that need many sources: market and competitor overviews, regulation, the state of research on a topic, comparisons of technologies or providers. It is not made for quick fact lookups.
Yes. Create an API key in your account and submit questions over a REST API. You can follow a run's progress as a live event stream and fetch the report as Markdown or PDF. Each report costs one credit, as in the web app. The interactive documentation is at research.ciata.io/docs.
Credit packages already lower the price per report. For larger volumes, an enterprise agreement or API terms, write to us through our contact form.
We implement the Art. 50(2) transparency obligations for AI-generated content: every report carries a visible "AI-generated" notice and an invisible, machine-detectable watermark (VSRMark). We assess the tool as limited-risk; formal risk classification remains a matter of legal evaluation.
AKI.io (AI inference, Frankfurt), Staan (web search, EU), Hetzner (hosting, Germany), Mollie (card payments, Amsterdam), Billie (business invoicing, Berlin), Qonto (invoicing, Paris), and Scaleway (transactional email, Paris). Every provider is EU-based; data processing agreements are currently being finalised.
Yes. We maintain a Data Protection Impact Assessment, a legitimate-interest assessment, and a technical briefing prepared for legal review. Contact privacy@ciata.io.
Create an account and run your first EU-sovereign research report in minutes.
AI-powered research reports — comprehensive, cited, and GDPR-compliant.
AI-powered research reports — comprehensive, cited, and GDPR-compliant.
The email contains the report's title, a short summary, a link and the date the report is deleted.
Programmatic access to the Deep Research API. Keys are shown only once when created.
| Label | Created | Last Used | |
|---|---|---|---|
| Loading… | |||
Permanently delete your account and all associated data. This action cannot be undone.