Guides and research on sovereign AI, GDPR-compliant meeting assistants, and sales call intelligence for European teams.
A cited, dated table built from the vendors' own DPAs and sub-processor lists: who processes your meeting audio, in which region, which sub-processors touch it, and how long they keep it. Covers Fireflies, Otter, tl;dv, Fathom, Metaview, Gong, Chorus, Avoma, Jiminny, Modjo, Clari Copilot, and Numi.
Metaview offers a DPA and SOC 2 and states it does not train on your data, but it hosts all interview data on AWS in the UK, outside the EU. What its own docs say, dated, and what the UK-hosting gap means for EU recruiting teams.
How recruiters use an AI meeting assistant across the funnel: hiring-manager intakes, candidate screens, and panel debriefs. What changes day to day, what to check on EU hosting and candidate data, and where Numi fits, stated honestly. Links the recruiting cluster together.
KI-Notetaker unterliegen der Mitbestimmung nach Paragraf 87 Absatz 1 Nummer 6 BetrVG, weil sie objektiv zur Verhaltens- und Leistungskontrolle geeignet sind. Warum das Recht ein Veto ist, was in die Betriebsvereinbarung gehört und wie Sie den Betriebsrat früh einbinden. Mit Checkliste und Muster-Gliederung. Keine Rechtsberatung.
Dürfen Recruiter Vorstellungsgespräche aufzeichnen und transkribieren? Praxisleitfaden zu Paragraph 201 StGB, DSGVO und BDSG: Einwilligung von Bewerbern, fertige Formulierungen, Aufbewahrung, Auskunftsrecht und was bei einer Absage mit den Aufzeichnungen passiert. Keine Rechtsberatung.
KI, die Bewerber bewertet, filtert, bepunktet oder rankt, ist Hochrisiko nach Anhang III (ab 2. Dezember 2027). Reines Aufzeichnen und Transkribieren sitzt in der Transparenzzone. Was ab wann gilt, welche Pflichten Recruiter treffen und wo Numi einzuordnen ist.
NIS2 and DORA push ICT third-party risk onto the buyer. Here are the 25 security questions to send any AI meeting or notetaker vendor, mapped to what the regulations expect you to collect, with Numi's answers to all 25 in writing, including where the honest answer is "not yet".
Welche Rechtsgrundlage nach Art. 6 DSGVO trägt die Aufzeichnung eines Meetings? Eine Matrix nach Teilnehmerkreis für interne, gemischte und externe Runden, mit Paragraph 201 StGB, Paragraph 26 BDSG und der Mitbestimmung des Betriebsrats. Vom Prinzip zur fertigen Vorlage.
Warum der Speicherort allein keine Souveränität schafft: der US CLOUD Act setzt an der Kontrolle des Anbieters an, nicht am Standort. Was BSI C5 und das neue BSI C3A trennen, wie der EUCS-Streit um die Immunität gegenüber fremdem Recht steht und woran sich Sovereignty-Washing erkennen lässt.
Wann die Stimme wirklich unter Art. 9 DSGVO fällt: der Unterschied zwischen Diarisierung, Verifizierung und Identifizierung, warum die eindeutige Identifizierung der entscheidende Zweck ist und wie sich Sprechernamen ohne gespeicherten Stimmabdruck vergeben lassen.
Wann ein KI-Notetaker eine Datenschutz-Folgenabschätzung braucht: die Schwellwertanalyse, die acht Kriterien der DSK und WP248, die vier Pflichtbestandteile und wann die Aufsichtsbehörde nach Art. 36 zu beteiligen ist. Mit kostenlosem Erforderlichkeits-Check.
Was in eine Betriebsvereinbarung für einen KI-Notetaker gehört: die 12 Pflicht-Bestandteile, warum § 87 Abs. 1 Nr. 6 BetrVG greift und die ehrlichen Grenzen einer BV. Mit kostenlosem Generator für einen fertigen Entwurf.
Stop matching candidates on titles and keywords. Match on trajectory: the career triggers that prove someone has already solved the problem you are hiring for.
A sourcing playbook for finding candidates who have already made the Seed-to-Series-A journey, using career triggers instead of job titles and Boolean strings.
Seven concrete signals in a career history that predict a candidate can handle hypergrowth, and how to spot each one in a LinkedIn profile.
Trigger-based matching is a way to match candidates to roles on career signals that predict fit, not on keywords or titles. A plain definition, examples, and how it works.
LinkedIn Recruiter and Boolean search match on titles and keywords, not on stage fit. Here is why that fails startup hiring, and how to source on career trajectory instead.
Most of the trajectory signal in a LinkedIn profile is in the dates, company stage, and sequence of moves, not the keywords. Here is how to read a profile for triggers.
A step-by-step framework for turning a client brief into a candidate shortlist by matching on career triggers instead of titles, with a repeatable worksheet you can copy.
An illustrative case study of trajectory matching: how a Series A fintech found a VP of Sales by screening for the stage transitions a candidate had already driven, not titles.
Culture fit rewards similarity and predicts little about performance. Here is the case for replacing it with trajectory fit, a concrete and more defensible hiring filter.
How to go from a long list of sourced profiles to a short list of real matches by systematizing trigger detection, and where an interview intelligence tool genuinely helps.
Interview intelligence records, transcribes, and structures what happens in interviews so hiring decisions rest on evidence, not memory.
Run the same structured phone screen every time so candidates are comparable and your notes drop straight into the submittal. Here is the full playbook.
Most teams look for a Metaview alternative when EU in-region data residency is a hard requirement. Here is what to evaluate and a shortlist.
Where does your interview data actually live? A recruiter's guide to EU vs UK data residency for AI interview notetakers, GDPR transfers, and what to check.
A structured interview scorecard template that reduces bias, plus how to capture the interview evidence behind each rating without breaking GDPR.
AI interview notetakers are not automatically high-risk under the EU AI Act. Annex III covers candidate scoring and ranking, not plain transcription.
Yes, you can record job interviews in Germany and the EU, but only with genuine, freely given, documented consent and a real no-recording option.
Take interview notes that record observable evidence tied to your scorecard, not personal impressions. Free interviewer template and a live note-taking method inside.
Run the debrief from the interview record, not from memory. Evidence mapped to the scorecard gives you a fairer, more defensible hire.
Most time-to-hire is lost between interviews, not at the top of funnel. Fix debrief lag, feedback chasing, and re-briefing to move faster.
Bad searches start with sloppy intakes. Here is how to run a hiring manager intake meeting, plus a copy-usable question checklist.
Stop asking managers to write feedback from memory. Give them a pre-drafted summary to approve in minutes. Nudge templates and worked examples inside.
Bewerbungsgespräche dürfen Sie nur mit echter, freiwilliger Einwilligung aufzeichnen. Leitfaden plus kopierbare Einwilligungsvorlage.
Most recruiting notetakers are great tools but host interview data outside the EU. The GDPR-first shortlist for EU and DACH teams.
Seven due-diligence questions every EU buyer should ask an AI meeting assistant before signing, mapped to the EU AI Act and GDPR. Storage versus processing, subprocessors, training, model control, Article 50 disclosure, deletion, and the DPA. Includes a printable one-page checklist.
Article 50 transparency has been in force since 2 August 2026, and most guides predate the Digital Omnibus and get it wrong. The deployer's playbook: what is live, who enforces it in Germany and the Netherlands, a practical checklist, and an avoid-list.
Seit dem 29. Juli 2026 ist das AI-MIG in Kraft und die Bundesnetzagentur ist Marktüberwachungsbehörde für KI. Was Unternehmen mit KI-Notetakern dieses Quartal prüfen sollten: Art. 50, Art. 4, Betriebsrat, § 201 StGB und Anbieterprüfung.
Voice-based emotion detection of employees has been a prohibited AI practice in the EU since February 2025. Where the Article 5(1)(f) line falls for call sentiment features, and why the input (how a voice sounds vs what the words say) decides it.
Trump v. Slaughter weakened the FTC independence the EU-US Data Privacy Framework relies on. A sober, no-FUD playbook for what happens to Otter, Fireflies and Gong if the DPF falls, and what to do now.
Teams and Google Meet are detecting and blocking third-party notetaker bots in 2026. What bot-based versus bot-free capture means, why the crackdown does not touch client-side capture, and the one question that now decides your meeting-AI buying decision.
On 2 August 2026 the AI Act's Article 50 transparency duties and fining powers go live, while the high-risk obligations slipped to December 2027. What actually applies to you as a deployer of a meeting-intelligence tool, and what to do before the date.
A source-quoted comparison of where Otter, Avoma, Jiminny, Modjo, Chorus by ZoomInfo, Fireflies, and Gong process your call data, and which keep it in the EU. Drawn from each vendor's own documents.
Otter.ai offers an Article 28 DPA with SCCs, but it is a US company hosting on AWS in the United States with every sub-processor US-based. What that means for an EU controller.
Google Meet shows a recording notification, but a notice is not consent. Here is what Google Meet actually does, and what GDPR still requires from you as the organiser.
You do not have to drop Google Meet for EU-resident meeting notes. The residency choice that matters is the notetaker, not the video platform. Here is why.
A GDPR-compliant AI notetaker for Microsoft Teams keeps its own transcription, storage, and analysis inside the EU. Here is how it works and where the EU Data Boundary stops.
Microsoft Teams does not ask for recording consent by default. Here is what Teams actually does, the policy you have to turn on, and what GDPR still requires.
A GDPR-compliant AI notetaker for Zoom keeps its own transcription, storage, and analysis in the EU. Here is how it works and where Zoom's own data still routes through the US.
Zoom prompts participants for recording consent, but the only way to refuse is to leave. Here is what Zoom actually does and what GDPR still requires from you.
You do not have to drop Zoom for EU-resident meeting notes. Zoom keeps account data in the US, but the notetaker layer is where EU teams have real, fast choice.
Is it really sovereign if the software is built with Claude, a US company? The tool that writes the code is not the tool that processes your data. Build-time versus run-time, with Numi's honest per-layer map.
The EU AI Act bans emotion recognition at work, adds transparency duties, and can make call-scoring high-risk. Here is what it means for AI meeting assistants and recordings.
A GDPR-safe AI notetaker for Google Meet joins your call as a participant and keeps transcription, storage, and analysis inside the EU. Here is how it works for EU teams.
How an AI meeting bot can join Microsoft Teams calls, transcribe and analyse them, and keep every step inside the EU with no US processors in the audio path.
Self-hosted meeting transcription runs the speech-to-text model on infrastructure the provider controls, so audio never leaves the EU. Here is how it works and why residency beats a US cloud API.
tl;dv stores meeting data in EU data centres, but its AI providers are US-owned and it offers a US processing option. Here is what that means for GDPR and data sovereignty.
Fathom is a free AI notetaker that stores all meeting data in the US, sends it to US AI vendors, and trains its own AI on de-identified transcripts by default. Here is where your data goes.
Otter.ai stores meeting data on US servers and trains its own AI on de-identified transcripts. Here is where your data goes, and the EU alternative.
Recording meetings with an AI notetaker in Germany needs all-party consent under §201 StGB and works-council sign-off. Here is what compliance actually requires.
Live status of every EU law that determines where AI can process private communications: Chat Control, AI Act Omnibus, Data Act, ePrivacy, EUCS, NIS2 and DORA. Primary-source citations. Updated when the radar fires.
A category-by-category guide to the tools that leave your communication scannable under Chat Control, and the end-to-end encrypted, EU-sovereign alternatives to switch to. Plus what changes if a stricter regime lands client-side scanning.
Lawful basis, per-country consent rules, data residency versus sovereignty, and retention. The practical 2026 checklist for recording calls and meetings compliantly across Europe.
Jurisdiction, not storage location, decides who controls your data. The strategic case for AI sovereignty in 2026: the CLOUD Act, supply-chain concentration, and why an EU data center is not sovereignty.
The EU-US Data Privacy Framework is still valid but under fresh legal pressure in 2026. A risk map of transatlantic transfers, the Latombe appeal and Schrems III, and how sovereignty removes the bet.
A buyer's scorecard to tell real AI sovereignty from sovereignty washing. Score any vendor on jurisdiction, sub-processors, model hosting, transfer mechanism, and audit rights.
Meeting memory, coaching at scale, and consistency are why AI meeting assistants are becoming essential. But the most common manager mistake is treating recording as coaching. It isn't.
A free notetaker is paid for in data. What the vendors' own privacy policies permit on training, US processing, and retention, with primary sources so you can verify every claim yourself.
Transcription is table stakes. The real category is grounded coaching: scorecards and feedback tied to what was actually said. The difference between a notetaker and a coach in 2026.
The definitive explainer on the EU's plan to scan private messages: the full timeline through the dramatic 9 July 2026 vote, what client-side scanning means for encryption, the error-rate math, and who is fighting on each side.
Chat Control normalizes scanning private communication at the edge. That makes one question decisive for every company: under whose jurisdiction is your meeting data actually processed, and who can reach it?
62% of EU companies pursue sovereign AI — but most still rely on US-owned cloud. Here's why a EU data center doesn't protect you from CLOUD Act subpoenas, and what actually does.
A fact-based breakdown of the five GDPR requirements for AI meeting tools — and how Fireflies (EU storage, US processing), Otter.ai (trains on your audio), and Gong (all data in the US) fail them.
Sovereign AI is a continuum across five dimensions: model, hosting, training data, operational control, and hardware. Here's a clear framework for EU executives navigating the term in vendor conversations and procurement.
The EU AI Act fines start at €7.5M. CRM tools are minimal-risk, chatbots are limited-risk, and HR/scoring AI is high-risk. Here's the exact compliance calendar for B2B sales and RevOps teams.
EU AI Act + GDPR now require specific contract clauses, audit logs, and data residency from AI vendors. These are the five questions that cover all of it — and what answers to accept or reject.
NIS2, DORA, and the EU AI Act now converge on one requirement: true data sovereignty. If you're in financial services or critical infrastructure, here's what that means for your AI vendor stack.
GLBNXT deployed a sovereign AI platform in 30 days vs 12-18 months for hyperscalers. Here's what the EU sovereign AI market looks like in 2026 and which companies are building the infrastructure layer.
US vendors offering EU data centers still can't protect your call recordings from CLOUD Act subpoenas. Here is what the law actually says, why SCCs don't close the gap, and what actually protects EU sales teams.
The best automated coaching tools for call center agents use call recordings to deliver real-time and post-call coaching at scale. Here are the top options in 2026 ranked by fit, accuracy, and time to value.
Most contact center platforms handle routing and recording but fall short on automated coaching and agent scorecards. Here is how to find the one that closes that gap in 2026.