KI-Snack Week 2026: learning for interactive and agentic systems
Learning resource · announced
A learning and ecosystem resource, not a blanket claim of a realised interactivation case. Announced for 9–13 November 2026: 9–10 November at LABCampus, Munich Airport, and 11–13 November online. Public-source review and Atlas publication: 27 October 2026. The event is still in the future at that date.
The offerings below are organised by their learning relevance to the existing Atlas taxonomy. Each explanation is an editorial classification of announced content, not independent evidence of deployment or success. These offerings do not count individually as cases, core dossiers or follow-up articles. Most sources come from the organiser and speakers; several pages from one provider do not constitute independent confirmation.
The programme and certification pages provide no reliably consistent publication date. Source publication is therefore recorded as undated, accessed 27 October 2026; event dates are 9–13 November, and Atlas publication is 27 October. Individual slots may change.
43 curated offerings in one entry; no additional core dossier.
Agent teams, sales and CRM
The AI agent as a colleague ↗
An introduction to delegating tasks and process steps. Atlas relevance: subsequent system actions and human oversight; the keynote itself does not demonstrate production use.
Sales agents with approval gates ↗
An announced demonstration of inbox triage, qualification, product matching and follow-up. The relevant loop is signal → next sales action → approval. Failure reports and customer reactivation remain speaker claims.
AgentSpot: controlled workflows ↗
The announced example connects a business goal, context, tools and approval. It supports comparison of authority and traceable decisions; it does not independently establish a reliable customer deployment.
HubSpot Breeze, Surfe and Claude ↗
A target-account list, enrichment and prospecting are presented as a connected sales workflow. Data alter prioritisation and the next CRM step. Claimed pipeline effects are not independently measured outcomes.
LinkedIn campaigns and CRM ↗
A learning resource on turning contact and campaign signals into CRM processing. An integration alone does not establish interactivation. The on-site version is linked as a variant, without double counting.
My team of AI agents ↗
An announced workshop building specialist agents for research, marketing, meetings and content. The question is how roles, handovers and results alter the next step; successful participant implementations have not yet been demonstrated.
Agents across three businesses ↗
The announced live demonstration covers LinkedIn, media buying, podcast publishing, CRM and project coordination. These process handovers are relevant; ROI examples and claimed operation remain provider statements.
Gradion: CRM with an agent at its core ↗
An announced account of an agent-centred CRM. Margince describes source-linked changes and approvals. Its repository explicitly says source-available, not open source; independent deployment evidence and a robust chronology are still needed for a core dossier. The product site announces a November 2026 launch; this is not a confirmed market launch.
Customer journey with mail, chat, voice and RAG ↗
The announced implementation links channel requests, company knowledge, follow-ups and a dashboard. Context may influence processing and the next action. The programme documents the description, not independently verified integration, local storage or outcomes.
Voice agent at wirkaufendeinauto.de ↗
An announced account of multi-country telephone service, according to the speaker. The research questions are request → processing or handover and actual system authority. Availability, cost and queue claims remain provider statements.
Adaptive customer guidance and agentic commerce
Neocom: Agentic Storefronts ↗
A learning resource on adapting product selection and guidance to dialogue signals. Neocom describes a shop agent, shopping quiz and configurator. Provider interviews do not replace independent confirmation of a specific deployment or its effects.
Context-aware Shopping Experience ↗
An announced shopping companion using shop data and the current customer context for navigation and subsequent actions. Relevant to adaptive selection and agentic actions; autonomous purchases or realised conversion effects are not inferred.
Commerce in an age of purchasing agents ↗
A conceptual outlook on assistance, delegation and potential machine procurement. The linked cima publication is a scenario and strategy resource, not evidence of a fully realised agent economy. The paid full text was not treated as reviewed.
The Future of Search ↗
Learning context on visibility in AI answers and measuring recommendations. The relevant mechanism is changed selection, rather than search marketing alone; claimed revenue contributions remain speaker statements.
GEO: brand and product visibility ↗
Examines questions, sources and product data as inputs to generative recommendations. Useful for studying visibility allocation; guaranteed influence on rankings or purchases is not established.
Measuring AI traffic and customer enquiries ↗
A methods resource on recommendation → visit → enquiry and attribution limits. A measurement dashboard alone does not change an offering; this remains a learning resource.
Newsletters: audiences, tests and offers ↗
Announced examples of segmentation, subject-line testing and adapting offers. The question is whether responses actually steer the next communication. Audience and revenue claims are not adopted as verified effects.
AI along the guest journey ↗
Learning transfer for enquiries, tailored offers, guest feedback and resource planning. Relevant where guest signals change selection or the next processing step; text generation and general efficiency promises do not establish a separate case.
City marketing and destinations
Neumarkt: interactive city-centre applications ↗
An announced account of connected online/offline campaigns, prize draws and a knowledge chatbot. A case still needs an identified application, feedback loop, resulting system decision and independent confirmation; content automation alone is insufficient.
Gastein: location twin ↗
An announced winter 2026/27 campaign using a captured outdoor location. Production configuration is a research question. Changes driven by user signals, independently confirmed deployment and savings remain unverified; no separate practice case.
Implementation, data and organisational learning
Research, documents and agent workflows ↗
Announced examples including errors and oversight requirements. Useful for identifying decision handovers; a tool demonstration or document summary alone is not counted as an interactivation case.
From proof of concept to operations ↗
Background on data, responsibility, interfaces and scaling. Relevant to assessing agentic-system maturity; the format proves neither a specific production launch nor commercial success.
From Zero to Shipping AI ↗
An announced account of project sequencing, internal capacity and early governance. Relevant to organisational learning around agentic production; reported results are not presented as independently verified.
Simplifying before adding agents ↗
The announced framework distinguishes simplification, standardisation, automation, assistance and agents. It supports checking whether a meaningful feedback loop exists; the consulting framework itself is not a practice case.
ERP data as a foundation ↗
A learning resource on consistent data for service and workflows. Relevant to the provenance of machine signals; introducing ERP or cleaning data alone does not meet the interactivation definition.
From individual use cases to AI adoption ↗
Moving individual assistants and agents into organisational workflows. Relevant to implementation, roles and learning loops. The direct URL retains an older Google Ads slug; the current visible programme title is used.
AI leadership rather than tool proliferation ↗
Learning context for selecting initial use cases and supporting adoption. Helps assess implementation conditions; it does not demonstrate that a particular business model has been interactivated.
KAMIE / HAPEKO: transferring decision knowledge ↗
The programme claims a production recruitment process built from expert knowledge. The research question is knowledge → next process decision. Architecture, human responsibility and independent deployment evidence remain open; no outcome or suitability judgement is made.
Vibe coding: a lead-tool demonstration ↗
An announced on-stage lead-tool build. Relevant as learning about tool production. A quick demonstration establishes neither reliable operations nor data-driven allocation or successful sales.
Governance, approvals and limits
Owning knowledge and switching tools ↗
Learning about portable context libraries and provider dependence. Atlas relevance: control over the knowledge base and switching systems, without implying automatic learning or migration.
AI Dependency Footprint ↗
An announced framework on AI dependency, disclosure and reversibility. Relevant to evaluating human control; scientific validity and effectiveness are not inferred from the programme description.
AI platform: data handling and workflows ↗
An announced platform for enquiries, offers and documentation. Relevant to data access and controlled actions. GDPR compliance is a provider claim; German hosting alone is not treated as compliance evidence.
Legal tasks when integrating AI ↗
Background on responsibilities for business AI. Relevant to limits on delegated system decisions. The Atlas records an announced learning resource and does not certify a system as legally compliant.
HR compliance and legal AI ↗
An announced resource on data-minimising applications and human decision authority. Supports examining which steps may be automated; it is not evidence of a lawful automated personnel decision.
Cybersecurity and social engineering ↗
Phishing prevention and detection complement security assessment of agents with mail and tool access. The workshop is governance context, not a separate interactivated business model.
mintys: detecting and labelling AI content ↗
An announced tool demonstration for detection and labelling. The relevant loop is classification → next publication action. Legal reliability and detection quality are provider claims; blanket legal certainty and independent error rates are unverified.
INU learning path: seven core formats
01 Your AI compass ↗
Orientation on AI systems and maturity. Supports distinguishing announcements, assistance and actual system actions. The time differs between the programme and certification page.
02 Understanding AI ↗
Foundations covering models, context, RAG and hallucinations. Relevant to distinguishing knowledge retrieval from autonomous action and reliable outcomes.
03 Trust, checks and responsibility ↗
Output checks, data protection and human responsibility. Relevant to approval gates and limits on delegation to agents.
04 From prompt to agent ↗
From answering to acting: tools, automation and control. Directly relevant to subsequent system decisions and human approval.
05 AI in action ↗
Transferring examples into use cases. The programme says 13 November, the certification page 12 November. No definite slot is adopted; examples do not establish participant success.
06 Society and leadership ↗
Values, responsibility and leadership accompany changes to human and machine roles. A context resource, not a separate practice case.
07 Scaling AI ↗
Learning about strategy, data integration, governance and skills. Relevant to operational maturity and responsibility; confirm eligibility and time directly before booking the certificate.
INU AI competence pass: learning and assessment record
According to the organiser: at least six explicitly eligible formats, with at least 75% live attendance per session, a session code and a transfer question; recordings do not count. The final test has 15 questions, an 80% pass mark and two attempts in total. A bilingual PDF certificate with QR verification is promised within five working days. No ECTS credits or academic degree. The paid add-on requires an event ticket providing access to at least six eligible formats. It is not blanket evidence that a company or AI system complies with the law.
Attendance and assessment rules ↗ · Linked information page ↗
Dates and source discrepancies
Freshly observed discrepancies: Your AI compass is listed at 09:00 in the programme but 10:00 on the certificate page; AI in action is on 13 rather than 12 November; Scaling AI is at 13:30 rather than 13:00. Format 07 describes INU certification, while the eligibility badge is absent from the online list. Area counters and visible individual offerings do not consistently match. The on-site LinkedIn/CRM variant describes itself as online. These records are not combined into supposedly definitive times or certificate guarantees. Check detail pages and confirm with the organiser before booking.
No time shown here is presented as a definitive organiser confirmation.
Not included as separate cases
The opening, general future and leadership talks, networking lunch and AI meetup are not promoted to practice cases. Ticket categories, checkout, sponsors, the speaker directory and call for papers are not separate interactivation offerings. Programme blocks merely group their individual contributions. General social-media and influencer talks and the session still phrased as a submission, “Your agenda, our AI expertise”, provide no sufficiently specific feedback loop. The older indexed “The agent-operated company” was not confirmed in the current visible programme; its direct page returned no description during review. The older engineering session “Agentic AI without a Python team” is not a confirmed additional online slot.
Practice dossiers: evidence review remains open
The notes below identify missing evidence. The actual dossier drafts remain unpublished, outside the sitemap and case count.
- Neocom: provider descriptions explain adaptive selection; a dated product chronology and independent confirmation of a specific deployment are missing. The founder interview of 1 October 2026 remains self-report.
- Margince/Gradion: repository and product site describe signals, changes and approvals; both are primary sources. No independent deployment or outcome evidence. Do not relabel source-available as open source. Current product site: launch announced for November 2026; the event is still in the future.
- Innov-AI-tive, the HubSpot workflow and business agents: demonstrations are announced; outcomes and errors are speaker claims. Verify an implemented loop and independent confirmation separately for each system.
- Mail/chat/voice/RAG and voice service: meaningful programme descriptions, but independent architecture, authority and deployment evidence for specific implementations is missing.
- Neumarkt: interactive campaigns and a chatbot are speaker claims. Association and city pages establish the person or organisation, not the AI feedback loop; they do not independently corroborate the mechanism.
- Gastein: an announced campaign demonstration; signal → production change and independent deployment evidence remain open. Generated images alone are insufficient.
Each separate practice dossier still requires a demonstrated feedback loop and at least two reliable sources, preferably a dated primary source plus independent corroboration. Unresolved interpretations remain subject to editorial review.
Sources and review date
All linked offering pages: provider/organiser sources, publication date not stated, checked on 27 October 2026. The founder interview date is stated separately. Event dates and Atlas publication are distinguished above.
- Organiser: event and event dates; undated ↗
- Organiser: complete on-site and online programme; undated ↗
- Direct access to online offerings ↗
- Organiser: INU learning path, attendance and test; undated ↗
- Competence-pass information page linked from the event; undated, not independent evidence ↗
- INU: university recognition; this does not establish accreditation of this short course ↗
Dates and evidence scope
- Recorded event / reference date
- 2026
- Atlas first publication
- Not separately established
- Source review in this audit
- Not separately established
The editorial update date is not a claim that every source or current product feature was rechecked. Unknown dates remain open. Source publication dates are listed with the evidence above.
Structured evidence and geography register · Unclassified source roles and unassigned origin/market fields are explicitly marked.