Live meetups

Events

Our sessions bring 20-30 people together to share real use cases and talk through what's actually working - and what isn't. We use Luma to organize and RSVP.

Quick links:Upcoming events Past events

Upcoming events

Prefer to follow on Luma directly? Open our calendar →

Past events

Anonymized summaries of what members have presented at previous meetups.

6th edition - September 24, 2026 5th edition - September 10, 2026 4th edition - July 7, 2026 3rd edition - June 4, 2026 2nd edition - April 30, 2026 1st edition - April 27, 2026

6th edition

September 24, 2026

Automated B2B lead generation with a human in the loop

An automation specialist presented two AI-assisted lead-generation workflows built on one principle — "AI thinks, automation moves, and the CRM remembers" — where AI is reserved for judgment (researching a company, assessing fit, drafting a message) while simple rules handle everything deterministic (waiting, updating the CRM, stopping a sequence on reply), all anchored to a single CRM "system of record." The ICP-based flow builds a niche prospect list from a local business directory (adapting to a smaller market where Apollo/Sales Navigator don't have coverage), then uses AI to qualify leads and record an outreach angle before an automation enriches contacts and pushes them into a sequence; the signal-based flow starts from a real public buying signal, chains AI agents to draft and QA a comment and DM, and routes everything through a Slack thread where the team gives feedback or approves before anything goes public. Once a meeting is booked, the same automation updates the existing record, briefs the team, and captures needs, objections, and next steps as structured CRM fields — the key takeaway being that lead gen should end in clean context and a clear next step, not just a calendar invite.

Building self-hosted marketing data pipelines with AI

A digital analytics consultant argued that raw data from ad and analytics platforms shares an identical structure across clients, so the real business value lies in surfacing trends and decisions rather than in charts, and that off-the-shelf connectors and MCP integrations are costly, opaque, and limited for large historical pulls. The proposed alternative was to use AI to write custom Python that runs autonomously and cheaply in Google Cloud, loading raw data into BigQuery as a central data lake (via Google's free Data Transfer Service) and passing it through a rigorous ETL process with 150+ automated QA checks, organized into four reusable abstraction layers and documented in Markdown so any AI agent can pick up the work without losing context. The closing vision was a shift from static dashboards to autonomous AI agents that monitor curated data, flag anomalies, and answer business questions in natural language.

5th edition

September 10, 2026

Six rules for an expert-driven AI workflow

A member shared a case study on integrating AI into her work for a neuroplasticity start-up, showing with real examples that AI-generated content can "sound great" at first glance yet look unprofessional or flawed to an expert's eye. To manage this, she laid out a personal blueprint of six practical rules for building a reliable, expert-driven AI working environment. The talk closed on an inspiring parallel between human and artificial intelligence — highlighting the brain's natural capacity and arguing that AI is an excellent support tool while human creativity and critical thinking remain our ultimate competitive advantages.

A personal news agent as an information diet

A presenter reframed information overload as a diet problem — once everything is available, the challenge shifts from finding information to selecting it — and built a personal AI agent that delivers a single daily news brief at a set time instead of an endless feed. The core idea is that the prompt is an editorial policy: it encodes what to admit, how to verify it (at least two independent sources, since social media is a signal and not proof), how to rank items by consequence and probability rather than noise, and rules for bursting the filter bubble by requiring uncomfortable, well-argued opposing views. Framed around Postman's media criticism, the "information diet" idea, and Timothy Snyder on freedom, the takeaway was that controlling your own information environment is an ongoing practice — "freedom is a verb" — with the agent removing mechanical work so human judgment, conversation, and action have room.

Building a real app with AI and open data

A presenter demonstrated a public-transport app they built with AI despite not knowing how to write the code, using a city's free, real-time open transit data (published in the widely used GTFS standard, so the same approach works for most cities) to show live arrivals for a personal commute. The app is open source, stores data on the user's own device with an optional free cloud backup, and even hides a small game — a nod to the idea that not everything you build needs a reason. The takeaway was that you no longer need to be a developer to build something useful: you need an idea, curiosity, open data worth exploring, and AI to help you build it.

4th edition

July 7, 2026

Building an AI-powered CRM for community management

A product manager described building a fully functional CRM for a startup community entirely with Claude Code — roughly 57 hours of focused work without personally writing a line of code, at an AI cost of about 25 cents a month (the model estimated the same system would take a senior developer 600–1,100 hours). The design rests on a written "constitution" that always keeps a human in the loop, dual prose-and-compressed rule sets to keep the AI fast and accurate, and a three-layer validation system that auto-records hard facts, scores interpretations by confidence, and always pauses for human approval on personal data.

Local AI tools for cutting paperwork

A presenter showed how local AI tools (Codex and Claude for Work) reduce administrative overhead by keeping context in files within a local folder, with use cases such as extracting and categorizing invoices to save accountants time. Practical tips included always requesting interactive HTML for reports and CSV for data the model will re-read, "refreshing" an overloaded chat by exporting instructions to a Markdown file and starting a clean session, and using a browser extension to pull LinkedIn statistics into a CSV.

AI in PR, marketing, and media monitoring

A presenter shared how a university marketing team uses ChatGPT to cut board-report writing from two weeks to two days, categorizing raw media-monitoring exports into internal themes and comparing huge year-over-year course catalogs to flag changes. Other time-savers included asking the model to output finished HTML for the website, while always keeping final press-release edits in human hands to preserve tone, authenticity, and SEO performance.

3rd edition

June 4, 2026

AI for personalized English-language teaching

A language teacher shared a workflow for producing highly tailored corporate training materials, since global textbooks don't cover niche business terminology. By interviewing clients and feeding their answers plus company website content into Claude or ChatGPT, they generate reading texts, exercises, grammar, and role-plays, using ElevenLabs to create listening audio in different accents — while stressing that the human teacher remains irreplaceable.

AI Search and the future of SEO and brand management

A presenter outlined the shift from traditional SEO to AI Search, such as Google's AI Overviews and answers from ChatGPT. Brands are losing control because the AI assembles its answer with only about 35% of sources coming from the brand's own site and only around 5% of users clicking through; the recommended response is to map the questions customers actually ask, ensure the brand is present on the platforms AI cites (Reddit, Medium, LinkedIn, YouTube), and add an llms.txt file.

Communication analysis with NLP and automation

A presenter demonstrated using AI and NLP models for large-scale sentiment analysis of emails and meeting transcripts — surfacing signals such as overly aggressive selling, customer frustration, or slipping response times — most useful as a periodic management report rather than real-time micromanagement. The talk also warned about prompt-injection risks when uploading corporate data to personal AI accounts and recommended converting PDFs to clean Markdown before feeding them to a model to save tokens.

2nd edition

April 30, 2026

Personal AI assistants and context management

A presenter explained how to work around the way large language models begin to "forget" instructions once the context window fills up, which degrades the quality of their answers. The talk walked through practical ways to persist instructions — custom instructions, Claude Skills, custom GPTs/Gems, and Claude Projects — along with tool recommendations (Gemini for its Workspace integration, Perplexity for search) and a note on the environmental cost of AI.

Data analysis and research with NotebookLM

A presenter demonstrated Google NotebookLM for processing large volumes of source material without hallucinations, since it answers strictly from uploaded sources such as PDFs, YouTube videos, and audio. Use cases included a dedicated research notebook per company for investment analysis, summarizing multi-hour podcasts and multi-day offsites, and turning dry reports into an AI-hosted audio podcast — with the caveat that the tool is no substitute for critical thinking.

Rapid prototyping with Lovable

A presenter showed how AI can bypass corporate bottlenecks by quickly producing working mobile app prototypes. Handed a poorly designed, agency-built app to review, they used ChatGPT to structure the feedback into clear user journeys and design requirements, then rebuilt a modern, functional prototype in Lovable in under an hour — underscoring that results depend on the quality of the prompt and on keeping a human in the loop.

1st edition

April 27, 2026

School management system built with Claude Code

An HR specialist with no IT background used Claude Code (natural-language coding) to build a complete internal school-management system that handles parent–teacher communication, meeting scheduling, and payments. The system was developed over roughly five to six months and includes a module that automatically reshuffles the schedule when meetings are cancelled.

Marketing and sales automation with n8n

A marketing professional built an automated pipeline in n8n (a Zapier alternative) that captures leads, filters and categorizes them, and sends automated emails and alerts to the sales team. A central theme was the change-management effort required to convince colleagues and management of the value of the automation.

Employee experience portal built with Lovable

An HR manager used Lovable to create a 360-degree performance-review system and an internal expense-management module for tracking company-card payments, completing the work in hours rather than days. A full employee self-service portal is planned as the next step.