Welcome back to the NYC AI+ newsletter, where I share upcoming AI events in NYC, practical AI insights and tools, news, and research papers to connect you to the AI ecosystem in NYC and help you succe...
![🚀 NYC AI+ Newsletter: AI Events [3/9 - 3/15~] + AI Search & GEO + Community Updates 🚀](/_next/image?url=https%3A%2F%2Fduxbvowwaeiqsxajxrku.supabase.co%2Fstorage%2Fv1%2Fobject%2Fpublic%2Fcontent-images%2Ffeature-1773112435479.png&w=3840&q=75)
Welcome back to the NYC AI+ newsletter, where I share upcoming AI events in NYC, practical AI insights and tools, news, and research papers to connect you to the AI ecosystem in NYC and help you succeed in and with AI. 😀
Upcoming AI Events in Town
A deep dive into AI Search Tools and how they change the game of the market
GEO AI and what SOM is all about
Miscellaneous & Fun AI
Let's get started.
Mon, 9th March - NY Tech Meetup: Space Tech Edition You love space and defense themes? Robots? And you are a founder, engineer, researcher and investor? Then join this meeting and mix and mingle with the space and Earth crowd.
Mon, 9th March - Agent Con New York 2026 This is a full-day event and if you are a developer and want to build the future of intelligence and AI, then this event is for you.
Tue, 10th March - Future of DevEx: NYC This event is about how AI reshapes developer workflows, for founders, developers and toolmakers, with Posthog, Grafana and more
Tue, 10th March - Harvard XR NYC Chapter Kickoff Meetup Do you want to dive into mixed, virtual and augmented realities? Then this meetup is for you. There will be showcases, mix and mingling.
Tue, 10th March - Creators x Founders Lighthouse Dinner, hosted by Notion Are you a builder, creator, founder, love Notion and their AI workspace and want to connect, share and network? Then join this event to mix and mingle.
Tue, 10th March - Google SRE NYC Tech Talk Are you a site reliability engineer or are excited what one does? Then hop over to Google for 3 deep-dive talks on on-call burnout, AI agent assistant incident troubleshooting and reliability magic for AI platforms
Wed, 11th March - Startups Decoded - Live - This is a super interesting conversation about startups and the question - is it always good to speed up decision making and processes with AI and where is it risky and hurts the business? This is a morning event.
Wed, 11th March - MCPConference New York This is a full-day event (paid, let me know if you need a free ticket), and covers all things MCP with speakers from eBay, Snowflake, Grafana Labs and others.
Thu, 12th March Arize Builders Meetup - NYC - Boosting Claude Code performance with prompt learning You want to speed up your Claude programming? Then come to this meetup to get hands-on insights and prompts to do so.
Thu, 12th March - UGLY TALK x The Builder Series: B2C STARTUP FUNDRAISING AND SCALING Do you want to learn from community leaders of how they scale(d) their communities and what works in 2026 - with a bit of AI and a lot of networking? Then join that event here.
Tue, 17th March - NYC AI Users - AI Talks, Demo & Social: AI Investing & Coding Agents - This meetup includes two talks - one looking at what comes next after coding agent and the other takes a closer look at investing with AI. Mix and mingle and learn more about AI at this meetup.
Wed, 18th March - NYC Openclaw Meetup - Wanna share best practices, workflows and projects in AI, learn more how to optimize your experience with Openclaw or simply mix and mingle with engineers and AI-enthusiasts? Then this is your event. (in Queens)
Wed, 18th March - Trivia Night - Brains, Bytes & Bragging Rights! - Databricks, Sigma Computing and a night of AI and data innovations. If that sounds like fun, then this is your night.
Thu, 19th March - AI Builders Day NYC: Intercom x Antler You love building and as much as possible? Then this is for you. It's a full day of building, breakfast, talking to VC and maybe even finding your first investor that day ;)
Thu, 19th March - AI 2030 ONLINE-The Human Blueprint: Reclaiming Digital Dignity in the Age of Agentic AI This is a one-hour online lunch session all about building loyal, ethical AI agents and the benefits of them with the Head of AI Engineering at Oracle
Thu, 19th March - eNO badge: World's first mini AI bodyguard comes to NYC Do you care for protection as a women via AI? Then come to this event to learn from the founder of an AI bodyguard for women how she built the company and what she learned.
Fri, 20th March - Goodfin Futures: Spotting Unicorns with Ash Arora This is a one-hour online afternoon session on how to spot unicorns in the AI Enterprise space from learnings and investments at Gamma, Polymarket and Mistral AI
Fri, 20th March Uncorked: AI, Tech & Culture Join for a culture, creative, AI and tech gathering with wine, startup presentations and mix and mingling. this is a paid event.
Tue, 24th March Build & Deploy Voice Agent Workshop + Hackathon (Vapi AI x Cartesia) Wanna learn about Voice AI agents in short, practical workshops? then this is for you, an evening about learning, sharing and networking for CTOs, builders and technical CEOs
Tue, 24th March The NYC FDE & FDE-Curious Gala You care about Forward-Deployed Engineers and what they do? FDE leaders of Anthropic, Microsoft, ElevenLabs and more will be joining too for mixing, mingling and pizza.
Tue, 24th March Beyond the Prompt: Enterprise RAG If you want to build an Enterprise AI bot, then this workshop is for you, hands-on, with a computer needed, this meetup wil teach you new skills for sure.
Wed, 25th March ComfyUI Official NYC March Forum Wanna learn about real-time video AI and listen to speakers that talk about that? Then join this meetup
Thu, 26th March - No. 50-International Conference on Collaborative Intelligence (ICCI)-Science & StartupAccessible AI Research and Innovation -Virtual - Do you wonder how AI and humans can best collaborate and work with each other instead against each other? Then this online lunch session is for you
Thu, 26th March - Foundation Models for Structured Data: Prior Labs US Launch - Listen to a talk about "The Next Frontier in AI: Tabular Foundation Models Meet Enterprise Data" and celebrate with Prior Labs their launch in the US.
Thu, 26th March - raid 8: talks on research, ai, design - Come to Modal's HQ for a meetup on research, ai, and design, talks, and mix and mingling.
Sun, 29th March - TinyFish Accelerator: Build Sprint How again do I build an AI agent? This is a Hackathon/Build Sprint for you to build an AI Agent and have the ability to win prices with it - 9-week program and good thing - they have mentors ;)
Mon, 30th March - OneSixOne - Scaling AI Infrastructure This afternoon event is part of New York Tech Week with networking and a panel discussion on software and hardware AI infrastructure innovations
Mon, 30th March - DTNY Robotics Reception Connect with founders, investors and operators in the robotics ecosystem in New York and mix and mingle with like-minded people.
Thu, 7th April - MCP in the Wild: The Future of Data Agents in Production (NYC) You want to have Snowflake, Crew AI and Block share light on some of the hottest topics of AI right now? Then join this meetup and dive into all these topics, listen to conversations and mix and mingle
Mon, 6th April Developer Job Fair Are you a developer and are looking for a new job or simply to figure out what is out there in jobs or to network? Then this is for you. This is an afternoon event.
Not long ago, we would evaluate LLMs by their cutoff date - the date until which they were trained on latest. But soon after, LLMs could search the web and get increasingly good with it. So good so, that nowadays, we let AI agents handle most of our work searching the web and doing work for us crawling websites, researching content, consolidating insights and creating lists of data.
But have you asked yourself what powers those web searches and what happens under the hood of search when you asked Claude, ChatGPT or Gemini a question and they seamlessly search the web for you?
The search layer, the ability of any LLM to become a search-augmented AI lies in its power to not only share the knowledge it has stored and is trained on, but to retrieve the latest news for you, knows that your company just published a new internal memo and that you published a research paper yesterday.
And the AI achieves that search by connecting to a search API - an interface that directly interacts with a search engine. An LLM sends the query/your question to to that search API, which delivers the query to a search engine, processes the results and shares the final results with the LLM. And that process is called Retrieval-Augmented Generation (RAG).
RAG works in 4 steps: Ingestion, Retrieval, Augmentation, and Generation
Step 1 — Ingestion: All of that data in the web (it also works for your internal documents, emails etc) is broken into chunks of data, converted into numerical representations called embeddings/vectors and stored in a vector database.
Step 2 - Retrieval: When you now ask a question, your query is also converted into a vector in the same vector space. The system then finds the chunks with the most mathematically similar vectors. This is then called to be the most semantically relevant results.
Step 3 — Augmentation: Those retrieved chunks are added to your original question to form an enriched prompt: "Here's what I found. Now answer the question."
Step 4 — Generation: The LLM then takes the enriched prompt and generates a response grounded in real, retrieved information rather than memory alone.
And while RAG is the base layer for all AI search engines, they have become more sophisticated over time. Modern RAG systems now do iterative retrieval (multiple rounds of search for complex queries), hybrid search (combining semantic search with lexical search), they rewrite queries to improve recall and re-rank search results for more relevance.
The search engine market for AI agents is developing fast and these tools handle the heavy lifting of web crawling, indexing and returning clean structured results the LLMs can then use for their answers.
AI-Native Search APIs (Built for LLMs)
🟣 Exa.ai — "Google for AIs"
Formerly Metaphor. Instead of keyword matching, Exa uses embeddings to understand query intent and context. It actively filters spam and SEO-gamed pages, returning only high-quality sources. Supports semantic search, similarity search ("find me more like this"), domain filtering, and structured outputs with citations. Scored 94.9% accuracy on the SimpleQA benchmark — the highest tested among major search APIs.
Best for: Deep research, complex multi-hop queries, agent workflows needing nuanced results. Raised $85M at a $700M valuation.
🔵 Tavily — The developer's workhorse
Positions itself as "the web access layer for AI agents." Fast, simple, and highly reliable — trusted by over 800,000 developers. Tavily doesn't try to be the most semantically rich; it tries to be the most consistently useful. Handles search, content extraction, site mapping, and crawling in one clean package. Achieved 14-second average turnaround for correct web search and extraction in the MCP Benchmark (July 2025).
Best for: Production AI agents, RAG pipelines, LangChain integrations, teams that need quick setup without fine-tuning.
🟠 Perplexity Sonar API — Pre-synthesized answers, not just raw results
Unlike Exa and Tavily which return raw results for your LLM to process, Perplexity's Sonar models combine search with their own LLM to deliver synthesized, cited answers directly. Fastest major search API at ~358ms median latency. Available in Sonar (lightweight, quick facts) and Sonar Pro (complex multi-step research). Perplexity also now offers a separate raw Search API for developers who want just the results.
Best for: Applications where latency matters most, current events, factual Q&A where pre-processed answers save development time.
🟤 You.com Search API — Enterprise-grade, multi-step reasoning
Access to over 10 billion web pages with news content available seconds after publication. Integrates natively with AWS, Databricks, and OpenAI models. Known for handling complex multi-step research queries better than most competitors.
Best for: Enterprise applications with high accuracy requirements, complex research workflows, multi-step reasoning chains.
Independent Index APIs (Big-Tech-Free)
🦁 Brave Search API — Privacy-first, truly independent
The only major search API built on a fully independent web index — not powered by Google or Bing data. Brave crawls and indexes the web itself (30 billion pages, 100 million daily updates), making it the choice for developers who want zero surveillance-based business models. An AI grounding feature reduces hallucinations by prioritizing authoritative sources. Popular for healthcare, financial, and government applications where query confidentiality is critical. Priced at $5/1,000 queries.
Best for: Privacy-sensitive applications, teams building away from Big Tech dependency, anyone wary of Google's antitrust situation.
📦 SERP & Data APIs (Google-Powered)
If you need volume and don't need semantic AI-native magic, traditional SERP APIs wrap Google or Bing results in developer-friendly JSON. Key players: Serper (affordable, great for LangChain integrations), SerpAPI (broad search engine support including Bing/Yahoo/DuckDuckGo), Bright Data (enterprise-grade SERP extraction with GDPR compliance), and ScrapingDog. Note: Microsoft shut down Bing Search APIs in August 2025, pushing users toward Azure AI Agents — a good reminder to diversify away from platforms that can pull the rug.
🕷️ Web Crawling & Extraction APIs
Not strictly search, but increasingly part of the AI retrieval stack: Firecrawl offers a flexible pipeline combining search with full content extraction — you can get standard metadata or scrape clean structured content from the found pages in a single API call. Incredibly useful when you need not just the URL but the actual article content.
You see, there are some very powerful tools on the market and from experience I am so far especially impressed by Exa, but tried Tavily at a hackathon recently too and if you build with search or simply want to see the power of these systems - even without being baked into LLMs, they are powerful by themselves - Exa especially for people and company searches.
And if you wonder what the best one is, there is one thing worth flagging: accuracy varies dramatically by query type. Exa dominates semantic search queries. Tavily excels at factual verification. Perplexity wins for current events. And You.com handles multi-step reasoning best. There is no single winner - the right choice depends on your use case.
And what has that now all to do with search engine optimization? And the way we not only search for information in the web, but also retrieve information?
SEO or in long - Search Engine Optimization focuses on optimizing one's website, content and channels to be found by humans clicking through Google and looking for information on the web.
When I have a well-performing SEO strategy, my website and content appears on top of the Google searches and there is a higher likelihood that when for example I offer a CRM for small businesses and someone types in "I search for a CRM for small businesses" that my company and product show up for that person in one of the first results in Google.
But what happens now, when we do not search via Google anymore to get our information? When we use LLMS like Claude, ChatGPT and Gemini to search for our information and take the information that these LLMs give us to make our decision?
How does it change the way those of us that want to make our content, our companies, our insights been seen and ranked high in comparison to others, what do we need to be different to be not only found via Google but via LLMs?
GEO stands for Generative Engine Optimization, AEO stands for AI Engine Optimization and SOM for Share of Model. All three determine a new emerging field, similar to SEO when it emerged in 2002, GEO is now the new field emerging - experimental and yet powerful and emerging rapidly.
Today, only 16% of brands track their SOM metric - the % of one's brand showing up in an LLM's synthesized answer when asking a question and the LLM's answer includes one own's brand for that query, a la the one of "What is the best CRM for small businesses?".
But what is the difference between SEO and GEO? How do these new AI-Native Search engines search?
We grasped a first insight in the section looking at Search Engines - understanding how they are working and how they utilize RAG and additional new methods helps us to close the gap of how those systems find answers for us and the way of how we need to optimize our content to be showing up for the queries and questions we care about to be asked about us.
Traditional SEO
In traditional SEO, your focus was on optimizing for rankings and clicks. When searching in Google, you wanted to make sure to appear in the first 10 search results that Google would provide and success meant that someone clicked on your page. Keyword density was important and backlinks. And your performance was measured in impressions and CTRs. Content was written for humans to browse.
Future-forward GEO
With GEO, you are optimizing for citations and mentions. It is not about the top 10 ranks, but you want your brand to be included in that one synthesized answer the AI/LLM is giving you when searching for those query results to show you in its answer. Success now is determined about how AI speaks about your brand in its response. And here, statistics, quotes and structured clarity as well as searchable websites are becoming more important. The metric changes towards SOM, the Share of Model. And while we are still searching the web ourselves from time to time, the growth of AI searching the web for us requires us to create content that is there for AI to extract and cite.
The difference is fundamental. Traditional SEO optimizes at the page level, while GEO optimizes at the entity and authority level - are you the kind of source that AI systems trust enough to cite when synthesizing?
GEO is still in its early beginnings. Thus, you do yourself a favor of experimenting and researching for how your brand shows up in AI search results.
I attended a hackathon over the weekend building a GEO Audit Tool that would focus on optimizing the SOM for your brand. And the first thing we did was to audit a brand's current status of SOM across various Search AI APIs like Exa and Tavily plus the main LLMs - Claude, Gemini, ChatGPM, Perplexity - across more than one AI model version.
Once you have a great understanding of your strengths, weaknesses and those queries that move the needle for you and those that don't, here are some general tips to improve your SOM:
Generative engines prefer content that is dense in meaning, ease to parse, and well-sourced. Q&A formatted content for example improved chatbot visibility by 55% in independent tests, because you're essentially pre-formatting your answer for the AI.
Bullet points, clear headers, schema markup, and explicit source citations help too. Plus, it's not only about one website page, but about authoritative mentions in reputable publications, podcasts and expert roundups carry disproportionate weight - making GEO look less like SEO and more like a fusion of SEO, PR and thought leadership.
If you get one insight from diving deeper into search and GEO, then remember that both are two sides of the same coin.
Search is about how AI and LLMs are searching the web and how they fund the information we more care about, with our context and our search history and to make these search results as valuable as possible for us.
GEO is our ability and strategy to make sure we show up in these search results in a way we want to show up - with our brand, our value to customers and our offerings. And to be visible in the ever-extending complex web of search, information and attention.
I tried Exa over the last weeks and I am a big fan - when I tried to use LLMs to enrich and find information about people, I stumbled. And I did not wanted to spend too much money on Clay, Apollo and else - that I found Exa to be very strong in that domain - love your input and to exchange ideas on this one
If you haven't tried Cursor, I highly recommend to do so - not only for coding, but specifically for all kinds of work - Cursor paired with Opus 4.6 Claude allows you to not only do ad-hoc analysis, but saving your key insights, client work, everything and then directly runs analyses on it, does team planning with you, runs simulations on i.e. sales commissions, and builds you small tools for your operations - an absolute life saver - and you do not need to be able to code for that - just accept to look at code from time to time ;)
SaaS companies are fearful - talked to several ones over the last weeks - that startups prefer to build their own internal operations and tools and won't invest anymore. Curious if you see that trend as well?
Listened to a speech of a Partnership Lead at Google on AI Transformation - and realized once again - building AI tools is 10% of the process. The other 30% are to build tools that are useful and improve workflows and the rest is about change management. Tools can be as powerful as they may be - and AI may change change slightly as it may fully replace a human task, but still, change management is key to make sure everyone sees its value and the company and the mindset of its people leads to sustainable long-term change - a part that is often forgotten - in tech, in business, and yes, also in AI
I had been absent from the NYC AI scene a bit and are now back in the game, going to more events again and are excited to meet you at one or the other of the upcoming events. Soon sharing more updates here again about the community, faces and leaders that move the needle and inspiring voices in the space. Stay tuned.
Love the Newsletter? Have ideas for improvements? I love to hear from you. This newsletter is here to bring value to you and connect you to the AI ecosystem in NYC. If that works well - amazing - and if that can be improved - let me know too.
Thanks for being part of this community. And great to be back!
Best, Frederike