Mapkin is the personal travel concierge.

A map app that steers travelers to plans and places they’d never discover otherwise.

Try it now on mapkin.io

The wealthy already have personal travel concierges — humans who filter places and plan to fulfill their client’s tastes. Mapkin unlocks that level of personalization and curation for everyone with deep search capabilities and LLM-powered curation.

Why Now?

• Recent ML advances make a personalized travel concierge viable

  • LLMs + agentic search have passed the quality threshold to understand vague intent, reason over places and neighborhoods, and make plans that feel human.
  • Recommender systems (TikTok, Instagram ads) have proven that large-scale taste modeling works — the ceiling for personalized discovery is very high.
  • Speech models are entering the same exponential curve, enabling natural voice-first interfaces, perfect for personalized virtual tour guides.

  Mapkin sits at the intersection of all three: language reasoning, taste inference, and voice UX. This was not feasible 1-2 years ago.

• Travelers now want local, authentic, TikTok-discovered experiences — not top-10 SEO’d lists.

  • Post-COVID / the ChatGPT / TikTok era have accelerated a shift toward local, authentic, small-scale experiences. More about this in New York Magazine, for example.
  • Meanwhile, SEO and ad-driven platforms (Google Maps, Tripadvisor, Reddit) have saturated the internet with recommendations that undermine user trust, leaving space open for driving growth by building trust.

  Mapkin is specifically building towards a great single-player tool that drives high trust while being high variance.

Progress

• I have shipped a v0 in under 50 days, including best-in-class social media indexers, LLM-driven search & a voice “tour guide” agent

  • Mapkin search indexes over sources like TikTok (including on-screen text in videos), Instagram, Google Maps and other sources, making it unmatched relative to apps like ChatGPT, Gemini etc.
  • Mapkin LLM-driven search & memory systems enable simple, conversational search and route planning, “vibe-search”, if you will.
  • Mapkin Voice enables a “tour-guide” experience for travelers: just put in airpods and walk around — Mapkin knows where you are, what’s interesting nearby and what you’d enjoy.

• 350+ users onboarded, over 50 DAU

  As of November 17. Our only goal thus far has been to iterate quickly towards discovering signs of product-fit and growth-strategy fit. We ran an experimental TikTok UGC program, which onboarded the vast majority of our users (nomads and frequent travellers).

• Early users love the app

  • “I can see how people enter LLM psychosis… I have been on a call with my mapkin for over 2 hours now”
  • “Honestly mapkin bangs, i think im almost a dau”
  • “AMAZING search… I’d pay for access to that alone. Mapkin found the best places I’ve found after researching over an hour instantly!”

Growth

• Come for the tool, stay for the network

  Mapkin begins as a high-trust, single-player utility: a better way to search for places and plan trips. Once users rely on Mapkin for taste-based discovery, we can start bootstrapping a powerful network effect, including:

  • proprietary signals from where people actually go,
  • user-generated lists, reviews, and saved spots,
  • social interactions around shared plans and lists

  The wedge is utility. The defensibility comes from aggregated taste and traveler behavior.

• UGC-Driven Growth

  • The closest existing channel with demonstrated PMF for place recommendations is TikTok/IG Reels: creators feature cafes, restaurants, bars, hotels, and experiences and users collect these videos into their “visit some day” list. We plan to use this mechanism as our main driver of growth.
  • Weaponizing our search capabilities for UGC-driven growth**:** By building the best-in-class search over TikTok and IG Reels, we gain the ability to discover the best creators before they blow up, enabling us to partner with them early and funnel their followers into Mapkin.

Competitors

• The Bootstrapping Problem has killed every “Place Recommendation” startup before

  Previous attempts (ex. Foursquare, Loopt, Corner) failed because they tried to build network-first instead of utility-first, struggling with:

  • Cold-starting cities: new users open the app and see nothing
  • Misaligned user incentives: early users don’t meaningfully benefit from contributing

  Mapkin inverts the model: we start with a best-in-class single-player experience, using ML to solve the taste problem before relying on network effects.

• Incumbents have a data moat but lack focus

  • The form factor for Google Maps has not meaningfully changed in ~20 years, largely because their usage is driven by day-to-day utility (navigation, searching franchises etc.). See tweet for more discussion.
  • ChatGPT is not currently optimizing for hyper-local search tools (ex. doesn’t know opening hours for most places near you) and doesn’t have access to your live location.

Team

• Nalin Bhardwaj. Researcher, engineer, ex-IOI

  Previously CTO and co-founder of Daimo, leading stablecoin payments startup backed by Vitalik Buterin and others, currently doing millions of dollars in payment volume every month.

  Before that, ran machine learning at Lunchclub (with Cognition / Devin founder Scott Wu), and led research at 0xPARC Foundation advancing ZK proving systems cryptography.

  More on Twitter and LinkedIn.

Monetization