Standard Spaces

Standard Spaces is a members-only city guide that combines editorial curation with tools to discover, save, and share spaces. Beyond helping members find somewhere to go, the platform brings people together by enabling members to host their own events and attend those organized by other members.

Miami

The case for curation

Finding somewhere to go shouldn’t be a difficult decision.

Local discovery is fragmented across products that solve different parts of the problem. Google and Apple Maps are comprehensive but unopinionated. Editorial guides like The Infatuation have taste but are static. Instagram and TikTok are great for inspiration, but they aren't designed around curated spaces and events. Recommendations from friends are personalized but get lost across message threads.

The problem is no longer a lack of information.
It is a lack of curation.

I built Standard Spaces around a different premise: what makes a night out memorable is the quality of the place and the people you're with, not the number of options you had to choose from.

In any dense city, people are already looking for somewhere to go and someone to go with. What's missing isn't options, it's a product that connects the two.

From interest to a plan

Most people begin with an intention, not a venue: dinner with friends, drinks nearby, somewhere to co-work, or somewhere worth exploring.

Standard Spaces combines curated inventory, the map, filters, and search into a single discovery system that progressively narrows the city into a useful set of choices.

Once a member opens a space, the product shifts from discovery to evaluation. Photography establishes the atmosphere first, because the question people are actually asking is what a place feels like, not what it serves. Description and tags follow. Practical details answer whether the space fits the occasion.

Reservation links connect to the venue's existing booking service rather than introducing another booking layer inside Standard Spaces. Owning booking would capture real signal — who is going where, and when — but that isn't worth building against the tools venues already use and members already trust.

The objective is to reduce the distance between “what should we do?” and “this is where we're going.”

But narrowing the city still leaves the question of what to surface first.

Taste as a signal

A recommendation becomes more useful when you trust who it comes from.

Standard Spaces keeps the signals members create distinct.

Saving a space means I want to go here.
Checking in means I went here.
Rating it captures what I thought afterward.

Keeping those states separate makes a member’s history more informative than a single collection of bookmarks.

Ratings begin with a simple judgment, from “Loved it” to “Never again.”

Reviews are optional by design. Visiting a space should not create homework. A check-in, rating, or written review can each contribute useful information at different levels of effort.

Profiles bring those signals together through saved spaces, check-ins, favorites, and activity. Following another member adds context to their recommendations by showing the broader pattern of places they choose.

The result is not just a list of opinions, but an evolving picture of someone’s taste.

From places to people

The quality of a space is only part of what makes going out worthwhile.
The people around you matter too.

Standard Spaces extends discovery into live events so members can do more than find somewhere to go. They can attend experiences organized by other members or host their own around spaces already in the guide.

This creates a progression from place discovery to social discovery. A member might first find a restaurant, gallery, or bar because it fits their taste, then discover an event there with people who share similar interests.

Event filter

Curation matters on both sides. Standard Spaces is selective about which spaces enter the guide, while membership creates a more deliberate social environment than an open event directory.

The goal is not to maximize the number of events or attendees. It is to make the experiences — and the people around them — more relevant.

Hosting allows members to create those experiences themselves. Attending gives others a lightweight way to participate without organizing something first.

Most membership clubs run programming through a community organizer, which limits what happens to one person's calendar and taste. Members here host their own. These are not produced events that need a theme and a budget. They are a place and a group of people, without the overhead of spinning up a new thread every time.

Events anchor to spaces already in the guide. Hosting at an arbitrary address is solved — group chats, Luma, and Partiful all do it. What those tools cannot offer is a venue that has already been vetted, and a group of people who have been too.

Standard Spaces therefore moves beyond answering where should I go? toward a broader question: where should I go, and who should I go with?

Building the system

Editorial curation establishes which spaces belong on the platform. Everything else is the system built to support that judgment at scale.

I built Standard Spaces end to end as a solo builder across the brand, product interface, application architecture, and internal editorial tooling.

Brand and UI system

I created the brand and reusable UI system alongside the product itself. Rather than designing complete screens before implementation, I designed primarily in code — building components, interaction patterns, responsive behaviors, and visual rules as the product evolved.

The conventional process maintains two systems: a design file and the code that implements it. They drift, and keeping them aligned is continuous work. Designing in code leaves one artifact instead of two. Agentic coding made the tradeoff clear — changing a component in code is now fast enough that there is little reason to change it somewhere else first.

The same system now extends across discovery, spaces, profiles, ratings, recommendations, planning, and events.

Mobile darkmode

Product architecture

Stack: Next.js, React, TypeScript, and Tailwind on Vercel with Supabase providing Postgres, authentication, storage, row-level security, and Edge Functions. Mapbox powers maps.

The data model is organized around spaces and members, with separate relational tables for the actions connecting them. Wishlists represent intent to visit. Check-ins represent visits. Ratings capture opinion. Follows create the social graph. Events connect a host and attendees back to an existing space through RSVPs and invitations.

Keeping these states separate makes them independently useful while allowing them to compound into a richer model of member taste.

Permissions are enforced at the database layer with row-level security. Public discovery data can be read broadly, member-owned records are scoped by identity, and event visibility distinguishes public events from private invitations. Administrative operations that require elevated access remain server-side.

Search combines deterministic filtering with semantic ranking. A query such as “drinks in Wynwood” is first parsed for hard constraints like category and neighborhood. The remaining candidate set is then ranked using 384-dimensional embeddings stored in Postgres with pgvector and indexed with HNSW.

This avoids a common failure of pure vector search — a semantically similar result cannot override an explicit geographic or categorical constraint.

AI-enabled curation

A member recommendation starts with something simple: a link.

Behind it is a multi-stage ingestion system that turns an unstructured URL into a nearly complete listing.

Perplexity Sonar Pro researches the space across its website and external sources and returns structured fields for the product schema — description, category, neighborhood, hours, pricing, booking information, and other visit details.

Photography is assembled separately from the venue website, Google Places, and other available sources. Low-value assets such as logos, icons, promotional graphics, and poor-quality images are filtered from the candidate pool.

Claude then evaluates the remaining photography across visual quality, ambiance, offerings, brand fit, informational value, and authenticity. Rather than simply selecting the highest-scoring images, the system balances the final set across useful visual roles such as interiors, exteriors, food, drinks, and atmosphere.

Once a listing is approved, its content is converted into an embedding through a server-side Supabase Edge Function using gte-small and stored alongside the space for semantic retrieval.

AI handles the repetitive research, classification, and media-processing work.
Editorial judgment still determines what enters the guide.

Submit recommendation

What comes next

Standard Spaces is live with members and now spans the full loop from discovering a space to planning a visit, recording an experience, contributing recommendations, and meeting other members through events.

The product deliberately trades comprehensiveness for a stronger point of view. That makes the quality of the inventory, membership, and ongoing curation more important than the absolute size of the platform.

The taste graph is built ahead of the membership that will fill it. The signals work as designed — what they need now is density.

The next stage is about validating which behaviors create durable value. Do members return to saved spaces? Do recommendations improve the inventory? Do ratings make discovery more useful? Do events turn shared taste into real-world relationships?

The move into events expands the original thesis. Standard Spaces began by curating where people spend their time. It is increasingly becoming a way to curate who they spend it with.

The larger question is whether those two forms of curation reinforce one another — and whether a better guide to a city can become a better way to experience it.