Communities · Engineering · 2025-2026 · Working prototype

Who is nearby, on your terms

Pods / pods.boston

A local discovery app: see who is within a few streets of you, filter by the interests people choose to show, and reach one of them without either of you becoming a product.

Fig. 1 — Nearby, in the app's demo mode: people grouped by rough distance band, never an exact distance. Every person shown is invented. Own screenshot, 27 October 2026.

Role: Designer, developer
Years: 2025-2026
Project status: Working prototype
Categories: Communities · Engineering
Sources: I08

Nearby grid, pods and invitations: Running at pods.boston
Location and sensitive-category protections: Built · not independently audited
NYC Climate Week matching run (separate script): Run · figures pending data export
Whether people meet and keep using it: Unverified

04Situation

Apps that show you who is nearby are usually built to keep you scrolling and to collect what they learn about you. Location-based apps have also been broken, repeatedly and publicly, by people working out exactly where their users are. With Pods I wanted the opposite: a way to see who is actually within a few streets of you, filtered by the parts of yourself you choose to show, and to reach one of them deliberately, without either of you becoming a product.

05Design

A pod is an interest, not a group. When you join, the app asks a few short questions, suggests pods from your answers, and you confirm what is true. Nothing is applied behind your back.

Some categories are dangerous to be findable by. Being visible as queer, as an organiser or as a person of faith can be a safety risk, so those pods are never inferred, start hidden, and are shown only to an audience you choose.

Nobody is scored or ranked. People nearby appear in rough distance bands, never an exact distance, and are shuffled differently for each viewer, so where someone appears gives nothing away. You join with a one-day invitation from someone already in, and the server keeps no record of who used it. A first message is a short, bounded exchange that then moves to Signal. There is no advertising, no analytics and no notification designed to pull you back.

Pods Invite someone screen on a phone: a QR code, a Copy the link button, and a note that the code lasts a day and does not record who used it — Fig. 2 — Invitation: a one-day code that records nobody who uses it. Local test run, demo account. Own screenshot.

Pods onboarding step 2 of 6: one statement card, I have a side project I'd talk about unprompted, with Not really and That's me buttons — Fig. 3 — Onboarding asks what is true of you rather than asking you to categorise yourself. Own screenshot.

Pods onboarding step 3 of 6, Your pods: interest groups suggested from earlier answers, shown for confirmation, above a separate Sensitive pods section that is never inferred — Fig. 4 — Suggested pods are confirmed by the member; sensitive pods are never inferred and start hidden. Demo account. Own screenshot.

06The open question

Whether people use it to meet, and keep using it, is untested. Pods records nothing about what its members do, by design, so any evidence will have to come from asking them directly.

Separately, for NYC Climate Week I wrote a matching script that grouped attendees into small cohorts and collected their feedback. It is a different experiment from the app, which deliberately does no matching, and its figures will appear here only from a direct data export.

07Status

Running at pods.boston. Nothing in it has yet been reviewed by an independent security researcher, so its privacy protections are stated intentions backed by readable source code, not audited guarantees.

References & documentation

1. I08
`portfolio_master_inventory_2026-10-21_revised.md` — Working inventory, earlier chat reconstruction and unverified legacy projectsfirst-party
Editorial research, no independent verification