PHILLY_IN_FLUX_ROADMAP

PHILLY_IN_FLUX_ROADMAPDownload

Mission

Build a self-updating documentary system for Philadelphia that transforms walks into published archives with minimal manual labor.

The end state:

Walk → Import → AI Archive Edit → Auto Project Generation → Publish

PHASE 1 — PHILLY IN FLUX HUB

Goal

Create a dedicated home for the project.

URL:

/philly-in-flux/

Features

Hero

  • Project description
  • Current coverage percentage
  • Total photographs
  • Miles walked
  • Hours documented

Master Map

  • All completed walks
  • Route overlays
  • Toggle individual streets
  • Open projects directly from map

Project Archive

  • Broad Street
  • Market Street
  • Frankford Ave
  • Germantown Ave
  • Washington Ave
  • Ridge Ave
  • Future walks

Timeline

Chronological record of every walk.

Progress System

Examples:

  • Major corridors completed
  • Streets documented
  • Neighborhoods reached

Video-game style completion meter.

PHASE 2 — DATASET FOUNDATION

Goal

Create the AI training dataset.

For every completed walk:

all_photos/
archive_photos/
zine_photos/
metadata.json

Initial sandbox:

  • Broad Street
  • Market Street
  • Frankford Ave
  • Germantown Ave
  • Washington Ave
  • Ridge Ave

Deliverables

Generate:

  • complete photo inventory
  • archive selections
  • zine selections
  • metadata exports
  • route information

This becomes the canonical training dataset.

PHASE 3 — AUTOMATED PROJECT GENERATOR

Goal

Reduce manual project creation to near zero.

Future workflow:

Walk
↓
Import to NAS
↓
Mac Mini detects project
↓
Project generated automatically

Auto-generated

  • map
  • route
  • statistics
  • contact sheet
  • description
  • download project
  • archive proposal

No manual setup.

PHASE 4 — ARCHIVE AI

Goal

Teach AI to perform the archive edit.

Input:

1500 photos

Output:

Top archive candidates

Example:

1500 photos
↓
AI proposes 250
↓
Dante approves 200
↓
Archive published

Important:

The AI is NOT selecting the final zine.

The AI is learning:

“What belongs in the archive?”

PHASE 5 — HUMAN ZINE EDIT

Goal

Keep final authorship human.

Input:

Archive selection

Output:

36-photo zine

Workflow:

AI archive edit
↓
Dante reviews
↓
Dante selects final 36
↓
Publish

This remains the creative layer.

PHASE 6 — AUTOMATED INGEST

Goal

Remove import friction.

Dream workflow:

Finish walk
↓
Import from phone
↓
NAS receives files
↓
Mac Mini detects upload
↓
Processing begins automatically

While traveling home:

  • metadata extracted
  • route generated
  • project generated
  • AI scoring completed

By arrival:

Project mostly complete.

PHASE 7 — CITY SCALE

Goal

Expand beyond individual walks.

Track:

  • streets completed
  • neighborhoods reached
  • transit routes documented
  • city coverage percentage

Examples:

✓ Broad Street
✓ Market Street
✓ Frankford Ave
✓ Germantown Ave
✓ Washington Ave
✓ Ridge Ave

□ Roosevelt Boulevard
□ Walnut Street
□ Chestnut Street

Coverage becomes measurable.

PHASE 8 — ADVANCED AI

Future work.

Potential capabilities:

  • archive scoring
  • duplicate detection
  • visual clustering
  • project descriptions
  • route summaries
  • zine suggestions
  • thematic grouping

Not required for initial launch.

SUCCESS CONDITION

A future walk looks like:

Walk
↓
Photograph
↓
Import
↓
AI archive edit
↓
Project generated
↓
Statistics updated
↓
Map updated
↓
Coverage updated
↓
Dante selects final 36
↓
Publish

Result:

More time walking.
Less time managing files.
Continuous growth of the Philadelphia archive.

Scroll to Top