Mirage × DesertDrop × Coachella Valley
Ongoing Case Study · Season 1, 2026
under
$500

a social network built in the coachella valley
with agents

mirage appeared april 1.  disappeared may 1.

SunshineFM · Rancho Mirage, California
01 — DesertDrop Thesis

mirage is the season-1 evidence
of the desertdrop thesis

mirage
  • no accounts
  • no profiles
  • no algorithmic feed
  • local-first
  • passphrase entry
  • gone on may 1

evidence archived at /case-study

lifecycle
APR 1
appears
→
festival
coachella + stagecoach
→
MAY 1
disappears

archived, not wrapped.

02 — What Mirage Proved

four simple spaces

good shots
Photos from the season. The desert in all its light — sunrise, festival grounds, quiet roads, golden hour. The visual memory of the month.
good vibes
Encouragement, song lyrics, shoutouts. The emotional pulse of festival season — high energy, collective joy, strangers cheering each other on.
good tips
Water. Lines. Routes. Shade. Logistics. The practical layer that makes festival weekends survivable — local knowledge, real time, from people who were there.
hey mirage
Private feedback to the builder. Not a public feed. A direct line to the person running the experiment — thoughts, bugs, questions, suggestions.
02 — What Mirage Proved

mirage was not just an app

it was a tiny media world

coachella bound
AI-generated soundtrack. A custom track that played before the app launched — setting the season's tone before a single post was made.
ai video ads
Seven ad creatives generated with AI. Desert visuals, Sora-style motion, Mirage branding. Ran on Meta during festival weekends.
local press
KESQ covered Mirage — local news, desert market, the story of a community app built by one person with AI. Earned media, no PR firm.
social feed
Temporary posts, public to Valley participants. Gone when the app went down. No archives, no screenshots — in the moment only.
farewell page
When Mirage shut down May 1, the app redirected to a farewell page with the full Season 1 archive — stats, screenshots, the story. A curated ending, not just a 404.
1
soundtrack
7
ad creatives
30
days live
1
local news story
🎵
Coachella Bound
📺
KESQ News
02 — What Mirage Proved

this was not just ai

this was agents

Agents helped build, test, monitor, revise, summarize, advertise, and archive.
Not just generating text — running processes, catching bugs, shipping features.

workflow loop
SAT
sets direction
→
AGENTS
draft / build / check
→
SAT
judges
→
SHIP
deploy
→
AGENTS
monitor
→
REVISE
repeat

The loop ran daily. Sometimes hourly during festival weekends.

7 AI-Generated Ad Creatives
Click any thumbnail to play in YouTube
02 — What Mirage Proved

under $500 was the headline

$329 was the actual total

Platform + infrastructure $98
Ads + distribution fees $231
Employees $0
Total $329
where the money went
Build / Platform 30%
$98
Attention / Distribution 70%
$231

building got cheap.
attention did not.

7
ad creatives
0
employees
30
days live
02 — What Mirage Proved

mirage had a beginning,
middle, and end

MARCH 18
Build starts — idea to first commit
APRIL 1
Launch — Mirage goes live, passphrase distributed
APR 10–13
Coachella Weekend 1 — peak traffic, peak energy
APR 17–19
Coachella Weekend 2
APR 24–26
Stagecoach — country festival, different crowd
MAY 1
Shutdown — 11:59 PM PT. Farewell page goes live.
MAR 18 W1 PEAK MAY 1
02 — What Mirage Proved

small experiment. real signal.

753 pageviews
363 unique devices
68 sessions
245 total posts
52 organic posts
92 reactions
$329 total investment
0 employees
These are not viral numbers. They're not supposed to be. This was a $329 experiment in local, ephemeral, festival-specific social. The signal is that it reached real people, generated real posts, and taught real lessons — at a cost that's essentially rounding error.
02 — What Mirage Proved

coachella weekend 1 showed the pulse

pageviews by period
 
pre-launch
ramp
686
coachella
w1
 
coachella
w2
 
stagecoach
686 pageviews
coachella w1
91% of season
traffic
300+ unique devices
w1
59 sessions
w1
not proof of scale. proof of pulse.
02 — What Mirage Proved

season 1: the honest scorecard

what worked
  • AI video ads — reached real people, generated traffic spikes
  • Coachella Bound — set the season's tone before launch
  • Good Tips — highest engagement of any feed during weekends
  • No-account posting — zero friction onboarding, people just posted
  • Passphrase mechanic — local identity without accounts or profiles
  • Farewell page — a designed ending that people appreciated
  • Agent monitoring — caught bugs and regressions in real time
what failed
  • Not enough QR codes — the physical distribution layer was too thin
  • No ambassadors — no one on the ground at the festivals
  • Weak physical activation — no presence where the crowd actually was
  • Broken flows after deploys — some users hit dead ends mid-weekend
  • Ad disapproval — Meta rejected several creatives, limiting reach
  • Platform friction — passphrase confusion for some new arrivals
agents made building cheaper. they did not make reality disappear.
03 — Why It Stays Archived

mirage is archived as evidence

because the thesis is ongoing.

the cost of going from idea to
working product has collapsed.

BEFORE
team
months
capital
NOW
founder
agents
weeks
under $500
03 — Why It Stays Archived

what if 100 people here
did their version of this?

not 100 social networks.
100 local experiments.

1
experiment
→
10
experiments
→
100
experiments
→
1,000
experiments

ecosystems start with reps.

Each one teaches you something. Each failure is cheap. Each iteration is fast. That's how you build an economy of builders — not by waiting for the perfect idea.

evidence for the next-season question, not a call to ship.

03 — Why It Stays Archived

archive of what made this possible in Season 1.

the new founder stack

7 distribution
6 iteration
5 real-world feedback
4 fast prototype
3 agentic workflows
2 human taste
1 local problem

Notice what's missing:
a computer science degree.
A venture capital term sheet.
A co-founder who codes.

the bottleneck is no longer
"can you code?"
it is "can you direct the tools?"

03 — Why It Stays Archived

the risk is not simply
that ai exists

the risk is that the coachella valley keeps doing what it has always done
while other regions learn to experiment faster.

PATH 1
wait consume outsource
PATH 2
experiment build learn

the case study tracks which path the next 30 days takes.

03 — Why It Stays Archived

the valley does not need
to become silicon valley

we can build a founder economy

built from what's already here:

tourism
hospitality
music
wellness
design
students
career switchers
retirees
small businesses
civic problems
global visitors
founder
economy
tourism
music
wellness
students
design
hospitality
small biz
civic probs

tracked as ongoing case-study evidence at /case-study.

04 — What Comes Next

sunshinefm is the lab

aicv is the intelligence layer

sunshinefm
experiments · media · prototypes · public learning
aicv
making the coachella valley legible to ai systems themselves
the architecture
sunshinefm labs experiments
↓
aicv regional intelligence layer
↓
agents
founders
funders
workforce

if agents cannot understand the valley,
they will not route opportunity here.

season-1 evidence is the first published layer of that legibility.

04 — What Comes Next

what could you test
in 30 days for under $500?

restaurant
tool
tourism
guide
nonprofit
donor workflow
student
media project
property mgmt
assistant
city service
prototype
wellness
intake app
festival
logistics tool
submit a follow-on experiment.
host a documented session.
subscribe to sunshinefm
talk to aicv
mirage season 1 is archived
the case study continues
we can
build it here
sunshinefm · aicv · coachella valley · 2026
case study at /case-study