← Default view
SPEC SHEET · 2026

Vibe-coding stack for 20 attendees · 3 hours · shared wifi

SKU WS-VC-2026-Q2
REV 2026-05-23
SRC 78 citations
Host spend · 20 attendees
€0 – €46/ $0 – $50
Per-attendee marginal cost: €0. Solo vibe-coders burn €92–€230/mo unnoticed [27]; workshop scale is well below a single user's burn.
Network plan
A → B
Plan A: venue wifi. Plan B: instructor phone hotspot. 100+ laptops saturate APs not provisioned for power users [39].
Signup window
T-48h/ off-venue
Single highest-yield mitigation [52]. Shared NAT triggers Turnstile [50] and "too many sessions from 1 IP" [69].

§ 01 · Primary & Backup units

2 · keep both warm
▣ PICK Lovable

Lovable

CHAT-ONLY UI · SSR · *.LOVABLE.APP

Polished output on first prompt, hides the file tree, ships TanStack SSR so heavy work happens server-side and only HTML chunks stream back — the thin-client profile that tolerates low bandwidth best [38]. The only vendor with a real K-12 / no-account classroom track [7].

FREE
5 cr/day · 30/mo[11]
PRO
€23 /mo · 100 cr[12]
SIGNUP
Google SSO
PUBLISH
1 click · *.lovable.app[55]
UPTIME RISK
314+ outages since 5/2025[71]
VS NOVICE
★ Workshop default[78]
▣ BACKUP Bolt.new

Bolt.new

WEBCONTAINER · NO ACCOUNT · *.BOLT.HOST

Lowest signup friction in the field — open a tab, type a prompt, see a result [4]. Landing page in 2–4 min, to-do app in 8–12 [5]. But the WebContainer hates locked-down laptops: ad-blockers, VPNs and non-Chromium browsers break the cross-origin-isolation handshake [31]. Screen-test laptops first.

FREE
1M tok/mo · 300K/day[15]
PRO
€23 /mo[16]
SIGNUP
None (free tier)
PUBLISH
1 click · *.bolt.host[58]
BOOT TIMEOUT
30 s hard cap[32] ⭐ 16k
VS NOVICE
Failed photo-booth 6/6[1]

§ 02 · The rest of the field

4 · context only
⚠ POSSIBLE v0

v0

UI-FIRST · VERCEL MENTAL MODEL
Free7 msg/day + €4.60 cr[13]
Pro€18.50 /mo
SignupGoogle/GH/Apple[46]
BandwidthSSR · thin
Publish1 click · Vercel[57]
Daily burn7 msgs = 1 bug fix[14]
Shipped working photo-booth from one prompt alongside Lovable; weaker full-stack [1].
✕ SKIP Replit

Replit Agent

FULL IDE · CLOUD-HOSTED
Free10 ckpts/mo[17]
Core€23 /mo · €23 cr[18]
SignupStandard
BandwidthHeavy · continuous sync[37]
PublishSleeps @ 5 min[17]
Per-checkpt~€0.23 · billed on err[19]
IDE overwhelms non-coders; free tier near-unusable [3]. Students: €9.25/mo first 6 mo [26].
✕ SKIP Cursor

Cursor

FULL IDE · LATER TOOL
Free2k compl · 50 prem[20]
Pro€18.50 /mo
Student3-7d SheerID[49]
BandwidthHTTP/2 · drops[34]
NetworkCorp proxy issues[33]
vs novice"start somewhere else"[6]
Useful later — wrong first tool. SheerID throws "limit exceeded" on bulk .edu signups with 72h cooldown [74].
⚠ FALLBACK Claude Artifacts

Claude Artifacts

IN-CHAT · ~70% TO SHIPPABLE
Free~15-40 msg/5h[22]
Pro€18.50 /mo · 5×[22]
Age gate18+ enforced[41]
SMSRequired (email)[75]
BypassGoogle SSO[42]
ShareLogin wall[61]
Runs apps inline; ChatGPT Canvas only shows code [8]. VoIP & Google Voice rejected [43].

§ 03 · WiFi survivability

first-load weight · drop tolerance
Lovable / v0
SSR HTML chunks[38] · each request independent
Claude Code (CLI)
5.5× lighter than Cursor[36] · silent hang on drop ⭐ 126k
Bolt.new
<1 MB WASM runtime[29] + npm · runs offline once booted[30]
GitHub Codespaces
Server-side VM[40] · session preserved across drops
Cursor
HTTP/2 streaming[33] · chronic disconnects, IDE restart[34]
Replit
Full cloud IDE · continuous sync · loses access on drop[37]

§ 04 · Signup gates · the silent workshop killer

most attendee minutes lost here, not at the prompt
Tool Account Phone SMS Age gate Credit card Walk-in?
Bolt.new NO NO none[5] NO ✓ Walk-in friendly
Lovable YES NO (Google SSO) none visible NO ✓ Walk-in friendly
v0 / Vercel YES NO none at signup NO[47] ✓ 5 SSO providers[46]
Cursor YES NO none at signup NO Hobby[48] ⚠ Student Pro = 3-7d review[49]
Claude.ai YES YES (email)[75] 18+[41] NO ⚠ Google SSO skips SMS[42]; VoIP rejected[43]
ChatGPT YES YES 13+ · parental 13-17[44] NO ⚠ IP + phone-country dual-blocked in ~20 countries[45]

§ 05 · Cost build · 20 attendees × 3h

three host-spend tiers · EUR (USD shown for ref)
Tier A · Free path

Both free tiers

€0/ $0
Lovable 5 cr/day = 1 app, no iteration[11]
Bolt 1M tok/mo, 300K/day cap[15]
Per-attendee €0
Tier B · Comfort ★ recommended

1 Lovable Pro

€23/ $25
Host Pro seat · 100 cr/mo[12]
Attendees stay on free, watch demo
Per-attendee marginal €0
Tier C · Insurance

+ static fallback

€46/ $50
Lovable Pro €23
Cloudflare Pages €0 · unlimited bw[63]
Buffer €23 for credit top-up

§ 06 · Known failure modes · field-tested mitigations

keep two tools warm and stay flexible[73]
Venue wifi saturates / drops[39]
Instructor phone hotspot · cellular plan B[76]
Shared NAT → "too many sessions from 1 IP"[69]
Pre-warm accounts off-venue[52]; split between two tools
Lovable outage (314+ since 5/2025, up to 10h+)[71]
Bolt.new warm as backup; pre-built starter repo
Claude SMS rejects VoIP / no-phone attendees[43]
Continue-with-Google bypass[42]; pre-provisioned host accounts
Free-tier credit exhaustion mid-build[2]
Constrain scope to ≤1-2 prompts on Lovable; switch to Bolt for iteration
Non-coder hits paywall on first prompt (Cursor, Replit)[70]
Don't pick those for first-touch
Demo collapses under testing (no planning step)[77]
Force 10 min of "what does the app do" before the first prompt
Beautiful prototype, broken on second prompt[72]
Frequent commits + forked starter so the LLM can pin bugs to a commit

§ 07 · Runbook · the three timestamps

T-48h · T-2h · T-0
T-48h
PRE-FLIGHT · OFF-VENUE
  1. Email attendees: create Lovable + Bolt.new accounts using Google SSO. Include screenshots [52].
  2. Build one reference app in Lovable yourself; save the share URL.
  3. Fork a public starter repo (any minimal Vite/React) — backup if both generators fail [72].
T-2h
ROOM CHECK · DEVICE PREP
  1. Charge your phone. Enable hotspot. Tell venue IT how many SSIDs you need [39].
  2. Print 3 QR codes: Lovable login, Bolt login, the workshop's shared starter repo. QRStuff or Canva — both free, unlimited scans [67] [68].
T-0
120 MIN OF BUILD · 20 MIN OF SHIP
  1. 10 min: "what does your app do?" — forced planning step, ungoverned by LLM [77].
  2. 90 min: build in Lovable. One prompt → preview → one revision. Discourage debug rabbit holes.
  3. 20 min: hit Publish. Generate QR. Show your app to the person next to you.

§ 08 · Related units in this playbook

4 sibling pages
▣ Expedition · Sibling
Workshop-scale app ideas for laymen
Five reliable app-idea menus a non-developer can ship in 2-3 hours, plus the failure modes to swap out.
▣ Survey · Sibling
Prompting workflow: ChatGPT-refine → Lovable-build
Plan a structured PRD in ChatGPT (free, no credits), then paste once into Lovable's Agent mode.
▣ Expedition · Sibling
Take-home deployment for non-technical participants
The deploy click is solved; database pauses, credit caps, and auth rot are what break the app after.
▣ Recon · Sibling
Facilitation logistics and failure modes
Pre-flight checklist, room/tool fallbacks, and the failure modes that actually sink 2-3 h workshops.
SPEC SHEET · alternate view of the canonical research page · 78 citations · prices shown in EUR (~0.92 EUR/USD May 2026), USD per source for reference · last revised 2026-05-23 · all figures inline-cited; full ledger lives on the canonical page