LoopcardMarket study — September 2026
Date of record: 18 September 2026
Method: Demand first. A job is real only if a named person already feels it, already pays someone or has a documented invoice plus a shipping price, the incumbent is named, they buy the work not a protocol, and 100–250 paying accounts at $19–99/mo is plausible from this domain with PLG only.
TAM slides are not demand. Bain’s ~$100B US “agentic glue” TAM (7 May 2026) and MarketsandMarkets’ $19.33B “agentic AI” 2026 figure (20 Aug 2026) are analyst envelopes. Nobody swipes a Visa into a percentage of those.
Raw evidence: /root/cryptobook-next/research/ and /root/cryptobook-next/decision/MARKET-STUDY.md.
1. Where money actually moves
1.1 Consumer / prosumer (card, this week)
Menlo Ventures, 2026 State of Consumer AI (GlobeNewswire 16 Sep 2026; 5,067 US adults, July 2026):
- Global consumer AI spend $40B in 2026, from $12B in 2025.
- 55% of US AI users pay for at least one AI product.
- 14% of payers spend $100+/month and account for ~60% of US consumer AI spending.
- 41% of US AI users have tried an agent; 24% use one regularly; 92% of regular agent users pay for AI.
What they swipe (vendor pages opened 18 Sep 2026):
| Product | Public price | Job |
|---|---|---|
| Cursor Individual | Pro $20/mo; Teams $40/user | Write code with an agent |
| Claude | Pro $20/mo; Max from $100; Claude Code included | Chat + coding agent |
| GitHub Copilot | Pro $10/mo | Code in the editor they already use |
| Lindy | Plus $29.99/user/mo | Inbox / Slack / meetings teammate |
| Lovable | Pro from $25/mo | Chat → web app |
| n8n Cloud | Starter €20/mo annual | Run workflows |
| Langfuse Cloud | Core $29/mo (MIT self-host $0) | Traces for an LLM app they already ship |
1.2 Builders (habit, not a third-party SKU)
LangChain State of Agent Engineering, 12 June 2026 (1,340 responses):
- 57.3% have agents in production.
- Top barrier: quality 32%. Cost dropped vs prior year.
- Observability: 89% already have some form.
- Daily agents: coding agents dominate (Claude Code, Cursor, Copilot, Windsurf).
1.3 Enterprise governance (real, not our buyer)
- Okta Agent SSO GA 24 Aug 2026, $0 extra on core SSO (~20,000 customers).
- IBM watsonx Orchestrate / AgentOps, Broadcom AgentMinder, Microsoft Agent 365: sales-led.
- A 1-VPS product that looks like “governance” is compared to free Okta and ignored.
1.4 Product Hunt, September 2026
Monthly leaderboard (opened 18 Sep 2026): #1 tiun (auth/billing for AI builders), coding agents, Slack bridges, Naoma (demo agent, $249/mo — out of our price band). Nothing in the top 10 is an MCP gateway, blast card, signed receipt, or quantum classroom.
PH upvotes ≠ cards. AI Toolbox 3.0: 410 upvotes; maker wrote 23 people reached the pricing page.
1.5 Quantum
IBM Quantum (opened 18 Sep 2026): Open Plan free (≤10 min QPU/month, unlimited simulators). PayGo $96/min. Classroom Accounts (28 May 2026): 5–100 students, no credit card, real QPUs. This VPS is a ≤32-qubit simulator. Kill.
2. Jobs that look hot and do not clear a card
| Pretend job | What people do this week | Card evidence |
|---|---|---|
| English MCP permit + spend cap + receipt | Cursor spend limit, Claude caps, Anthropic console | Permit.io MCP Gateway Show HN: 2 points |
| Signed deny receipt | Screenshot, Loom, git diff | Receipt Show HN: 9 points |
| Agent budget as MCP | Hit Cursor’s own ceiling | OSS after a $1,263 Cursor bill; HN 6 points |
| Blast-radius / MCP scan | mcp-scan, CodeFlowOps free 30s grade |
ShieldMCP $49 PDF, 5 PH upvotes, 0 reviews |
| Quantum classroom | IBM Composer | Negative WTP vs free real QPUs |
The $1,263 Cursor bill is real pain. The founder shipped OSS, not a $19 subscription. The incumbent for “stop the runaway loop” is the vendor they already pay.
3. Distribution on this domain
nginx access log for cryptobook.space, seven days to 18 Sep 2026:
- ~1,503 hits
- ~65 unique IPs
- Paths:
/and/api/posts/...(old social)
This is not a Cursor audience. SEO from a crypto-social brand into “MCP permit” converts like a casino site selling CI tooling.
4. Application jobs (do the work)
A dedicated pass found zero application jobs that were simultaneously: paid this week, 1-VPS, 60-second, not high-risk AI Act, not US-illegal, not a clone of Intercom/Sierra/Glean/Perplexity/Cursor, and profitable to the user.
Closest misses (all killed): Monday competitor brief (Figo launched 17 Sep; Outmano $49), agency Friday PDF pack (Claude vision), invoice chase drafts (ChaseAI $9; sending is legal kill), Google review replies (RepliFast $29), email → SOW (ChatGPT is the 60s product).
Pattern: if a human is paid this week, a $9–$49 agent already exists; if not, ChatGPT is the SKU; if the agent must act in the world, you hit Sierra / Naoma / CFAA.
5. Conclusion
NO-GO. Demand exists in the agentic stack. It does not exist, at $19–99 self-serve, for a new sidecar on cryptobook.space. Building Slip (or any governance collage) would be a unused website.
Sources opened 18 Sep 2026 include: Menlo GlobeNewswire, Bain PR (7 May 2026), MarketsandMarkets PR (20 Aug 2026), LangChain State of Agent Engineering, cursor.com/pricing, claude.com/pricing, lindy.ai/pricing, langfuse.com/pricing, permit.io MCP gateway pricing + Show HN 47426690, ibm.com/quantum/pricing, Product Hunt September 2026 leaderboard, codeflowops.com, ShieldMCP Product Hunt, cryptobook.space nginx logs.