zh

研究内容为英文. 本页的研究内容（标题、简报、来源）以英文撰写，未经机器翻译。只有界面已翻译。

新想法Developer tools

Cloud iPhone farm for AI agents

Rented cloud iPhones that AI agents drive via MCP/API; crowded (≥8 vendors); Hold.

1周前 · 评分于 2026年11月6日 · ID HBV-261106-02

暂定标题： Cloud iPhone farm for AI agents (working name: Poki-powered cloud iPhone for LATAM agents)

摘要

简报已验证

Hosted farm of real iPhones agents drive over MCP/API. Benchmark poki (JosssphZhou/poki, AGPL-3.0) 358 stars ~1 day after creation (Verified, github.com HTML 6 Nov ~11:16 BRT). Crowded: ≥8 hosted agent-iPhone/iOS vendors live (Invi Labs, Phone Harness, Mobilebox, Mobilerun, Mobile Next, Limrun, Fonino, Agentfy) — Verified on vendor pages.

描述已验证

A hosted farm of real iPhones that AI agents drive over MCP/API. Who: developers and agent builders who need a real always-on iPhone through one MCP endpoint without owning a spare device or Mac/Xcode. Pain: local poki needs macOS 15+, full Xcode, USB pairing; physical support verified only on Chinese-UI iPhones, portrait only (Verified, poki README).…

阅读其余部分

Monetization: market anchors $59/mo (Invi Labs), $64–80/mo (Phone Harness), $0.05/phone-min (Mobilebox) — Verified vendor pages; no HBV price approved. Build: hardware-heavy farm OR software-only wedge on third-party devices after CD. Recommendation: Hold — do not advance as HBV hardware business.

目标客户

Developers and agent builders who need a real iPhone (not a simulator) driven by Claude Code, Codex, Cursor or similar MCP clients.

痛点

Need a real always-on iPhone for agent tasks without owning spare hardware, a Mac with Xcode, or USB pairing.

为何是这个分数

分数较低的原因： Crowded category (≥8 hosted agent-iPhone vendors) + hardware/ops fit weak for HBV software thesis; AGPL licence risk if code reused.

护城河假设

No clear moat identified. Possible defensibility (Hypothesis): localized LATAM niche (local SIM, app store region, language) or self-learning routes per app. Status: Hypothesis — not validated.

需求信号

X 上的帖子（7 天）

514

信号趋势

未知

网络需求

n/a this run

市场与竞争对手

TAM

Unknown — Scope: Agent-driven remote mobile devices. Method: no public sizing found this run. (Basis: n/a)

SAM已验证

Unknown — Scope: Developers/agent builders needing a real iPhone driven by Claude Code/Codex/Cursor. Method: price anchors exist ($59–80/mo dedicated; $0.05/min shared — Verified vendor pages); user counts unknown. (Basis: invilabs.io/rent · phone-harness.com · mobilebox.dev/ios)

SOM估算

Unknown — Scope: Share reachable by a late entrant against ≥8 incumbents. Method: not estimated. (Basis: n/a)

竞争对手已验证自行申报

Invi Labs ($59/mo hosted iPhone); Phone Harness ($64–80/mo, YC-backed per its site); Mobilebox ($0.05/phone-min iOS); Mobilerun; Mobile Next (claims 1,000+ real devices); Limrun (cloud iOS simulators); Fonino (US-only hosted iPhone); Agentfy (iPhone fleets) — Verified on vendor pages; figures Self-reported by vendors where noted. Plus BrowserStack, Sauce Labs, LambdaTest, Kobiton device clouds.

商业模式

变现方式

Hosted device rental / usage (market anchors only; no HBV price approved).

收入假设（每月）假设

Not sized — crowded category; Hold (no revenue hypothesis advanced).

待完成的任务

When my AI agent needs to finish a task inside an iPhone-only app, I want to hand it a real, always-on iPhone through one MCP endpoint, so I can automate the task without owning, pairing or babysitting hardware.

HXC已验证

Agent builder without a spare iPhone or Mac (illustrative persona) · Uses Claude Code/Codex; wants one MCP URL over USB pairing. Buying signal: Hypothetical — competitors charge $59–80/mo per dedicated device (Verified, vendor pages).

构建与 GTM

技术栈

Benchmark: poki (AGPL-3.0) local MCP + agent; hosted path would need real iPhones, Mac hosts, MCP/API gateway. Stack for software wedge TBD after CD.

GTM 渠道

If pursued later: agent/MCP directories, LATAM niche SEO; not active while Hold.

风险

主要风险

Crowded category (≥8 hosted agent-iPhone vendors) + hardware/ops fit weak for HBV software thesis; AGPL licence risk if code reused.

全部风险已验证

Value: Attention — rising interest but incumbents already serve it | Usability: Supported — MCP-over-HTTP is category norm (Verified, vendor docs) | Feasibility: Attention — hardware + Mac hosts; Apple terms risk | Business Viability: At Risk — crowded; AGPL-3.0 if poki code reused in hosted service (licence risk, not a cut)

下一步验证

Decide whether a software-only or regional LATAM wedge is worth a fast Customer Discovery run; otherwise archive. Interview 10–15 agent builders on what current vendors fail at.

来源

  1. github.com
  2. poki.bot
  3. invilabs.io/rent
  4. phone-harness.com
  5. mobilebox.dev/ios
  6. mobilerun.ai/mcp
  7. mobilenext.ai/cloud
  8. github.com/limrun-inc/mcp
  9. fonino.app/docs
  10. agentfy.io
  11. usedevice.cloud list

ideanewiphoneagentmcpcloudagpl-riskcrowded

想在评分当天收到下一个想法吗？

分数、来源、风险和下一步，通过邮件发送。

免费。通过 beehiiv 发送。一键退订。

  • 新想法Developer tools· 0周前

    AgentParity — continuous parity

    A CI service that replays a real agent trajectory against every self-hosted framework release and asserts interrupt/stream/checkpoint/durability parity, failing the build before a regression reaches production — self-serve, $35/repo/month.

    分数较低的原因： Adjacent platforms (LangSmith, AgentOps, promptfoo) ship parity checks for free

    打开 ProductBots.ai →

  • 新想法Developer tools· 1周前

    Embeddable 2.5D avatar player SaaS

    Hosted drop-in 2.5D lip-sync avatar widget for sites (docs, support, marketing) with API/seat billing — player infrastructure, not a creation studio.

    分数较低的原因： Owner LivLiv already monetizes hosting and a partner Web SDK; thin whitespace

    打开 ProductBots.ai →

  • 现有业务Developer tools· 1周前

    NoMac-style copycat

    NoMac.app rents disposable Apple-silicon Macs to AI agents by the minute (US$0.90/h Verified list price); TrustMRR shows about US$2,617 revenue in 30 days (+1,217%). HBV angle: a NoMac-style copycat with LATAM billing and distribution.

    分数较低的原因： Apple licence + hardware economics (fleet cost vs utilization unknown; sub-24h terms Hypothesis) and crowded category

    打开 ProductBots.ai →

这是研究，不是投资建议。数字保留其发布时的标签。