jimmy·research
Home/Agentic Commerce/Magic
EN↔中文
Company

Magic

The mythological 2015 YC SMS concierge — one text can get you anything; today quietly pivoted into a B2B virtual assistant outsourcing company.

Disambiguation: this entity is the SMS concierge company Magic / getmagic.com (YC W15, 2015), distinct from Magic.dev (the AI coding assistant / long-context model company from 2024, unrelated to this page).

1. Core Product / Service

Original form (2015): a phone number — no App, no website login. Users texted any legal request (book a flight, buy flowers, find a plumber), and a human + AI backstage team would quote and execute within minutes; users replied to confirm and got charged. Fully parasitic on iMessage / SMS — the proto-template for no-app-commerce / imessage-commerce.

Current form (2026): completely pivoted to Magic Virtual Executive Assistants, focused on B2B remote virtual assistant outsourcing, monthly subscription, people-not-AI, positioned closer to Upwork / Fiverr-style middlemen. SMS conceptually still reachable (short code 83489), but product focus has moved away from text concierge [1][2].

2. Target Users & Pain Points

  • 2015 flagship era: Silicon Valley high net worth individuals / founders / executives. Pain point: "too lazy to run errands themselves", willing to pay a premium for time.
  • 2026 today: SMB owners and executives. Pain point: "can't afford or don't want to manage a full-time EA", needing ops outsourcing, not an agent.

Community feedback (HN 2022) consistently describes the company's trajectory as "high-quality concierge → low-quality VA" — neither cheap enough to compete with rote work, nor smart enough to handle complex research; squeezed in the middle [3].

3. Competitive Landscape

Competitor Fate Difference vs Magic
Operator (Garrett Camp / Uber co-founder, 2015) Native app + real human ops, closed 2017 Heavy UI, differentiated on "product discovery", couldn't make unit economics work
Mezi (2015) Pivoted from full-category shopping to travel, acquired by American Express for ~$150M in 2018 Took the path of professional vertical (travel) + selling to enterprise users; one of the few happy endings
Assist Similar local-life / travel concierge, now dormant Same-ecosystem ally
ChatGPT / Claude (2023+) The biggest current dimensional reduction attack LLMs natively handle "information organization + simple errands" needs, free + instant
attentive / OneText Another branch of SMS commercialization Concierge sells time; Attentive/OneText sells direct commerce

Conclusion: the entire "text-to-anything" race of 2015 ended one of three ways — acquired (Mezi), shut down (Operator), or pivoted to traditional outsourcing (Magic). The reason: after LLMs arrived, the AI + human hybrid middle business form got squeezed from both ends — complex / high-net-worth needs got captured by LLM + Notion / ChatGPT for the information layer; low-end rote-work needs couldn't beat Upwork / Fiverr on pure labor pricing.

4. Unique Observations

  • Historical significance outweighs current relevance: Magic is the same origin point of three threads — no-app-commerce / ai-human-hybrid / human-in-the-loop-ai — in 2015 when SMS bots had almost no LLMs, they brute-force ran the path of "natural language → human backstage → any service", and is the spiritual prequel to modern agentic-commerce.
  • YC early valuation anchor: Magic reportedly took the YC standard terms $125k / 7%, implying ~$1.78M post-money [4]. This is exactly the "seed floor" shared by early SMS / iMessage companies in agentic-commerce.
  • The AI-human-hybrid death spiral: judging from Magic's trajectory, in the LLM era pure hybrid models have only two survival paths — either push AI proportion above 95% (become a pure LLM product), or push human proportion to 100% and verticalize (become precision outsourcing). Magic chose the latter and lost its brand premium.
  • The winners of SMS / iMessage commercialization in the US today aren't concierge — they're attentive / Postscript / OneText, tools that sell marketing services to brands without directly telling users a story.

5. Financials / Funding

  • 2015: YC W15 batch; in May the same year announced Series A $12M, led by Sequoia Capital [1].
  • Subsequent rounds: no public large follow-on funding disclosed; the company operates quietly.
  • 2023 revenue: third-party SaaS database Latka estimates ~$97.9M ARR / 339-person team [2], implying $290k/year outsourcing revenue per head — a healthy human services company, but far from the original SMS concierge story.
  • Current valuation: undisclosed; no longer treated as an AI / agentic commerce name.

6. People & Companies

  • Founders: Mike Chen (CEO), Aaron Kemmer, David Merriman, Nic Novak, Michael Rubin — five who originally did the blood pressure monitor app Bettir in YC W15; Magic was a weekend side project that unexpectedly took off and they flipped the table to it [1][5].
  • Investors: Sequoia Capital (Series A lead, 2015); Y Combinator (W15 batch).
  • Historical competitors: Operator (shut down), Mezi (acquired by AmEx 2018), Assist.
  • Ecosystem-related: attentive (same SMS commercialization infrastructure layer, but goes via marketing rather than concierge).
  • Modern dimensional reducer: ChatGPT / Claude-class LLMs directly absorbed the "find info + automate" demand pool.

Sources

Related

attentive
Last compiled: 2026-05-09