Y Combinator's Orchid AI Draws Backlash Over Second Brain Assistant Ad

In late July, a three-person startup from Y Combinator's Spring 2025 batch posted a two-minute video to X, and by the first week of August it had become the most-hated AI ad of the season. The ad, published July 28, follows a woman whose partner forgets their anniversary; she texts Orchid, which contacts him, books a dinner reservation, arranges a flower delivery, and updates her on his progress. The company's tagline — "the first assistant that actually gets you. It remembers what you forget. Down to the flower" — was meant as a hook. Instead, it became a lightning rod for anxieties about what happens when AI stops automating workflows and starts automating care11.

What Orchid actually is

Strip away the cinematic framing and Orchid is a text-message-based executive assistant. It lives in iMessage and SMS, drafts email replies in the user's voice, books meetings, prepares users for calls, plans trips, sources gifts, handles invoices, and remembers personal details14. Its stated audience is professional, not romantic: the company's FAQ pitches the product to "founders, investors, lawyers, agents, and agency owners" whose time is better spent on substantive work, and its only published case study describes helping an executive prepare for client calls14.

The company behind it is small by any measure. Orchid was founded in San Francisco in 2025 by Nizar Abi Zaher and Adam Wazzan, previously the builders of Zero, an open-source AI-native email client, and it went through Y Combinator's Spring 2025 batch with a team of three27. Tracxn's company profile lists total funding of roughly $500,000 across a single seed round with one institutional investor, Y Combinator27. That's a detail worth dwelling on: this is not aunicorn, not a heavily capitalized lab, and not an incumbent — it is a pre-seed-sized project whose marketing momentarily set the tone of a global conversation about AI and intimacy.

The name "Orchid" has also created some confusion in funding databases, where at least four distinct companies share it: an AI-powered electronic health record for mental health clinicians that has raised about $17 million from investors including Andreessen Horowitz and a16z-adjacent angels1; a New York-based identity-security startup called Orchid Security that raised a $36 million seed round in January 2025 co-led by Team8 and Intel Capital, counting Costco and Repsol among its customers37; an Israeli-adjacent deep learning research lab in London; and the iMessage assistant at the center of this story27. None of these are the same company, and coverage of the backlash has largely kept them distinct. The assistant startup's modest raise stands in sharp relief against the larger AI funding environment, where venture investment into U.S. AI startups hit a record $97 billion in 20248.

The ad that crossed a line

The launch video is built "around a real situation," the team explained in a behind-the-scenes clip — a twentysomething man engrossed in video games while his anniversary slips by11. His partner texts Orchid, which books a restaurant and orders lilies, remembering her preference over roses. The kicker exchange, rendered entirely in lowercase and emojis, has already become shorthand for the whole controversy: the man texts that he can handle it, and Orchid replies, "you literally can't. that's why I'm here"1119.

The response was immediate and overwhelmingly negative. Instagram commenters asked what the point of a relationship is "if you outsource all the effort and compassion," called the product "disgusting and dystopian," and warned that outsourced affection "doesn't end humanity is doomed"1114. Critics on X and Reddit framed the tool as a service for people who want to subcontract the basic acts of care that constitute a relationship in the first place1421. One executive assistant on Reddit wrote that their "heart kind of broke" at the thought of the world turning toward tools like it18.

A second, since-deleted ad made the criticism sharper. It suggested an AI agent that could scan your food for allergens before you eat — with a script line about the list of things that have put the narrator in a hospital — and was panned almost as quickly for the obvious dangers of trusting a chatbot to vet food in cases of severe allergy1113[16. Even sympathetic readers of the campaign flagged the allergen pitch as a categorically different risk profile than restaurant bookings, and the company quietly pulled it16.

The technical wrinkle that both critics and defenders keep circling is that Orchid supports multiple people communicating with the same agent. Co-founder Abi Zaher has said the feature exists to enable shared contexts like group chats and meeting coordination, and that the anniversary scenario was built to showcase it14. But the ad shows a couple interacting with the assistant through two separate private conversations, passing information between partners and reporting one person's actions to the other — which raised questions the marketing never answers: what has each partner consented to share, can either access the other's messages, and how does Orchid decide what it may report back21. Orchid's own documentation is, by its account, conservative — email replies stay in drafts until sent, calendar holds require approval, data isn't sold or used to train models — but the multi-user case remains ambiguous14.

Where the coverage agrees and diverges

The reporting agrees on the essentials: the ad exists, it went viral for negative reasons, the backlash is about emotional labor rather than technical capability, and the allergen ad was pulled. Where accounts diverge is in interpretation. Euronews' write-up treats Orchid as symptomatic of a broader drift — consumers already use AI for mundane life tasks, and the startup is simply the most awkward expression of that11. Archyde's synthesis leans on the defense: the product is an administrative tool, the founders say the ad was a capability demo, and the outrage is aimed at marketing rather than function14. The most pointed reading, from AINave, argues the episode shows a real boundary in consumer AI — people will accept help with logistics but reject automation of effort in relationships, because the value of a gesture lies in the human work behind it16. Mashable lands closest to the cultural nerve: an assistant can book the table and order the flowers, but the gestures matter because a partner remembered, and in the video the human gets the credit while the machine does the remembering.

The uncomfortable context Orchid is riding

Whatever one makes of the ad, the demand it gestures at is measurable. Economists at OpenAI and Harvard analyzed 1.5 million ChatGPT conversations between 2022 and 2025 and found that roughly 75 percent involved non-work tasks — practical guidance, information seeking, writing, cooking help, finding a pharmacy1116. European survey data is more specific to the romantic edge: about a fifth of European adults have used AI chatbots for romantic conversations, and over a quarter believe young people could fall in love with one14. Dating app happn has branded the "AI situationship" an emerging trend, with chatbots serving as emotional mirrors and spaces for attentive listening1114. U.S. survey data points the same direction, with 26 percent of American singles reporting AI use in their dating lives in 2025 and 44 percent of married adults having sought relationship advice from AI14. Orchid did not create this market; it just made the pitch unignorable.

For an investor lens, the episode is a case study in capital allocation during the AI boom. With VC money flowing into everything from identity security ($36 million seeds7) to clinical scribes ($17 million1), the marginal iMessage concierge was always going to be judged on narrative rather than revenue — Orchid publishes no pricing at all[. The backlash delivered tens of millions of views in days, which in a venture context is not nothing14. Whether that converts into trust — the only currency a product with this much access to intimate communications can trade on — is the open question. As one Reddit skeptic put it, the ad read like a Black Mirror episode; as another noted, most of these products end up as tools that assistants use, with humans left to catch the hallucinations18.

The read here: Orchid's founders demonstrated a genuinely hard technical capability — multi-user shared agent context — and then wrapped it in the one use case guaranteed to make people feel replaced rather than helped. The backlash isn't about whether AI can book a table. It's about who gets to be thoughtful.