Why We Built RSVP'd: The Agentic Wedding Thesis

Table of Contents

• The Premise
• Why Weddings, Specifically
• Chat Is Not the Product
• Why Couple-Funded, Not Vendor-Funded
• What Agentic Actually Means in Our Stack
• The Cultural Depth Problem
• Where This Goes
• FAQ
• Sources and Further Reading

The Premise

Every few years, a category of consumer software gets a chance to reset. Travel did it around 2008 with mobile. Finance did it around 2015 with neobanks. Weddings have not had their reset. The Knot went public in 1999, merged with WeddingWire in 2018, and today looks and feels like a content portal glued to a registry and a vendor directory. Zola launched in 2013 as a registry-first brand and has spent a decade adding adjacent features around that core. Joy is the prettiest of the incumbents and still functions, at its heart, as a wedding website builder.

I started RSVP'd because the gap between what these products do and what a couple actually needs is enormous, and because the tool that closes that gap is not another content platform. It is an agent.

Why Weddings, Specifically

The US and Canadian wedding market is roughly $72B annually across 2.24M weddings. Ninety-two percent of couples are Millennial or Gen Z. Eighty-five to ninety percent plan digitally. Fifty-nine percent describe planning as overwhelming. Sixty-two to seventy-three percent go without a professional planner, mostly because the cost of a full planner ($5K to $15K) is unjustifiable next to a wedding that already costs $33K on average.

That last number is the entire opportunity. There is a massive middle between "DIY with seventeen browser tabs" and "hire a $10K planner." The middle has been served by software that organizes information but does not act on it. You can use The Knot to find a florist. You cannot use The Knot to email that florist, get three quotes, compare them, and flag the contract clauses that should concern you. That gap is where the money and the stress live.

Thirty-six percent of couples already use AI in some form during planning, and that number doubled year over year. Couples are not waiting for permission. They are already pasting vendor emails into ChatGPT and asking it to draft replies. They are already using Midjourney to mock up floral moodboards. The behavior exists. The product that serves the behavior does not, at least not from the incumbents.

Chat Is Not the Product

Most "AI wedding planner" features launched in 2024 and 2025 are chat interfaces on top of GPT-4. You type "help me write a vendor email" and it writes one. That is useful in the way a search box is useful. It is not a product.

The thesis behind RSVP'd is that chat is the least interesting thing AI can do for a couple. The interesting thing is agency. An agent is software that holds a goal, breaks it into steps, takes actions in the real world, waits for results, and adjusts. For weddings, the actions are:

• Discover vendors in your region that match your style, budget, and date
• Draft outreach emails in your voice and send them from your address
• Parse the replies, extract pricing, and queue follow-ups
• Read a contract PDF, surface the three clauses you should push back on, and draft the reply
• Track deposits, final balances, and deliverables against your actual calendar
• On the day itself, coordinate the timeline, text late vendors, and update the next person in line

This is not chat. This is staff work. The category we are building in is not "AI wedding planner." It is "the coordinator you would have hired if the coordinator were $49 a month instead of $2,000."

Why Couple-Funded, Not Vendor-Funded

The Knot, WeddingWire, and Zola make most of their money from vendors. Vendors pay to be listed, boosted, or recommended. The incentive structure points at the vendor: more leads, more ad spend, more upsell. The couple is the product, not the customer.

I wanted the inverse. RSVP'd charges the couple $49 per month for the wedding tier. Over a 15-month average engagement window, that is $735 of LTV per couple. It is enough to build a real product around the couple without taking a dollar from any vendor. When the agent negotiates on your behalf, I want you to trust that it is negotiating for you, not for the florist who paid for premium placement.

The math also works. A planner tier at $399 per month ($4,788 per year per planner) sits on top for the professional coordinators who onboard 10 to 30 couples a year at zero CAC to us. That is the second leg of the business. Planners bring couples, the agent handles the volume, and we never need to sell ads against search results.

What Agentic Actually Means in Our Stack

I get asked a lot what "agentic" means in the RSVP'd context, because the term has been hammered flat by marketing. Here is the concrete version.

When you connect your Gmail (OAuth, read-only by default, send-as opt-in), the agent does three things at once. It enriches vendors from the emails already in your inbox. It routes incoming vendor replies into the right pipeline stage (researching, contacted, meeting, negotiating, booked, contracted, completed). And it drafts replies using Claude Sonnet for the creative work and Haiku for the classification work. We fall back to Gemini Flash when latency or cost demand it. The drafts are reviewed by you. The sends happen from your address. Nothing leaves without your approval, for now.

That architecture is not glamorous but it is the actual product. Classification per email costs about $0.0003. A full planning cycle for one couple runs in fractions of a dollar of inference, not tens. That cost structure is what makes a $49 subscription viable without taking vendor money.

The Cultural Depth Problem

The other reason I built RSVP'd is that I am married to Negin Kazemian, a Persian woman, and our wedding had to thread two cultural frames. Most wedding software treats "cultural wedding" as a tag. They offer a checkbox for "Indian wedding" that surfaces the same article from 2018. The actual needs are structural. A Punjabi Sikh wedding has six to eight distinct ceremonies across three to five days, each with its own vendor set and timeline. A Persian Sofreh Aghd has an entirely different object list than a Western ceremony. A North Indian wedding and a South Indian wedding share almost none of their ritual vocabulary.

RSVP'd models these as first-class ceremony graphs: Western, Chinese, Punjabi Sikh, Hindu North, Hindu South, Persian, and Custom. The vendor discovery, budget splits, and day-of runbooks adjust accordingly. This is the kind of thing a content platform cannot fake and an agent has to get right, because if the agent drafts a Sangeet playlist inquiry to a DJ who does not do Bollywood, you have wasted a week of the planning window.

Where This Goes

I do not think RSVP'd is the final form of wedding software. I think it is the first version of what every planning category will eventually look like: a thin, opinionated interface over a fleet of agents that hold the goal and do the work, priced to the person benefiting from the work, not the person selling into them.

The next six months are about making the core agent loop better, not wider. Better vendor discovery, better contract parsing, better negotiation memory across the engagement window. After that, we open up the non-wedding events tier (free forever, that is the Story B pivot), because corporate and milestone events are a lower-stakes training ground for the same agents.

If you are a couple reading this, thank you for giving us your 15 months. If you are a planner, we built the $399 tier for you specifically, come see it. And if you are another founder in this space, I would rather compete on product than on ad spend. That is the whole point.

FAQ

Sources and Further Reading

• The Knot, "Real Weddings Study 2025," theknot.com
• WeddingWire, "2025 Wedding Industry Report," weddingwire.com
• Statista, "Wedding Services Market Size US 2024," statista.com
• Brides, "The Average Wedding Cost in 2025," brides.com
• McKinsey & Company, "The State of AI in 2025," mckinsey.com
• Crunchbase, "Wedding Tech Funding Rounds 2023-2025," crunchbase.com
• Pew Research, "Millennial and Gen Z Digital Behaviors," pewresearch.org