An event planner wants software that tells her what to charge

Sara Rojas's second product request is a pricing engine for events and catering: put in the rentals, the plates and dishes, the food, and the hours of labour, and have it tell you what you should be charging. She has learned to get this roughly right over years of practice. Her argument is that a newcomer should not have to.

What she asked for

Her framing was a dream tool — the thing that would make her wake up and think technology had done its job. What she described:

"It would be nice to know the pricing of things like rentals of plates and dishes and like the food aspect and the catering, like and how, you know, the hours of labour. Like if I can put this all and kind of they can generate, hey, this is what you're supposed to be making."

The output she wants is not a cost total. It is a price — what to charge, given what the job actually consumes.

The failure mode it prevents

This is the part worth understanding before building anything, because it explains why a calculator would not be enough. Her diagnosis of how event planners lose money is behavioural, not arithmetic:

"As an event planner you're only looking at it making pretty and like it looking nice, and you're like, if I got this one it looks better and I got this one, but really all that time and effort at the end it's not worth it."

The planner is optimising for how the event looks. Each upgrade is individually defensible. The cost of the accumulated upgrades, plus the labour they add, only becomes visible after the event. She describes the same pattern in her flower work — costing the stems and not the hours, and hitting herself over the head about it repeatedly.

So the requirement is not a sum. It is a tool that surfaces the labour and the add-ons at the moment of the decision, when they can still change the quote.

Why she wants it for other people

She is explicit that she has mostly solved this for herself, through what she calls practice and hitting herself over the head. Her reason for wanting it built is that someone entering the industry has no such buffer:

"It would be amazing for someone that just went into this industry to have that tool all in one, so they actually start making a profit."

That is a market signal, not a courtesy. The buyer is not the experienced planner who has already internalised the numbers. It is the one in year one, who is currently learning by underquoting.

The volatility that has to be in it

A static cost database would fail her within a season, because the largest input moves. Flowers are the recurring investment in event planning — she notes that equipment eventually pays for itself over enough events, while flowers are bought again every time. And the price is not stable:

• It moves with the season
• It spikes around holidays — she names Valentine's Day, when everything is hiked up
• She used to quote off-peak prices into holiday weeks and absorb the difference

She knows to check now. Knowing to check is not the same as being warned, and a pricing tool that does not carry live inputs would reproduce the exact error it is meant to catch.

What already works, and what it tells you

She has assembled a partial version of this herself. Payroll, sales ratios, food costs and store pricing all go into ChatGPT, and she gets a plan back in about five minutes. What that establishes is that the reasoning is not the hard part.

The hard part is the inputs: rental prices across suppliers, current wholesale flower prices, and a labour model that reflects how long an arrangement actually takes to build. Those are data problems, not model problems, and they are the reason this is a product rather than a prompt.

Where this comes from

S2E2: The secret AI strategy this small NYC business uses to save thousands — the full interview with Sara Rojas and Luisa Ramirez. Listen or watch: YouTube, Spotify or Apple Podcasts.