Nowadays co-founder & CEO Anna Sun discusses...
- Meeting planning frustrations
- How Nowadays differs from legacy
players
- Working with AI from the ground up
- Developing a roadmap on client
feedback
Nowadays, an
AI-powered meeting planning technology startup, took home both trophies up for
grabs last month at the Business Travel Show America Innovation Faceoff:
Innovator of the Year as voted by the Innovation Faceoff judging panel, and the
People’s Choice as voted by the Business Travel Show America attendees.
Nowadays prevailed against 11 other competitors spanning solutions for
transient travel, payment and expense management. Co-founder and CEO Anna
Sun sat down with BTN VP of content Elizabeth West to discuss the
win, the technology and the concept of addressing old problems from the ground
up with new solutions. The following is an edited transcript.
BTN: Like many meetings technology
innovators, your starting point was your own experience and frustrations with
organizing meetings.
Anna Sun: I studied computer science at Massachusetts
Institute of Technology, but I was planning meetings there as well. My sister
was planning meetings [as part of her job at] Google. We co-founded the company
recognizing the tedious back-and-forth emailing of trying to find and negotiate
with venues, collecting quotes and getting bumped afterward. We wanted a
process that could get all that information in one place where it was easy to
compare data apples to apples. I can remember spending hours copying things to
spreadsheets and figuring out the math to compare things. So Nowadays makes all
that easier by applying artificial intelligence to the problem.
BTN: How does the AI work?
Sun: The short version is that a
client will go into the platform and put in their event details and the AI will
just take it from there. We have a database of 400,000 venues, which is
inclusive of all hotels in the world, and the AI filters down to the best fit
based on what the client is looking for—for example, is there enough natural
light in the meeting space or are there microwaves in the guest rooms. The AI
narrows down the choices and best options, and it can automatically email and
call the venues with AI to start the request for proposal process, get the
quotes into email and then the AI reads the proposals, extracts key data like
nightly rates, attrition policies, cancellation, etc. And then it can compare
apples to apples for you in our dashboard.
BTN: Am I correct in remembering that
the platform will digest attendee home cities and location costs and return a
dashboard of cost and convenience options based on all that data?
Sun: Right. So we need to know
certain details. If it’s a nonprofit, they probably aren’t going for a
five-star hotel, but an executive offsite does want that. So taking those
factors into account, the user provides an anonymized attendee list of where
people are coming from, and then we can estimate how much the flights are going
to cost based on historical data, based on layovers; the location itself might
have a certain vibe or kind of weather you want, and then when we get down to the cost of the venues, the
average cost of meals and coffee breaks, we can ballpark that budget to
recommend the best fit.
BTN: Four hundred thousand is a lot
of venues. How are you capturing that granular level of detail in your
database? And how reliable is it?
Sun: We directly partner with convention
and visitors bureaus and hotel chains and independent venues to get the most up-to-date
data. We also partner with Northstar Meetings Group which sources all of their
venues through us. Through that partnership, we're able to make sure things are
up to date. But our agents are actively updating data all the time through
Google searches. “Hey, what are the top venues in Thailand that can fit 500
people,” etc. The agents also go through hotel websites, read what's updated
and then update our database. Also, when hotels email our system new proposals
[for clients], they often will send new promotions and new renovations that
have happened. Our AI automatically updates that in our database.
BTN: Just to be clear, when you say
“agent,” we should always understand that as “AI agent” or “bot”?
Sun: Correct.
BTN: SkyLink CEO Atyab Bhatti, one of
the Innovation Faceoff judges, said about Nowadays that you were solving old
problems in new ways and looking at it from a “first principles” angle, which
is unlike legacy tech that often overlays AI onto its existing solutions. Tell
me more about building natively with AI.
Sun: We've seen a lot of legacy
platforms integrating AI. Often that takes the form of, “OK, we have some data
our clients want to understand better. Let's build an AI chatbot on top of
that, so they can summarize data and extract the most important details.” That's
almost like a nice-to-have, because the hardest part [with meetings] is doing
the manual logistical work. It's hard to build AI agentic workflows if you
already have that manual workflow in place. You're ripping out all the software
you've already built. When we think about a problem now, we think about how
does this scale in 10 years so it's completely automatic. Our vision is that
there will be hotel AI agents that are communicating with our AI agents for
clients, and it will be completely automatic. We always try to build toward
that.
BTN: Many times in a negotiation
there’s an organic development in a conversation between two people. If a
certain thing is available, it could re-stack the deck of priorities for either
side. Can agentic AI account for that kind of non-linear decision-making?
Sun: We are building user profiles to
learn the typical decision factors. Do they care about the quality of the food
the most or the how nice the meeting spaces are? I think it really helps. Plus,
we have so much access to data from all the other contracts that our clients
have negotiated and what the standard list of concessions are, so less experienced
meeting planners who [occasionally] organize events might not ask those types
of questions. The technology does a lot of double-checks on priorities but also
saying to the organizer, “Hey, make sure you're not just signing the contract
as is, always try to negotiate and get the best rate.” Of course, if you're
more experienced, you have that list and those checks can be automated through
Nowadays.
BTN: AI runs on massive amounts of
data. You are collecting user data from their contracting and other activities
in the platform to inform your development. Some companies might not love that.
Sun: We definitely anonymize the
data. Some companies want us to learn from their experiences so the next time
they plan that VIP kickoff, we know exactly what they care about. It's almost
like they have their own AI model trained for them and accessible only to them.
In terms of other clients, we just make sure that once we anonymize that data,
it's mostly to collect hotel proposals and help us build up a database of how
we can predict what hotels are going to give us instead of having to request it
every time. We do that so we can give more real-time feedback to our users.
BTN: At the very end of your Faceoff
presentation, you offered—and maybe challenged—attendees to bring you problems
to solve. How do you stay agile to customize like that?
Sun: Our whole company is based on
client feedback, and we started the company by first doing 200 user interviews
before coding anything. We want to understand the client problem and how they
envision a solution for themselves. If we talk to 50 people and they've all
asked for this certain customization, of course, we will build it. There are
smaller customizations that might be geared more toward one company versus the
broader [universe], but because we are very fast at building—we're a team of
MIT and Ivy League engineers—it sometimes just takes us a couple of hours to
push out a change. We might as well make sure a potential client can have
everything they need to get the green light.
BTN: I assume you are using AI code
to help you develop?
Sun: Definitely. I don't know how
deep you want me to go into it. It's pretty crazy. You can imagine I have 34
monitors up and each one is coding while I'm doing something else.
BTN: Who is your target client? Is it
the small to midmarket, or are you working toward larger enterprises?
Sun: We were in Y Combinator and
we're in Silicon Valley, and [my sister] Amy and I both come from a tech
background. Our first 100 users were tech companies ranging from 10 to 500
people planning offsites and meetings. Their feedback really helped us iterate
quickly because they understand tech and they were willing to adopt it early. Now,
we've built the platform to a place where more corporate meeting planners [can]
understand it and we, also, understand the problem better so we can sell
better. On the other hand, we’re a team of engineers versus a team of business
finance people trying to solve a problem to make money. That’s a big difference
we’ve seen at conferences. I’m always the youngest in the room and usually the
only one with a computer science background.
BTN: But you want to make money, too—what’s Nowadays’ commercial model?
Sun: We charge a percentage sourcing
fee to our clients, and we also make commission from the hotel side. For
enterprise clients or those that are doing 10 or more events a year, we'll go
to an annual fee that makes more sense.
BTN: Are you always going direct to
corporate? Are agencies interested in your tool right now?
Sun: It's a mix of both. We
definitely started direct to corporate, but we've gotten a lot of interest from
third parties that have a bunch of human agents that either use Cvent right now
or manual processes and they want to streamline. Now that we've built the right
tool, it's a good time to start selling to them. We don’t necessarily want to
build completely to the needs of a larger third party. We have a lot of tech we
want to build to address broader client needs, so we are looking at how to
strike that balance.