Will generative AI replace the social graph? California-based local experience discovery startup Bigfoot hopes that adding a conversational interface to its weekend planner website (in the form of an AI chatbot it branded Littlefoot, which launched last week) will create a more personalized feel while also making it easier for people to find relevant activities in their neighborhood.
Co-founder Alex Ward explains the idea is to recreate the feeling of chatting with a “knowledgeable friend.”
This feature allows users to interact with Bigfoot's curated local events, activity spots, and nightlife options through text conversations, easily directing and guiding the AI to make the right recommendations for weekend fun: a scenic hike followed by a nearby BBQ lunch and an evening of live music close to home; or a cultural tour of a contemporary art museum followed by a painting class and wine tasting. The startup's mission is to empower people to do more in their local area in their spare time without making the weekend planning process tedious.
While the idea is by no means new (travel accommodation giant Airbnb, for example, has offered an experience discovery service for nearly a decade), the team at Bigfoot believes there is still an opportunity to build a brand around better weekend activity planning tools that cater to local leisure needs, whereas the big platforms have focused on travel and guided experiences (see also: Get Your Guide).
Bigfoot was founded in late 2022 and has raised a small amount of pre-seed funding from several angel investors and venture capital firms. The company claims to have built a supply pipeline of more than 10,000 events, restaurants, nightlife, sports and outdoor activities in 160 cities, which are showcased through a categorized search on the company's website.
A new addition to the company's product is the ability to engage users in contributing ideas that can be actioned locally via an AI chatbot powered by Large Scale Language Models (LLMs), which include AI technologies from Anthropic, OpenAI and Perplexity.
Ward, one of a trio of ex-Airbnb staffers behind Bigfoot, acknowledges that the GenAI interface feature is essentially a re-release of the product: The startup's local events and experience discovery platform was up to about 30,000 monthly users on version 1 of the product when it launched in New York and was investing heavily in marketing, he told TechCrunch over the phone.
As the team needs to secure initial funding to continue building towards their big vision of powering an offers platform to help small businesses bring in more foot traffic through local discovery, there are high hopes for Littlefoot to gain friends and make an impact on those looking for help organizing their next action-packed weekend.
“[At peak marketing] “We were growing at about 40% month-on-month,” he says, “so we proved that there was real interest in this type of product. But we couldn't keep spending money on marketing; we wanted to focus on building the product. We were getting consistent feedback from our early alpha users… [that] Essentially, what they wanted was a product that would let them talk to Littlefoot as if they were talking to a knowledgeable friend.”
The stakes are high, but the battle for consumer attention is likely to intensify, especially with general-purpose chatbots and AI search engines that are being made available for free by some of the underlying model makers whose APIs the startup uses.
For example, if you ask tools like OpenAI's ChatGPT or Perplexity's AI search engine to plan an activity-packed weekend in New York or Madrid, you'll quickly get a ton of suggestions. However, these general-purpose AI tools typically generate fairly general suggestions, choosing the most obvious and touristy ones.
The Bigfoot team believes that the combination of GenAI and over 50 curated “trusted” data sources has the opportunity to recommend more interesting and contextual activities for users and improve local leisure discovery.
Ward said the sources used to recommend weekend plans range from mainstream content accessible through APIs, like Google Maps data, to lesser-known sources like hiking blogs, and the idea is that this mix will allow Littlefoot to produce less inoffensive recommendations.
But they still hire mainstream law masters graduates to do the conversational grunt work of the tools.
“We get supply from over 50 trusted sources around the world, and we also use Perplexity to augment our supply. We connect to Perplexity's API,” he said. “The reality is, [great local knowledge] “It's in the hearts of the people who are saving these things, or in their pins on Google Maps, so we continue to get better and better at this curated supply.”
“But the core of it for us is providing curated information. What we're going to be building out is the ability to connect the Google Maps API, which will make it really easy to add collections if you save a Google Pin. So you'll be able to add new collections of things to do, or create new days based on what you've done in the past.”
“We can also connect to the Spotify API so we can understand the type of music users like. [e.g. to personalize recommendations for gigs and other local music events]And finally, something that's actually already available in the alpha is the ability to send Apple notes or Google maps or create entire collections just from lists.”
The product is designed to give users a rich experience, populating it with images to illustrate recommended events and activities, as well as showing related TikTok videos. So it's essentially the opposite of the dry, bulleted text list you get when you ask ChatGPT to make weekend plans for you. (Though some people who make weekend plans might actually prefer a basic list, and that level of information density makes it easier to cross-reference the suggestions with your own trusted sources.)
But not all general-purpose AI tools are so sterile on the surface. For example, Perplexity's AI search engine serves up a few images by default, with handy links to sources for easy reference. The service is similarly focused on turning your searches into compelling summaries and visual guides. So the difference in product experience from the competition is more baby steps than giant leaps.
Any AI recommendations for the weekend?
We compared Littlefoot’s suggestions with the recommendations provided by a generic LLM using the same prompts. Littlefoot generates some less clear suggestions compared to the generic chatbot.
While OpenAI's ChatGPT's recommendations for a weekend in Barcelona included well-known tourist spots like Park Güell and the Magic Fountain show at Plaça d'Espanya, Littlefoot suggested a variety of more off-the-beaten-path options, including a hike to Parc de Poblès Sec and an “Alternative Modernista Tour,” which promises “a way to discover Catalan modernism less known to tourists.”
But it's hard to say whether the average user will notice and appreciate these nuances. Plus, there was some overlap in some of the recommendations.
People looking for a more specialized weekend experience are also likely to be the hardest to convince, as they already have a large number of trusted sources (i.e. friends who are actually knowledgeable), so Bigfoot's competitors seem more likely to be mainstream, general-purpose search tools.
During testing, the startup's AI tool failed in some pretty obvious ways: for example, when asked for “outdoor” recommendations, the first suggested activity was “indoor go-karting” (according to the event description). In this case, the accompanying photo showed go-karts being used outdoors. So this issue could be attributed to Bigfoot's team (also) relying on GenAI to copywrite the event/activity descriptions, resulting in machine-generated text that was misleadingly general.
However, this kind of contextual mismatch can be jarring and can quickly cause people to lose interest in your new product.
Another example: when we asked the AI to recommend outdoor bouldering spots on natural rock, it returned a few indoor climbing gyms and general fitness gyms… meaning that this more specialized request was handled even more poorly.
Beneath each activity recommendation is a button that users can click to let the AI ”direct my day” — that is, if they like the idea. After a few seconds, a fuller version of their entire weekend plans appears, along with still and video images pulled from public sources like TikTok and a map view of the recommended locations.
It's not clear how the tool's logic works to create a complete plan (i.e., a plan with multiple to-do items) from a single suggestion that you select, but the suggested daily schedule can be edited to swap out different elements (for example, replacing one restaurant with another), and there's a copyable link so you can share the plan with friends.
Overall, the interface feels a bit rough around the edges. Obviously, this is MVP territory and needs a lot of fine-tuning and fine-tuning, especially in the presentation of the plan, the product logic, and the timing of transitions between visual elements. The smoothest aspect of the UX is undoubtedly the text chat with the AI, which is of course the most commoditized.
If the team can iron out some of these pain points and improve usability as they iterate — if they can find that sweet spot between compelling visuals and ease of use — they might be on to something good, but it's certainly a tough problem they're tackling in a highly competitive field.
Even offering higher quality alternatives to weekend plans might not be enough, as people often fall back on familiar platforms out of habit. In other words, scaling a consumer startup today might be like climbing a very big mountain.
“I think the trap that a lot of startups fall into when it comes to local search is expecting users to do the work of adding supply,” Ward acknowledges. “And the reality is, people just want to do something perfect, amazing, unique, curated. So it takes a lot of work to ensure that supply is there and directly address the cold start problem of not being successful.”
“If you expect other people to generate supply or create supply for you, it doesn't work.”
Bigfoot is backed by a small team of six people with ideas for continually enhancing functionality and improving personalization, like allowing users to upload notes and map pins that are then turned into shareable visual plans and collections.
Another idea they're considering is offering a version of Littlefoot that runs inside WhatsApp, giving users access to group chats between friends making weekend plans, but they'll need to raise more funding to continue this effort.
“We are in active discussions about seed funding,” Ward asserts. “We are looking for great partners who share this vision and can help us continue to build. We are a small team that has built a lot, but finding a really great partner that wants to join us on this journey and continue to build with us is what's most exciting.”