seeklocal.

Inspiration

Earlier this year we attended panel sessions with people planning LA28 and Los Angeles' major events, across energy, transportation and infrastructure. Two themes kept coming up: moving equipment between cities is a major source of emissions, and working with local vendors in every new city is slow, manual work.

When we looked at Prompt A, we saw the same pattern in events that tour the world. F1 teams, concert tours and trade show exhibitors ship the same chairs, glassware, fridges and screens to every city they visit, even though each city already rents them. For F1, moving equipment was 45% of the sport's entire carbon footprint in 2019, while the race cars were 0.7%. Renting locally is cheaper and cleaner, but finding, contacting and negotiating with new suppliers every week is so much work that teams ship everything instead.

What it does

seeklocal. is an AI agent that handles local sourcing for touring events. The organizer uploads the equipment list they normally ship, and the agent:

1. Decides which items must be shipped (performance-critical, certified or branded) and which can be rented locally
2. Searches the web for real suppliers in the host city
3. Drafts a quote request for each supplier
4. Negotiates within a budget the organizer sets
5. Shows an impact dashboard: freight removed, CO₂ avoided, bookings and fees

The organizer's only job is to approve each booking.

For one F1 team over a season, we estimate this avoids 258 tonnes of CO₂, saves over $1M in freight and coordination costs, and moves $3.6M of spending to businesses in host cities. Across all 11 teams, that is about 2,840 tonnes of CO₂ a year.

Business model: B2B, pay per booking. $100 per supplier deal, no subscription, and no charge if no deal is made. Suppliers join free, so the agent always negotiates for the event.

How it solves the problem

Organizers aren't ignoring emissions. The low-carbon option just costs them more effort than the default. seeklocal. removes that effort, so choosing the cleaner option needs no extra motivation.

• Reduces emissions: Every item rented locally is freight that never gets shipped, and the biggest cuts come from replacing air freight and duplicate equipment sets.
• Benefits host cities instead of draining them: Touring events usually bring their own supplies and leave little behind. Here, spending goes to local suppliers, so the city gains income and a reason to welcome the event.
• No change in behavior: Organizers upload the list they already have and approve bookings. No audits and no new process.
• The event stays the same: Only commodity items move to local rental. Critical and branded gear still ships, so nothing the team or audience sees gets worse.

How we built it

• Research: F1 and DHL logistics reporting, the Tyndall Centre's live music touring roadmap, UFI exhibition industry statistics, destination management company pricing, and published freight emission factors
• Impact model: CO₂ avoided = tonnes removed × distance × emission factor (air 1,054 g CO₂ per tonne-km vs sea 19), plus freight cost and coordination time saved. We ran it twice with independent cross-checks and compared our baseline against published per-team estimates.
• Prototype: built on Replit with [Anthropic / OpenAI]'s API and its web search tool. The demo uses an F1 team's hospitality equipment list for the Las Vegas Grand Prix. Classification, live supplier search and outreach drafting are working; negotiation is simulated.

Challenges we ran into

• Narrowing a very broad prompt. We explored waste recovery, self-cooling structures and full event-planning platforms before focusing on the one decision organizers control directly: what they buy and ship.
• Getting the carbon story right. Sea freight is efficient per tonne-kilometre, so "shipping is bad" was too simple. We focused our claims on removing air freight and reducing the duplicate equipment sets teams keep.
• Checking our own numbers. Our first model run had a unit error that overstated carbon savings; cross-checking against F1's published totals caught it.
• Keeping the demo responsible. We chose not to send real quote requests to real Las Vegas businesses for an event we aren't running, so negotiation is simulated until a live pilot.
• Pricing. We moved from savings-based pricing, which is hard to verify, to a simple per-booking fee that is easy to count and doesn't lock customers in.

What we learned

The biggest barrier to low-carbon event operations is often the coordination work that makes the low-carbon option harder than the default. An agent that removes that work makes local sourcing the easy choice, and the carbon saving comes with a cost saving.

What's next

• A live pilot in Las Vegas with 3 customers and 20 local suppliers, measuring bookings, supplier response rate, staff hours saved and CO₂ avoided
• Real email outreach and reply reading, with a person approving every booking
• Interviews with logistics managers to test our main assumption that about 10% of shipped equipment can be rented locally

All savings figures are modelled estimates.

Built With

• anthropic
• css3
• html5
• node.js
• react
• replit
• typescript
• web-search