Why we're building Wendir

Quick answer: Wendir was built after a four-person trip where one of us was up at 6am cross-referencing receipts in three currencies and the group chat had 46 unread messages. The root problem isn't trip planning — it's that everyone uses one primitive (a group chat) for two different jobs (decisions and feelings). The fix is a workspace alongside the chat, run by seven specialised AI agents with citations. iOS-first, closed beta, built in Australia by Securight.

This is a short note about why we're building Wendir. If you came here for product features, the landing page is more useful. If you want the actual reason — the trip — read on.

The trip

Four of us. A city we'd all wanted to visit for years. Eight months of planning that should have been three weeks. Forty-six unread messages in the planning group chat the morning we flew.

By Day 2 of the trip, one of us was up at 6am cross-referencing receipts in three currencies on a phone in a hostel kitchen. By Day 4, two of us were quietly fuming about a dinner one of the others had vetoed without explaining why. By the time we got home, the Splitwise had A$340 in unresolved entries and stayed that way for six weeks.

The trip itself was great. The planning broke us.

We sat down a few weeks later and tried to figure out what had actually happened. The conclusion wasn't "we picked the wrong tools." It was that we'd run the entire trip with the same primitive everyone uses: a chat that is also the planning workspace. And a chat is the wrong primitive for the planning part.

The pattern

The more we looked, the more we noticed it.

Every group trip we'd ever been on had ended the same way. One person exhausted. One Splitwise abandoned. One chat full of "yeah I'll send it tonight" that never arrived. The names changed; the pattern didn't.

The friends who organised trips well were doing the same thing — they were just carrying more of it themselves. They were the unpaid project managers, fronting the money, owning the spreadsheet, chasing the votes. They were great hosts. They were also the most tired person on the trip.

That's the problem we're trying to solve. Not "trip planning" in the abstract. The specific thing where the planning works only because one person eats the cost.

Why a workspace and not a better chat

You could solve part of this with a better group chat — threading, polls, structured pinning. But the underlying issue is the merge of two different surfaces:

• The chat is high-frequency, ephemeral, social. It's for jokes and presence and "I just landed."
• The workspace is low-frequency, durable, structural. It's for dates and votes and bookings and expenses.

When you collapse them, the planning gets scrolled past. The vote gets relitigated. The expense never gets logged because the message about it is two hundred messages deep.

The fix isn't a better chat. The fix is a second surface — a workspace alongside the chat. That's what Wendir is. Detail in the planning manual and the chat-tools piece.

Why specialised agents and not a chatbot

We could have built a slick chatbot front-end on top of one of the frontier LLMs. It would have shipped faster. It would have looked impressive on first contact.

It would also have failed in all the same ways every other "AI trip planner" has failed — hallucinated venues, no expense math, no multi-user model, no persistence. Our AI reality check series is the long-form argument; the short version is that a chatbot is the wrong primitive for a multi-stakeholder workflow.

So Wendir has seven specialised agents — Scout, Local, Moneybags, Booker, Concierge, Reshuffler, Memorykeeper — each doing one job, with citations and confidence levels on every output. It's a workflow that happens to use LLMs for the parts they're actually good at, and uses other tools for everything else.

That architectural decision is the reason we expect Wendir to do what the wrappers can't.

Why iOS first

Pragmatic, not ideological. iOS first because:

• We can ship a higher-quality first version on one platform than a mediocre version on three.
• The closed-beta audience we're targeting (Australia, US, UK travellers, group-trip-organisers) skews iPhone.
• The interactions that matter most (camera for receipts, location for in-trip moments, Apple Pay/PayID for settlement) are higher-fidelity on iOS in 2026.

Android and a full web app come later. The longer reasoning is in Why iOS first when Splitwise scaled on Android.

What we're not

A few negatives so we're honest:

• Not free. Wendir Plus is A$49/year for the group — covers every member of every active trip. Free tier exists with limited AI generations. The economics work because it's a group seat, not a per-user subscription.
• Not a booking site. We don't earn commissions on bookings. The booker agent helps prepare booking intents; the user confirms with one tap. Same booking, but you're not the product.
• Not a chatbot wrapped in marketing. Real verification, real expense math, real multi-user state.
• Not done. We're in closed beta. There are bugs. There are agents (Booker, Reshuffler, Memorykeeper) that are in earlier stages than the live ones. We say so.

What's next

Closed-beta cohorts open every few weeks. Australian release first, then US/UK/SG/NZ. Android sometime in 2027. A full web app after that.

If the system works — really works, not just demos well — we keep building. If it doesn't, we'll tell you that too, and pivot or wind down. We'd rather be honest about both than sell a third-rate "AI trip planner" that lets one person remain the unpaid PM.

Waitlist, if you want an invite when the next cohort opens. No spam. We'll email you when there's actually space.

More from the Founder Dispatches

• Why iOS first when Splitwise made $30M on Android — the platform-strategy reasoning.

And the manual the trip in this post should have run on:

• How to plan a group trip without becoming the unpaid PM
• The 4 roles every group trip needs
• The 80% consensus rule
• How to split travel expenses

---

Written by the Wendir team. Last updated: 15 May 2026.