Introducing TruTxt: Building Utility In The Unloved Parts Of The Open Internet
As FPC pens this edition’s missive, the news has broken that Anthropic has raised a Series H (yes, the 8th letter of the alphabet) round of $65B, giving it a post-money valuation of $965 billion. AI may well be the second coming, but you can’t help feeling the arse is going to fall out of this market at some stage. History never repeats; it rhymes.
A combination of corporate pullback and skyrocketing computational cost will no doubt take a rusty hacksaw to darling standalone Frontier LLM companies like Anthropic and OpenAI. FPC is a big believer in the tech, but open-source models will win out in the end.
Do you really want to get trapped by another Silicon Valley behemoth? Absolutely not. That’s why most corporations built on Linux during the OS wars. It’ll be no different here.
AI Can Build Utility For Publishers
It is strange to introduce a piece about “publisher utility” with an AI narrative when these frontier models were built - for free - off the blood, sweat and tears of content producers. No thanks. No compensation. Just fodder for the hyperscalers.
Despite all the negativity, though, AI is starting to present some interesting applications for publishers. FirstPartyCapital has been helping to incubate a new startup that addresses an overlooked part of the publisher inventory stack - with AI as a core capability.
TruTxt launched itself out of FPS this week. Led by Nick Fletcher and Dave Barr, the company aims to turn “distressed” international inventory into higher-yielding premium supply.
It is a big problem both Dave and Nick have seen throughout their long and illustrious careers - and have been busily building a solution over the past six months to help publishers better monetise international traffic.
Ad Tech Alchemy: Turning “Distressed” Into Something Of Value
Premium publishers have long struggled with how to properly monetise international traffic. How best to boost yield on non-native speakers who engage with content? AdSense and house ads are typically the go-to, paying pennies on the dollar/pound/euro/yen/rupee. But that is a terrible waste of attention.
There are publishers in the UK, for instance, with huge traffic in Africa, mostly from their always-on coverage of the Premiership. The traffic often spikes when an African footballer of note has played well - a Salah hat-trick or a Semenyo back-heeled goal. The problem for these publishers is that non-native speakers do not translate into meaningful revenue.
TruTxt is a solution that does real-time translation of publisher content across multiple languages to ensure better pricing. How does it work?
TruTxt has three key components:
Scaled translation: Post-ad call, Google Translate can translate most text at a functional level, allowing users to read the content. The issue here is that it doesn’t solve the problem of localising this user for buyers and plugging the signal into the programmatic demand stack. Publishers are making next to nothing in this instance. Now, publishers could build something functional of their own, but they would run into computational problems that make the process insanely expensive. That is where open-source models hosted on O&O servers play a big role in reducing costs. Tough to do. Even harder at scale. TruTxt has built this capability.
The signal: One of the big factors weighing on international pricing is the availability of anonymised geo data that can be shared with buyers. Location data is a key part of enriching the international impression. The TruTxt solution plugs this into buyers, maximising yield.
The demand stack: Having sales people in the market - from Sau Paiulo to Lagos to Hanoi - selling premium publisher supply (IO, curation and enriching open market inventory) to buyers is a critical piece of the TruTxt proposition. The objective is to build a comprehensive demand stack built around existing programmatic infrastructure, working with globally scaled partners like Index Exchange, on web and app.
Boring, Boring Ad Tech
It’s worth stressing again that boring is beautiful. TruTxt is a new model that does one boring thing really well. This will be a trend from FirstPartyCapital for the rest of the decade.
What can we build that addresses a real problem instead of chasing the latest trend (AI algos, CTV SMB, etc)? What can we build that isn’t necessarily a fast follower of US ad tech solutions developed specifically for the US market? It doesn’t always translate (excuse the pun) to the ROW.
This type of business is not something US ad tech will build anyway because it’s not a priority. In an insanely fragmented non-US market where we operate, TruTxt becomes a critical piece of infrastructure for the world’s biggest publishers.
If you are interested in partnering with TruTxt, reach out to Nick & Dave. We are doing a small seed round right now to help with dev hires and the global rollout. FirstPartyCapital has already invested and will be committing more in the coming months.
Boring, boring ad tech indeed. Have a great ad tech day.