AI built for people.
Mountain View, CA
Joined August 2021
- Tweets570
- Following204
- Followers12.7K
- Likes1K
Pinned Tweet
Inworld is your new favorite inference provider.
We already process tens of trillions of tokens a month because we understand consumer AI different, and we've built our inference from the ground up for it.
Today, the inference top consumer AI apps run on is opening up.
Status, the AI social simulation game built by Wishroll, was growing fast. Its five-person team needed to bring AI costs down without compromising the experience players expected.
We partnered with them to make that growth sustainable.
Key results:
- 95% reduction in AI costs
- Scaled to 500,000+ daily active users
- 35x revenue growth to tens of millions in annual revenue
- Sub-second response times, even at peak demand
With Inworld’s support, the team could keep expanding the game’s audience while spending more engineering time on new features.
“Working with Inworld has been the most positive experience I've had with another company... They've been incredibly hands-on, responsive, and fast. It's like having an external ML team dedicated to helping us solve our problems.”
- Fai Nur, CEO and co-founder of Wishroll / Status
Read the full case study for a closer look at how we worked together.: inworld.ai/customers/status
Inworld AI retweeted
@thestatusai has scaled to tens of millions in annualized revenue. Working with @inworld, we cut our AI costs by about 95% while maintaining or improving content quality.
When you build consumer AI, you want people to be obsessed with your app. Then you have to afford that obsession.
Our users spend an average of over an hour and a half a day on Status. They’re building relationships with characters, creating stories, and getting absorbed in these worlds. Every interaction needs to feel worth coming back for.
Inworld’s team got into the details with us: which models to use, how to structure prompts, and how to test whether the content was actually funny and engaging. They’ve been hands-on and fast, and we’ve kept working together to improve performance and lower costs.
That gave our team more room to keep shipping. More features, more things for players to do, and a business that could support the experience we wanted to build.
There’s so much opportunity in consumer AI when you understand what people enjoy. Making the economics work gives you room to keep finding more of it.
Read more on how we scaled with Inworld: inworld.ai/customers/status
With @thestatusai, we worked on prompts, evaluations and serving to cut AI costs by over 95% while maintaining or improving content quality.
@latitudegamesai ran 25M+ inference calls in its first six weeks on Inworld.
We recently announced our acquisition of Ultravox. GeekWire covered the news, with some background on the company and its journey from Fixie to Ultravox.
If you missed the announcement, read the story here: tinyurl.com/UVInworld
We recently shared our perspective in USA TODAY on the infrastructure behind consumer AI.
The article explores what it takes to keep AI applications fast and affordable as people use them more often.
It covers how serving costs change as usage grows and why realtime voice requires coordination across multiple systems. There’s also a look at Inworld’s origins in interactive experiences and how that work shaped our focus on realtime AI for consumer-facing applications.
Read the full story: tinyurl.com/InworldUSAToday
In his blog post “WTF is Voice Steering?”, our Head of Research, Aleksey Tikhonov, explains how voice steering in TTS-2 lets you give a speech model instructions about how to deliver a line.
The piece focuses on how we measure whether those instructions actually work. Aleksey explains where automated audio judges fell short in our tests and how the team measures changes in the speech itself.
He also explores why instruction wording matters and why the same direction can produce different results across voices. These measurements help the team track changes between model versions, with limits on what they can tell us.
Read the full piece: tinyurl.com/wtfisvoicesteeri…
Inworld AI retweeted
In Voyage, the narrator and every character you meet have their own voice, generated live for whatever you do.
Our SOTA World Engine gives every line of dialogue a delivery direction (warmly, breathless, urgently) and @inworld TTS performs it live.
Here's what that unlocked:
4x faster voice at a quarter of the cost. Those savings helped us offer unlimited play to all our subscribers.
Our founder and CEO @nickwalton00 sat down with the Inworld team to talk about why a robotic voice breaks a story, and how we made every voice in Voyage feel real.
Join us in Berlin on Friday, October 23 for Backchannel Berlin, an afternoon of conversation about the research on speech-to-speech models.
Our team will share how we're building and serving full-duplex speech-to-speech in realtime, including the things that haven't worked. Then the floor opens: bring your own work or a problem you're stuck on, or just come and listen.
We'll be getting into turn-taking and interruptions, cascaded versus end-to-end, keeping an LLM's reasoning when you teach it to speak, cross-lingual transfer, and what we should actually be measuring for conversations.
If you work on speech-to-speech, TTS, ASR, audio codecs, conversational modelling, LLM serving or evaluation, we'd love to have you. PhD students welcome.
Food and drinks throughout. Register at luma.com/backchannel-berlin
We’re excited to announce that @ultravox_dot_ai is now part of Inworld.
Ultravox is the platform developers use to build real-time voice agents. We've worked with the team for a while through our TTS partnership, and today members of the team that built it are joining Inworld to keep developing it.
Together, we can improve the full conversation: understanding what a user says, helping them get something done, and responding in a voice that fits the moment. The first improvement is live today. Every built-in Inworld voice on Ultravox now runs on Realtime TTS-2, with existing voice IDs unchanged, no code changes required and no additional cost.
This is a big step in our work on speech-to-speech experiences that bring speech understanding, reasoning, and expressive voice generation closer together, and we're only getting started.
Full post on why we came together, what you get today, and what's next: tinyurl.com/inworldultravox
