Regular updates on the latest VC-backed AI startups. Follow along to stay informed!
Inworld, a generative AI platform for creating NPCs, raised $30M in venture capital at a $500M valuation. Lightspeed Venture Partners led the round, with Stanford University, Samsung Next, Microsoft’s M12 fund, and Eric Schmidt’s First Spark Ventures participating.
Problem to be Solved
Scripted NPC interactions in video games get old fast. Inworld makes gaming more fun by using AI to give NPCs dynamic dialogue.
How They Use AI
Inworld uses multiple transformer-based LLMs to generate language. What’s interesting is how they go beyond basic LLMs to add configurable safety, knowledge, memory, narrative controls, multimodality, and more. Inworld claims the company’s NPCs can “navigate chats with memory and recall,” which makes us believe Inworld connects foundation models to external data sources, perhaps through a vector search tool like Pinecone. Inworld also offers a tool called 4th Wall that attempts to preserve immersion by preventing NPCs from talking about locations, people, social constructs, professions, and periods outside a game’s lore. This could be done using RLHF, similar to how OpenAI prevents ChatGPT from responding to certain questions (the dreaded “as an AI language model, I cannot…”).
Business Model
Inworld sells access to its platform on a subscription basis to creators, gaming studios, publishers, and other businesses. The company has worked with brands including NetEase Games, Niantic, LG Uplus, and Alpine Electronics to build AI-driven NPC experiences. It was part of the 2022 Disney Accelerator and has a grant from Epic to integrate its platform with Epic’s Unreal Engine. To enable more hobbyists to use its tech - and, presumably, convert some of those hobbyists into paying customers - Inworld plans to launch an open-source version of its character creation tool. Add the fact that Inworld’s founders sold their last startup to Google, where it became Google’s flagship conversational AI design platform, and you can see what has VCs excited.
Stay Ai, a customer loyalty operating system for Shopify brands, raised $15.1M in Series A funding. Telescope Partners led the round, with participation from Boulder Food Group, Riverpark Ventures, and Vanterra Capital.
Problem to be Solved
DTC e-commerce brands always try to increase growth by attracting and retaining customers. There is a whole Shopify App Store full of subscription tools that promise to increase AOV and LTV while reducing churn. Stay Ai’s founder was familiar with these tools from previous work building retention programs for 100+ brands as Managing Director of the retention division at Lunar Solar Group, and built Stay Ai “to solve the limitations in legacy subscription tools.”
How They Use AI
Stay Ai’s system leverages predictive analytics and machine learning models to enable merchants to rapidly deploy A/B tests across their marketing mix and optimize business outcomes. Honestly, this is more of a data science application than AI. The core of the company’s product seems to be causal inference with randomized control trials. The name Stay Ai is a recent rebrand from “dba Retextion,” which makes us think the company is cashing in on the AI hype without having any technical secret sauce.
Business Model
Shopify brands pay Stay Ai $499 per month plus 1% + $.19 per transaction. The company’s website says it is “Trusted by 350+ Brands Managing Millions of Subscribers”, and includes case studies from popular brands like Olipop.
Akkio, a generative analytics platform for data analysts and business operators, raised $15 million in Series A funding. Bain Capital Ventures and Pandome led the round.
Problem to be Solved
AI-powered BI tools that enable less data-savvy operators to create real-time insights are like flying cars - long promised in science fiction but still absent in real life today. Akkio aims to be the go-to generative AI solution for data analysts (like ChatGPT for copywriters, Dall-e for designers, or CoPilot for software engineers) with consumer-level usability.
How They Use AI
The core chat-with-your-data BI feature is built on a third-party API like ChatGPT or Cohere. Other features enable users to build and deploy predictive models with no code, an interesting example of “use AI to speed up your creation of AI.” In general, we at AI Deal Watch are bullish on AI dev tools that can be used by customers to train smaller custom models. It is unlikely they are training any large in-house models (they raised $15M, not $150M).
Business Model
Classic SaaS. With free trials and a “Starter” product tier starting at $49 per month, Akkio appears to be attempting product-led growth.