Natural language processing—NLP for short—is our favorite type of AI. We research NLP methods to ensure that our conversational AI understands banking, with all its esoteric terminology and specific customer requests. It’s our idea of a good time.Learn More
We don’t just keep up with the technology; we take it to new places. Our research creates new possibilities for digital banking conversations, so your AI assistants continue to get smarter.
Our deep learning models are second to none. Our datasets are always growing, too. That to more powerful predictions that solve for more business needs.
Rule-based AI gets a bad rap. Sure, it’s a simpler approach to machine intelligence—but it’s fast and precise. Statistical AI is more exciting. It’s powerful, but takes a while to get up to speed. We use both, solving every problem with the right tool for the job.
Memory persistence helps conversational AI follow speakers as they leap from topic to topic and back again. Human speech isn’t always linear; Posh AI is okay with that.
We find new ways to turn information into insight for our team and our clients alike. Our client-facing analytics platform automates comprehension gap analysis, topic organization, and more—with powerful new features always on our radar.
"For our team, it isn’t just about creating successful conversational AI solutions, but also empowering our clients with the knowledge and tools to be AI-adept in an increasingly AI-centric world."
Our mission is set in stone: We’re here to make digital banking a nicer experience. But there are lots of ways to get there, and our research team explores them all. Here are just a few of our current tech obsessions:
Our systems answer many client-specific incoming requests without training models, by automatically collecting unlabeled data from databases, documents, and websites—and leveraging them with semantic search.
We’re leading the way in building language models, embedding resources and ontologies to understand the personal finance and banking domain.
Our analytics dashboard is fueled by unsupervised machine learning to point out areas for improvement, and offer recommendations on content and design.
Annotating the content of a human-AI conversation makes it easy to verify success—and figure out where we could do better.