Three New Speech Models & Smarter Chat Widgets

Product Updates July 2026

History, Everywhere You Need It

There is now one global /history page with surface tabs, an owner filter, and a type filter. Cards tell you who sent a run and what state it is in, so a shared workspace reads like a shared workspace.

Alongside it, each app gets a Usage tab listing that app's runs across the whole workspace, reachable straight from the view selector in the bar. Handy when you want to know how one agent is doing rather than everything at once.

Three New Speech Recognition Models

SraVaani 1.0 is now available, an open (MIT) 430M-parameter FastConformer model from IISc's SPIRE Lab and ARTPARK, deployed on Modal. It covers 65 Indian languages and dialects and identifies the spoken language automatically, so you do not need to tag it at inference time.

Google Chirp 3 replaces the deprecated USM, with its documented locales, auto-detection, and isolated recognizers.

GPT Transcribe also joined the list and is now the default ASR model. Chirp's transient polling failures now retry instead of failing your run.

Chat Widgets: Themes, Media Memory, and Cleaner Exports

Widgets can now use builder and WhatsApp themes, and the chat preview reskins in place instead of remounting, so switching themes stops feeling like a page reload.

Self-Hosting and Platform Independence

We are happy to share that Gooey.AI is working towards DPGA certification. Self-hosting and platform independence are now ready to use with a single terminal command. If you are interested in self-hosting a Gooey-server and want to know more, you can read all about it.

Gooey.AI Selected for the 100x Accelerator

Gooey has been selected for 100x Impact's 2026/27 Cohort at the The London School of Economics and Political Science (LSE) School of Economics, joining 11 other ventures (from 2000 applicants!) working to scale proven solutions to global challenges.

https://drive.google.com/file/d/1ItiVcILfttZBobOzgCz8nhLOdCiIo%5Fop/view?usp=sharing

AIJF Session with Rajdeep Sardesai

As part of the AI Journalism Fellowship (AIJF) 2026, fellows participated in an interactive session with veteran journalist Rajdeep Sardesai. Drawing on over three decades of reporting experience, he shared insights from his field assignments and discussed the evolving role of AI in journalism. He emphasised that AI cannot replace the human nuances, unpredictability, and lived experiences that are integral to journalism. The session enabled fellows to engage directly with his experiences through a thoughtful Q&A and discussion.

https://drive.google.com/file/d/15374YLfU8FEzwWMC5TkNXUGkyVWmcxul/view?usp=sharing

Reflection in Creative Experience Workshop 2026

Archana Prasad presented the paper Beyond Bias: Participatory and Reflective Approaches to Cultural AI at the first Reflection in Creative Experience (RiCE) Workshop at UAL London, held as part of the ACM Creativity and Cognition Conference 2026.

Co-authored under the supervision of Tom Simmons, with research assistance from Isha Singh, the paper explored creative AI as a tool for ethical reflection and AI literacy, collaborative dataset creation, and culturally grounded AI practices. The paper has been published at arXiv: https://arxiv.org/pdf/2609.05102