Gradium raised $70 million in seed funding. FirstMark Capital and Eurazeo co-led it. Participants include Xavier Niel, Eric Schmidt, and DST Global Partners. The Paris startup spun out of Kyutai lab. It started in September 2025.

The backers support new audio language models for voice AI. Founders from DeepMind and Meta fix latency and accuracy issues. The tech unifies generation, transcription, and talk in one system. It covers English, French, German, Spanish, and Portuguese.

Funds help commercialize via APIs as a voice layer for other AI. It fits healthcare, support, gaming, and ads. Revenue started six weeks after launch.

Neil Zeghidour, CEO and ex-DeepMind researcher, said current voice AI feels brutal. They reengineer for seamless talks. The team includes CTO Olivier Teboul, CCO Laurent Mazaré, and CSO Alexandre Défossez, all from DeepMind or Meta.

Gradium develops audio AI for real-time voice. Based in Paris, France, it offers models with low latency and natural flow for developers worldwide. The B2B APIs wrap around text or vision AI for sectors like gaming and customer service.

News from TechCrunch.

Cloneable, a startup that uses AI to shadow human experts in heavy industries such as energy and replicate their specialized workflows into autonomous agents, has raised $4.6 million in seed funding. Congruent Ventures led the raise, which included participation from First In, Overline, Bull City Venture Partners, and St. Elmo Venture Capital.

The idea for Cloneable traces back to a bottleneck its founders encountered years earlier while working in the field. The startup's founders realized that heavy industries face a “knowledge crisis” as experienced workers retire faster than they can be replaced.

Cloneable aims to capture and preserve that kind of institutional knowledge. The funding will also support expansion into infrastructure-heavy industries such as public utilities, vegetation management, construction, rail, mining, agriculture, and manufacturing.

The company claims that a process that typically takes a human engineer eight hours can be completed by a Cloneable agent in under two minutes.

In a move to revolutionize industrial maintenance, Edmund, a Czech startup, has landed €2.5 million in funding. This investment, led by FORWARD.one and backed by University2Ventures and T, underscores the potential of AI in transforming the factory floor.

The funding will likely be used to further develop Edmund's AI-powered debugging platform, designed to streamline troubleshooting processes in industrial settings. By leveraging AI, Edmund's platform could significantly reduce downtime and increase overall efficiency.

Edmund's approach to industrial maintenance highlights a growing trend towards adopting AI-driven solutions in manufacturing. As factories become increasingly complex, the need for intelligent troubleshooting tools has never been more pressing.

FORWARD.one's involvement in the funding round suggests a strong belief in Edmund's technology and its potential impact on the industry. With the participation of University2Ventures and T, Edmund is well-positioned to capitalize on its innovative approach.

The application of AI in industrial maintenance is still a relatively new field, but one that holds immense promise. Edmund's success in securing funding is a testament to the viability of its solution and the interest it has generated among investors.

As Edmund moves forward with its plans, the focus will be on how effectively it can implement its AI-driven troubleshooting platform across various industrial settings. The potential for growth and the impact on the manufacturing sector are significant, making Edmund a startup to watch in the coming years.

With its innovative approach and newfound funding, Edmund is poised to make a meaningful difference in the world of industrial maintenance. The use of AI to enhance troubleshooting capabilities is a step towards a more efficient and productive factory floor, and Edmund is at the forefront of this change.

Ringg AI has closed a USD 5.5 million Series A round. Arkam Ventures led the investment. Groww Founder Fund, Kunal Shah, White Venture Capital, and existing backer Capital2B also took part.

The startup will use the money to add people to engineering and go-to-market teams. It plans to create new products and grow its presence outside India. Part of the capital will go toward building its own AI models and internal GPU setup. This will allow on-premise options for large enterprises that need strict compliance.

Siddharth Shankar Tripathi, the founder and CEO, said voice is the best way to express thoughts. The platform lets companies handle operations through conversations, without code, and at a scale old systems can’t reach. The new funds will help make the product stronger, develop in-house models, and support on-premise setups for more control. In the long run, the team wants to handle 1 billion business conversations.

Rahul Chandra, managing director at Arkam Ventures, said AI is changing how enterprises use voice across their work. Voice is easy to adopt and opens up much more automation than before. Ringg AI already has the most live enterprise cases among similar companies. It should reshape how workflows get built.

The company processes around 1.5 million voice conversations a month. About three-quarters get resolved without any human help.

Ringg AI is a voice AI startup based in Bengaluru. It runs a no-code platform that lets enterprises build and run AI voice agents for customer tasks. The agents work in areas like support, lead checks, collections, scheduling, and screening, across multiple languages.

San Francisco-based Higgsfield announced an $80 million Series A extension, bringing its total Series A funding to over $130 million. The round included participation from Accel, AI Capital Partners, Menlo Ventures, and other investors, valuing the company above $1.3 billion.

The funding follows rapid growth for Higgsfield. Since launching in April 2025, the platform has gained over 15 million users and now generates 4.5 million videos per day. Its content has reached more than 3 billion social media impressions, making it one of the top GenAI platforms for reach.

Most of Higgsfield’s usage comes from social media marketers, with 80% of that segment producing commercial content. The platform is increasingly used as a full production tool, letting teams handle ideation, storyboarding, animation, editing, and publishing all in one system. Some customers are already spending over $200,000 annually on the platform.

CEO Alex Mashrabov said traditional video production can’t keep up with modern marketing needs. He explained that Higgsfield lets creators iterate quickly, giving smaller teams or individuals a chance to outperform large studios online.

Board member Jeff Herbst noted the platform is moving from pilot projects to daily production use across enterprises, signaling real adoption. Antoine Blondeau of Alpha Intelligence Capital said the company has turned speed into a major advantage, letting teams build, ship, and iterate in days instead of weeks.

Higgsfield plans to use the new funding to expand internationally, grow its enterprise offerings, enhance AI models for commercial video, and continue R&D. The company is also improving its API and marketing automation tools for high-volume video production workflows.

Source: Higgsfield ai

Singapore-based GrowthPal has raised USD 2.6 million in funding to scale its AI-powered platform for mergers and acquisitions deal sourcing. The round was led by Ideaspring Capital, with participation from a group of global angel investors.

GrowthPal is focused on helping corporate development teams find acquisition targets beyond traditional networks. Its platform uses AI to surface relevant opportunities faster, including off-market companies that are not actively seeking buyers. The goal is to help teams move from strategy to execution with more clarity and less reliance on personal relationships.

The company said the fresh capital will be used to build out the product and support expansion across the United States and other international markets. GrowthPal is seeing rising demand from companies looking for more structured and repeatable ways to approach M&A.

Founded by Maneesh Bhandari, Shalu Mitruka, and Amaresh Shirsat, GrowthPal builds an AI-driven M&A copilot for corporate development teams. The platform supports enterprises, startups, and private equity-backed firms across sectors such as IT services, SaaS, fintech, and vertical software. The company operates from Singapore and serves customers globally.

Deepgram, a voice artificial intelligence company based in San Francisco, said it has raised USD 130 million in a Series C funding round valuing it at USD 1.3 billion. The round was led by AVP, a global investment platform focused on fast-growing tech businesses. All major existing backers joined the round, and several new partners also took part in the investment.

Deepgram’s API platform lets developers and enterprises build live voice functions that can hear, speak and understand speech with low delay and strong accuracy. The company says more than 1,300 organisations already use its voice AI tools.

The startup also announced the purchase of OfOne, a voice AI platform originally built for restaurants and drive-thru services. The OfOne team will now help Deepgram build specialised voice AI for hospitality and food service.

Deepgram plans to use the funds to grow its products, expand into new markets, and support partners building voice AI features. As part of its push, it will open a new Voice AI Collaboration Hub in San Francisco.

Executives said the round shows strong belief in voice as a way for people to talk with machines, and will help Deepgram build tools that work more like the human voice.