In the era of artificial intelligence (AI), enterprises are eager to leverage large language models (LLMs) to optimize critical internal functions. Despite significant investments, achieving a substantial return on investment (ROI) remains a challenge. Today, New York-based startup Hebbia, which focuses on simplifying information retrieval through AI, announced it has secured $130 million in Series B funding from notable investors including Andreessen Horowitz, Index Ventures, Peter Thiel, and the venture capital arm of Google.
Hebbia is developing a straightforward yet powerful LLM-native productivity interface that streamlines the extraction of value from data, regardless of its type or size. The company is already collaborating with major financial services firms, including hedge funds and investment banks, and plans to extend its technology to more enterprises soon.
“AI is undoubtedly the most important technology of our lives. But technology doesn’t drive revolutions– products do. Hebbia is building the human layer – the product layer – to AI,” George Sivulka, the founder and CEO of Hebbia, stated in a blog post. Prior to this funding round, the company raised $31 million through several rounds.
While LLM-based chatbots can utilize internal documentation or be prompted with documents, they often fail to answer complex questions about business functions accurately. This can be due to limitations in the context window, which may not handle the size of the document provided, or the sheer complexity of the query itself. Such errors can erode teams’ confidence in the capabilities of language models.
Hebbia, founded in 2020, addresses these challenges with its LLM-linked agentic copilot called Matrix. This tool operates within the business environment of companies, enabling knowledge workers to pose intricate questions related to internal documents—ranging from PDFs and spreadsheets to audio transcripts—with an infinite context window.
The Matrix platform allows users to input queries and associated documents/files. The system then decomposes the prompt into smaller, manageable actions that the underlying LLM can execute. This process enables the platform to analyze all the information contained in the documents simultaneously and extract the necessary data in a structured format. Hebbia claims that its platform can process and reason over any volume (from millions to billions of documents) and modality of data, while also providing relevant citations to help users trace each action and understand the platform’s decision-making process.
“Designed for the knowledge worker, Hebbia lets you instruct AI agents to complete tasks exactly the way you do them – no task too complex, no dataset too large, and with the full flexibility and transparency of a spreadsheet (or a human analyst!),” Sivulka elaborated in the blog post.
Feature | Description |
---|---|
Infinite Context Window | Handles extensive data volumes, ideal for enterprises with vast documentation. |
Comprehensive Data Analysis | Processes various data formats simultaneously. |
Actionable Intelligence | Breaks down complex queries into manageable tasks for accurate results. |
Transparency and Traceability | Provides citations for every action, ensuring decision-making transparency. |
Versatility | Adapts to various industries, including financial services and beyond. |
Initially, Sivulka created the platform to streamline the workload of financial industry workers who spent substantial time sifting through documents for relevant information. Over the years, Hebbia has expanded its reach, gaining traction in other sectors as well. Currently, the company boasts over 1,000 use cases in production with several major enterprises, including CharlesBank, American Industrial Partners, Oak Hill Advisors, Center View Partners, Fisher Phillips, and the U.S. Air Force.
“Over the last 18 months, we grew revenue 15X, quintupled headcount, drove over 2% of OpenAI’s daily volume, and laid the groundwork for customers to redefine how they work,” Sivulka noted. It remains uncertain whether OpenAI is the sole model used within the Matrix platform or if users have the option to select other LLMs.
With the recent funding, Hebbia aims to build on its successes and attract more large enterprises to its platform, simplifying how their workers retrieve knowledge. “I’m excited for a world of unbound progress– one where AI agents contribute more to global GDP than every human employee. I believe that Hebbia is going to get us there,” Sivulka added, highlighting that the company is developing what he considers the most important software product of the next 100 years.
Despite its innovative approach, Hebbia is not alone in the AI-based knowledge retrieval space. Other companies are also exploring similar technologies. For instance, Glean, a Palo Alto-based startup, achieved unicorn status in 2022 and has developed a ChatGPT-like assistant specifically for workplace productivity. Additionally, Vectara is working on enabling generative AI experiences grounded in enterprise data.
Hebbia’s recent funding round and its expanding list of enterprise clients underscore its potential to significantly impact the field of AI-driven knowledge retrieval. As the company continues to innovate and enhance its platform, it is well-positioned to lead the market and drive substantial improvements in how businesses manage and utilize their data.
With its focus on creating practical, product-driven AI solutions, Hebbia is poised to help enterprises navigate the complexities of data management and retrieval, ultimately unlocking new levels of efficiency and productivity.
Synthflow, a Berlin-based startup, has announced a $7.4 million seed round to further develop its no-code AI voice assistance platform designed for small and medium-sized enterprises (SMEs). The company, which focuses on automating repetitive tasks for busy business owners, has raised a total of $9.1 million since its inception around spring last year, highlighting the increasing investor interest in generative AI applications.
Investor Confidence and Customer Growth
Synthflow has made significant strides since its launch, approaching 1,000 customers and boasting “double-digit” monthly growth rates since unveiling its browser-based no-code tool in December 2023. This rapid adoption suggests a strong demand among SMEs for generative AI tools that can enhance productivity through automation.
The latest funding will be dedicated to research and development, with CEO and co-founder Hakob Astabatsyan emphasizing the importance of maintaining the startup’s early momentum. “We have very many ideas. We know exactly what the customers need,” Astabatsyan told TechCrunch, indicating a clear vision for the future of AI in the SME sector.
The Founding Team
Astabatsyan, an ex-Rocket Internet entrepreneur, co-founded Synthflow with his brother Albert and Sassun Mirzakhan-Saky. Albert brings experience from a previous no-code startup, while Mirzakhan-Saky contributes his software engineering expertise as CTO. Together, they aim to make AI technology accessible to non-technical users, particularly SMEs.
Multi-Language Capabilities
Initially, Synthflow’s product focused on English-language call handling, catering to its largest markets. However, the startup has since introduced beta versions in German and French, signaling an expansion into European markets.
End-to-End Experience for SMEs
Synthflow’s no-code platform targets service industry SMEs, offering an end-to-end experience that automates core tasks like appointment scheduling. This allows business owners, who might otherwise miss calls and potential business, to benefit from AI-driven efficiency. Astabatsyan explains, “The AI can do it in a more affordable manner, more reliably, and humans can do other stuff.”
Key Benefits of Synthflow’s AI Voice Assistance
Technology and Integration
Synthflow builds on OpenAI’s GPT language models, incorporating its own AI models fine-tuned to specific customer needs. The startup has developed a “voice orchestration layer” that converts speech to text, processes it with AI, and converts responses back to speech. This technology ensures that even non-technical users can design voice agents tailored to their business requirements.
Future Developments
Looking ahead, Synthflow plans to enhance its capabilities with features like “live actions” or “connections,” allowing AI to check live inventory or pull requested information during a call. The startup envisions a scenario where task-focused AI systems could collaborate, handing off calls to other specialized AI agents or human operators as needed.
Challenges and Opportunities
Astabatsyan acknowledges that while AI can increase productivity, it also raises questions about resource allocation. “If there’s so much capacity — and productivity gets unleashed — how do we channel this human resources in other sectors of the economy?” he pondered, highlighting a key challenge for managers and leaders.
Funding and Investor Participation
The $7.4 million seed round was led by Singular, with participation from existing investor Atlantic Labs and AI-focused investors, including the founders of Krisp AI. This robust financial backing underscores the confidence in Synthflow’s potential to transform SME operations through AI.
Summary of Synthflow’s Progress
Aspect | Details |
---|---|
Funding Raised | $9.1 million total, including $7.4 million seed round |
Founding Date | Around spring last year |
Customer Base | Approaching 1,000 customers |
Growth Rate | Double-digit monthly growth since December 2023 |
Core Focus | Automating repetitive tasks for SMEs |
Technological Foundation | OpenAI’s GPT, proprietary AI models |
Languages Supported | English (main), German and French (beta) |
Key Takeaways
Synthflow’s innovative approach to AI voice assistance positions it as a promising player in the SME sector, with substantial backing from investors and a clear roadmap for growth and development.
When Ironspring Ventures launched in 2020 to back startups in industrial sectors like construction and manufacturing, it was one of very few early-stage venture firms paying attention to these capital-intensive sectors. Now, the firm is doubling down on its initial commitment.
The Austin, Texas-based firm raised $100 million for its second fund, focusing exclusively on industrial startups. This marks a significant increase from the firm’s $61 million debut fund that closed in 2021. The latest raise has enabled Ironspring to expand its team by hiring its first principal, Colleen Konetzke, and a head of platform, Stephanie Volk. With Fund II, the firm plans to invest in 20 startups, supporting four to five companies annually.
“What we saw back then was as true as we see today,” Ironspring co-founder and general partner, Ty Findley, told TechCrunch. “There is a big gap in the venture industry that deeply studies and has genuine GP market fit with these industrial markets and can help them navigate a pretty challenging go-to-market process. When you really roll these industries up, they are over half of the U.S. GDP. My strong opinion is, we as a country simply cannot afford to let the U.S. get left behind.”
Findley refers to industries including manufacturing, construction, transportation, and energy. The firm backed 16 companies in its first fund, among them Solvento, a payments infrastructure startup for trucking companies in Mexico; OneRail, a last-mile logistics startup; and Prokeep, a communications platform for distributors.
Ironspring has already backed six companies with Fund II and has deployed about a quarter of the fund. Findley notes that the main difference between Fund I and Fund II is the additional capital, which allows the firm to write larger checks, ranging from $2 million to $4 million. This financial capability helps Ironspring stay competitive as seed rounds have grown larger.
Fund | Amount Raised | Number of Companies Backed | Check Size |
---|---|---|---|
Fund I | $61 million | 16 | Smaller initial checks |
Fund II | $100 million | 20 (planned) | $2 million to $4 million |
Findley expresses enthusiasm for having a fresh pool of capital to invest now, given the macroeconomic tailwinds affecting the industries they focus on. Supply chain constraints that began during COVID-19 persist, and new challenges have emerged due to conflicts in the Middle East. Policies like the Inflation Reduction Act and CHIPS and Science Act are also bringing attention and government funding to these sectors. Additionally, advancements in AI could significantly impact these industries.
“We are seeing more top-tier tech and innovation talent flood into these industries,” Findley said. “Whether they are recirculating from recent tech unicorns, or just other tech talent that simply wants to make a big impact on their career that’s not based on photo sharing or adtech or chasing the next crypto coin, that is what the macro trends are.”
GoodShip, a freight orchestration and procurement platform started by former operators at Convoy, exemplifies the type of company Ironspring is keen to support. Ironspring co-led the firm’s 2023 seed round alongside Chicago Ventures and re-upped at the Series A earlier this year.
While Ironspring was one of the first early-stage firms focused on industrial startups, the space has become more crowded as deep-pocketed firms like Andreessen Horowitz, General Catalyst, and Bessemer have entered. However, Findley views the entrance of these name-brand firms not as competition but as a positive development.
“I’m a believer that the more capital flowing into these industries, the better,” Findley said. “Those are great allies. We wouldn’t be able to do our job at the seed stage if we didn’t have great downstream growth.”
Findley emphasizes the collaborative nature required for these types of startups to grow successfully, appreciating the different perspectives other firms can bring to their portfolio companies. Ironspring even invites these other firms onto its podcast, Heavy Hitters, to create a valuable resource for their portfolio companies and beyond. Notable VCs such as Katherine Boyle, a general partner at a16z; Aaron Jacobson, a partner at NEA; and Lior Susan, the CEO and founder of Eclipse Ventures, have been featured.
Findley believes Ironspring stands out among the growing competition due to its sector expertise and unique LP base. The firm’s LP base includes operators in the industries they invest in, such as owners of construction companies and manufacturing plants. These LPs can provide guidance and advice to companies and serve as potential customers.
Being based in Austin is also a significant asset for Ironspring, according to Findley. Contrary to the perception that Austin is merely an emerging tech hub, he points out that many of the industries Ironspring focuses on have deep roots in the area. With Tesla moving its headquarters to Austin and the recent approval of $6.4 billion from the infrastructure act for Samsung to build semiconductor chips there, the city has the right talent to drive the digital industrial revolution.
Findley’s commitment to ensuring that critical U.S. industries are not left behind is unwavering. “The U.S. can’t allow these critical industries to be left behind,” he said. “We are here for the long haul in ensuring that will never happen.”
With a fresh pool of capital and a strategic focus on industries vital to the U.S. economy, Ironspring Ventures is poised to make a significant impact on the industrial startup landscape. The firm’s approach, combining sector expertise and collaborative investment strategies, ensures that it remains a key player in fostering innovation in these critical sectors.
In an inspiring twist to the usual post-graduation story, 18-year-old best friends Christopher Fitzgerald and Nicholas Van Landschoot are trading summer fun for hard work in the world of startups. The recent high school graduates from Boulder, Colorado, have secured a $500,000 pre-seed investment from Varana Capital to launch their innovative venture, APIGen.
High School Achievements:
While most of their peers are enjoying their last summer before the responsibilities of college or careers, Fitzgerald and Van Landschoot are immersed in the entrepreneurial grind. They have set up shop in a venture capital (VC) office in Boulder, focusing all their energy on APIGen, a startup that aims to simplify API creation through natural language processing.
Future Plans:
The concept for APIGen blossomed from the duo’s shared passion for coding and their early exposure to the challenges of API development. The two met on their school’s debate team and quickly bonded over their love for technology. Their first collaborative project was a chatbot designed to facilitate data interaction, but they soon realized the uniqueness of their idea was limited. This realization led them to a pivotal insight about the complexities and difficulties of designing APIs.
“We discovered that creating APIs was cumbersome and not straightforward,” said Fitzgerald. “It was during this project that we decided to focus on simplifying API generation.”
With a prototype in hand, they began showcasing their idea to local programmers and industry professionals in Boulder, a city known for its vibrant tech community.
Their perseverance paid off when they caught the attention of Philip Broenniman, the founder of Denver-based Varana Capital. Varana Capital, which has grown from a family office to a firm managing $400 million in assets under management (AUM), saw potential in the young founders’ vision. Initially planning to provide mentorship, Broenniman and his team were so impressed by the teens’ presentation that they decided to invest.
Investment Details:
Broenniman recalled, “We walked into the meeting intending to offer some advice, but their pitch was the best we had heard in years. Their insights were incredibly sharp for their age.”
APIGen’s mission is to develop a platform that allows users to create custom APIs using natural language prompts. This approach simplifies the complex process of API creation, making it accessible to users without extensive technical knowledge. The platform aims to serve various needs, from basic data connectors to intricate business logic integrations.
“We’re not just talking about simple data exchanges. Our goal is to generate code for APIs that incorporate complex functionalities and business logic,” explained Van Landschoot.
The founders envision a future where creating APIs is as simple as giving a verbal command. For instance, an e-commerce business could request an API to connect its online store to its inventory database, and APIGen would deliver a functional solution.
APIGen’s Capabilities:
Feature | Description |
---|---|
Natural Language Processing | Users can request APIs using everyday language. |
Complex API Generation | Builds APIs with integrated business logic. |
Multiple Task Handling | Capable of generating APIs for serial or multiple tasks. |
IoT Integration | Supports APIs for controlling IoT devices. |
Custom Functionality | Allows for highly tailored API solutions. |
Fitzgerald and Van Landschoot have quickly integrated into Boulder’s close-knit tech scene. They joined the local AI Meetup, which boasts over 1,400 members and received enthusiastic support from the community. Their early involvement and the positive reception of their product demos have been pivotal in building a network of supporters and potential users.
Despite facing stiff competition from established players like Salesforce’s MuleSoft and RapidAPI, the founders believe APIGen has a unique value proposition. Their approach to API creation, centered on natural language and tailored to specific user needs, sets them apart in the burgeoning API market.
Competition and Market Potential:
“We’re entering a competitive space, but we’re carving out our own niche,” said Fitzgerald. “The support we’ve received from the local community and the backing from Varana Capital give us confidence in our path forward.”
Although APIGen is still in its early stages, the founders are focused on refining their product. They are working tirelessly to develop a minimum viable product (MVP) and plan to release a beta version later this month. The feedback from this release will be crucial as they aim to attract their first set of users and demonstrate the platform’s capabilities.
Development Milestones:
Broenniman, who will take a seat on APIGen’s board, is optimistic about the startup’s prospects. He highlights the founders’ ability to build a community around their idea and their potential to make significant returns on investment.
“APIGen is more than just a startup; it’s a partnership with Christopher and Nicholas. They are tapping into a massive market and have already shown they can rally support,” he said. “The potential for returns is remarkable.”
As Fitzgerald and Van Landschoot navigate their journey from high school graduates to startup entrepreneurs, they remain focused and determined. Their story is a testament to the power of youthful ambition and the vibrant support networks available in tech-centric communities like Boulder.
With their first major funding secured and a clear vision for APIGen, the young founders are poised to make a significant impact on the API development landscape. Their summer may be spent in an office rather than on the beach, but they are building something that could change the way businesses interact with technology.
As the tech world watches, these two young innovators are proving that with a compelling idea and the right support, age is just a number in the realm of entrepreneurship.
Generative AI startup Stability AI, known for its innovative text-to-image technology, Stable Diffusion, has appointed Prem Akkaraju, former CEO of Weta Digital, as its new full-time CEO. This leadership change follows the departure of founder and former CEO Emad Mostaque amidst investor concerns about the company’s financial health.
Prem Akkaraju Steps In
Stability AI, the cutting-edge generative AI firm, is set to welcome Prem Akkaraju as its new CEO, according to reports from The Information. Akkaraju, who previously helmed visual effects powerhouse Weta Digital since January 2020, will take the reins from interim co-CEOs Shan Shan Wong (COO) and Christian Laforte (CTO). Akkaraju’s leadership background includes co-founding Screening Room, where he serves as executive chairman.
This leadership transition comes almost three months after Emad Mostaque, Stability AI’s founder, resigned in March 2024. The interim period saw Wong and Laforte guiding the company through a turbulent phase marked by investor skepticism about its financial sustainability and business strategy.
Interim Co-CEOs Shan Shan Wong and Christian Laforte’s Contributions
During the interim period, Wong and Laforte provided stability and continuity, focusing on sustaining the company’s operational momentum and technological advancements. Their stewardship ensured that Stability AI continued to innovate and expand its AI capabilities despite the leadership vacuum.
Concerns Over Financial Viability
Mostaque’s departure was fueled by investor concerns regarding Stability AI’s financial health. Reports indicated that the company had incurred losses exceeding $30 million in the first quarter of 2024, with revenues falling short of $5 million. This financial strain raised alarms about the company’s long-term viability and prompted a search for new leadership and additional funding.
Stable Diffusion’s Market Impact
Despite these challenges, Stability AI has maintained a strong presence in the AI market, particularly with its flagship product, Stable Diffusion. This text-to-image generative AI technology has continued to evolve, with the latest iteration, Stable Diffusion 3 Medium, launched on June 12, 2024. This version promises faster performance and enhanced capabilities, reinforcing Stability AI’s commitment to pushing the boundaries of AI innovation.
Investment and Funding Prospects
Amidst the leadership changes, Stability AI is also reportedly on the cusp of receiving a significant financial boost. The Information claims that an investor group led by tech entrepreneur Sean Parker is preparing to inject new capital into the company. This investment could provide the much-needed financial stability and fuel further growth and development.
Stability AI’s portfolio extends beyond image generation, showcasing a broad array of AI models and tools across various domains. This diversification strategy is evident in their offerings in code generation, text language models, and multimedia applications.
Stable Code and Stable LM
Stability AI has ventured into code generation with its Stable Code model, which saw a significant release at the beginning of 2024. This model aims to streamline and enhance software development processes by generating high-quality code based on textual descriptions.
In the realm of language processing, Stability AI’s Stable LM offers a range of models with different parameter sizes, including 1.6 billion, 3 billion, and 7 billion parameters. These models cater to diverse applications, from natural language understanding to complex text generation tasks.
Audio and Video Innovations
Stability AI’s foray into audio and video generation is marked by the release of Stable Audio 2.0 on April 3, 2024. This advanced model allows users to create up to three minutes of audio content, opening new avenues for creative and commercial applications. Additionally, Stable Video provides robust text-to-video generation capabilities, including the ability to produce 3D content through Stable Video 3D (SV3D).
Membership and API Access
To support its growing suite of AI models, Stability AI introduced a new business model in December 2023. This model includes a membership system where organizations wishing to use Stability AI’s technologies for commercial purposes must subscribe. This approach not only generates revenue but also ensures that commercial users contribute to the sustainability of the company’s operations.
Beyond the membership model, Stability AI offers access to its AI models via an API, allowing organizations to pay based on usage. This flexible access model caters to a wide range of users, from individual developers to large enterprises. The Stable Artisan discord bot, launched on May 9, 2024, provides a user-friendly interface for accessing these powerful AI tools.
Product | Description | Release Date |
---|---|---|
Stable Diffusion | Text-to-image generative AI technology | June 12, 2024 (v3) |
Stable Code | AI model for code generation | January 2024 |
Stable LM | Large language model with multiple parameter sizes | Various (1.6B to 7B) |
Stable Audio 2.0 | Audio generation model allowing up to 3 minutes of content | April 3, 2024 |
Stable Video 3D | Text-to-video generation, including 3D capabilities | Ongoing |
Membership Model | Commercial usage subscription model | December 2023 |
API Access | Usage-based access to AI models | Continuous |
Stable Artisan Bot | Discord bot for AI model access | May 9, 2024 |
Competitive Landscape
Stability AI faces significant competition in the rapidly evolving AI market. In the image generation space, competitors like OpenAI’s DALL-E 3, Midjourney, and Ideogram are constantly innovating, intensifying the market dynamics. However, Stability AI’s diverse portfolio and continued advancements position it well to navigate these challenges and capture market opportunities.
Strategic Diversification
The company’s expansion into various AI domains, including code generation and multimedia content creation, demonstrates its strategic intent to diversify its offerings. This approach not only mitigates the risks associated with relying on a single product line but also opens up new revenue streams and market segments.
Mostaque’s New Initiative
Even as Stability AI moves forward with new leadership, Emad Mostaque, the company’s founder and former CEO, remains active in the AI community. Mostaque has announced his involvement in a new venture, Schelling AI, which aims to advance the concept of decentralized AI.
While details about Schelling AI are sparse, Mostaque has hinted at a vision that promotes broader access to AI technologies and supports decentralized innovation. More information about this initiative is expected to be unveiled in July 2024.
Anticipation for Schelling AI
Mostaque’s announcement has sparked interest and speculation within the tech community, as many await to see how his new venture will shape the future of AI development and deployment. His experience and insights from leading Stability AI are likely to influence the direction and goals of Schelling AI.
Conclusion: A New Chapter for Stability AI
As Stability AI embarks on this new chapter under Prem Akkaraju’s leadership, the company is poised to navigate its financial challenges and continue its trajectory of innovation in the AI landscape. With a diversified product lineup and a strategic business model, Stability AI remains a key player in the competitive field of generative AI.