AI & Startups October 21th - October 27th

AI Product News

?? Midjourney Launches New Image Editor

Midjourney has debuted a new AI-powered web editor, allowing users to modify, retexture, expand, and stylize both generated and uploaded images using text prompts.

The Details:

??? New Editing Features: The update introduces capabilities like expand, crop, repaint, and modify for both Midjourney-created and uploaded images, driven by natural language prompts.

??? Compatibility with Previous Features: The editor integrates seamlessly with existing tools such as personalization, style references, and more.

??? Re-Texturing Tool: This feature lets users adjust lighting, texture, and materials while preserving the image's original shape.

??? Limited Access: Initially, the editor is available to yearly subscribers, 12-month members, and users with 10,000+ generations for testing, before broader release.

?? Why It Matters: As AI-generated content becomes increasingly sophisticated, distinguishing between real and AI images will be even more challenging. While this new editor empowers creatives with incredible flexibility, it also opens the door to deepfake and manipulation concerns, making us question the authenticity of images more than ever before.

?? Google DeepMind Open-Sources AI Watermarking Tool

Google DeepMind has announced the open-sourcing of SynthID, an advanced watermarking system for AI-generated content. SynthID is already in use across Gemini and other Google products.

The Details:

??? Tournament Sampling: SynthID uses a ‘tournament sampling’ method to embed undetectable watermarks in AI-generated content without compromising quality.

??? Extensive Testing: SynthID was tested across 20M real Gemini interactions, showing no impact on response quality or user satisfaction.

??? Multi-Content Support: SynthID works across various formats, embedding watermarks in audio waveforms, image pixels, video frames, and text outputs.

??? Open Source: DeepMind has made SynthID’s code available to developers and companies, aiming for it to become a watermarking industry standard.

??? Deployment: SynthID is integrated into Google tools such as Gemini, ImageFX, VideoFX, and Vertex AI’s image generation capabilities.

?? Why It Matters: As AI-generated content continues to blur the line between real and artificial, SynthID offers a promising step toward maintaining transparency. While it’s not a silver bullet for misinformation or deepfakes, SynthID represents a significant advancement in AI safety. By open-sourcing the technology, Google hopes to lead the industry in standardizing watermarking tools.

?? Anthropic 's AI Now Navigates Computers Like a Human

Anthropic has introduced a groundbreaking new capability called ‘computer use,’ alongside upgrades to its AI models, enabling Claude to autonomously navigate computer interfaces by viewing screens, typing, moving cursors, and executing commands.

The Details:

?? Autonomous Navigation: Claude can now perform complex tasks across multiple applications and websites, mimicking human-like computer interactions.

?? General Computer Skills: Rather than creating a standalone tool, Anthropic trained Claude on general computer skills, making its functionality more versatile and human-like. ?? Sonnet 3.5 Upgrade: This upgrade significantly improves coding and tool use, outperforming competitors like o1-preview on key benchmarks.

?? Haiku 3.5 Model: Haiku 3.5 offers the capabilities of previous high-end models at a lower cost and faster speed.

?? Testing Phase: Anthropic notes that computer use is still in development (with some amusing hiccups), and suggests testing it on low-risk tasks as it continues to improve.

?? Why It Matters: Although many were anticipating Opus 3.5, the Sonnet and Haiku upgrades are no small feat. With the introduction of computer use into its foundation models, Anthropic has sent a strong message to automation startups—its models are evolving rapidly, and while the tech isn’t perfect yet, the potential for revolutionizing automation is huge.

?? 微软 Reveals Autonomous Copilot Agents

Microsoft has announced the upcoming release of new agentic capabilities for Copilot and Dynamics 365, allowing users to leverage both pre-built and custom autonomous agents to enhance business processes.

The Details:

?? Pre-Built Agents: Ten pre-built agents will be introduced in Dynamics 365, focusing on areas like sales, service, finance, and supply chain management.

?? Independent Operation: These agents can initiate tasks and respond to business signals autonomously, reducing the need for constant human oversight.

?? Copilot Studio: Users will soon be able to create their own agents with Copilot Studio, which will move from private to public preview next month.

?? Powered by OpenAI : The agents utilize OpenAI’s o1 model series, equipped with enterprise-grade features like encryption, data loss prevention, and safety guardrails.

?? Why It Matters: The agent revolution is no longer just on the horizon—Microsoft's Copilot agents represent a major leap forward. Described as "the new apps for an AI-powered world," these autonomous agents could soon make workflows as simple as selecting the right agent for a given task, marking a major shift in how businesses operate.

?? xAI Opens Grok API to Developers

Elon Musk’s xAI has launched a public beta of its API, allowing third-party developers to integrate the Grok language model into their applications.

The Details:

?? Grok-Beta Model: Now available via the xAI API, priced at $5 per million input tokens and $15 per million output tokens.

?? API Capabilities: Developers can generate text, write code, and call functions to connect with external tools and databases.

?? Model Options: While the ‘Grok-beta’ model is currently available, xAI’s Grok 2 and Grok Mini models are also mentioned in the API documentation.

?? Funding and Expansion: This API launch follows xAI’s recent $6B funding round, which increased the company’s valuation to $24B. Future capabilities could include a vision model for analyzing both text and images.

?? Why It Matters: xAI’s entry into the API market challenges major players like OpenAI and Anthropic. While Grok’s ‘rebellious’ nature may not suit traditional use cases, the API’s potential to leverage less restrictive and censorship-free features makes it appealing to developers looking for flexibility, especially with the addition of the upcoming Flux image model.

?? Meta Reveals New AI Models and Tools

Meta FAIR has unveiled a range of new research models and datasets, including advancements in image segmentation, language modeling, and solutions for accelerating large language models (LLMs).

The Details:

?? Spirit LM: An open-source multimodal language model that combines speech and text, generating more natural and expressive speech.

??? SAM 2.1: The latest update to Meta’s popular image and video segmentation tool, which has already seen over 700,000 downloads in just 11 weeks.

? Layer Skip: A novel solution that accelerates LLM generation by nearly 2x without requiring specialized hardware.

?? SALSA: A new tool designed for security testing, alongside other artifacts like Meta Lingua for language model training and a synthetic data generation tool.

?? Why It Matters: Meta continues to advance the AI landscape with open-source systems that challenge the dominance of closed models. As these tools evolve, the gap between open-source and proprietary solutions is closing fast, making it difficult to see how closed systems can maintain a lasting advantage.

?? IBM Introduces Granite 3.0: High-Performing AI Models Built for Business

At its TechXchange event, IBM unveiled Granite 3.0, its most advanced family of AI models to date. Released under the Apache 2.0 license, these models are designed to deliver top-tier performance across various business tasks while prioritizing transparency, safety, and efficiency.

The Details:

?? Granite 3.0 Models: The Granite 3.0 family includes general-purpose language models (8B & 2B), Granite Guardian for enhanced safety guardrails, and Mixture-of-Experts models for efficient inference and low latency.

?? Time Series Model: Granite’s Time Series model sets new performance records in zero/few-shot forecasting, beating models 10 times its size.

?? Granite-Powered watsonx Code Assistant: IBM also introduced an updated version of the watsonx Code Assistant, leveraging Granite models for general-purpose coding and application modernization.

?? Enterprise Integration: Granite 3.0 models will become the default for IBM’s Consulting Advantage platform, helping its 160,000 consultants deliver AI-powered solutions more efficiently.

?? Why It Matters: With Granite 3.0, IBM is pushing the boundaries of enterprise AI, offering open-source models that rival larger, more expensive alternatives. The models' efficiency, safety features, and enterprise integration mark a significant step forward in AI-driven business solutions, challenging the industry status quo.

Capital Watch: This Week's Funding Highlights

?? Perplexity AI Seeks Valuation of About $9 Billion in New Funding Round

Perplexity AI, an artificial intelligence search engine startup aiming to compete with Google, has initiated talks for a new round of fundraising, seeking to more than double its valuation to $9 billion, according to sources confirmed by CNBC .

The Details:

?? Valuation Jump: The company, previously valued at $3 billion in June 2024, is now looking to raise approximately $500 million in this new funding round.

?? Competitive Edge: Perplexity is positioning itself to challenge Google’s dominance in the search engine market, leveraging AI to offer generative search capabilities.

?? Funding History: Perplexity began 2024 with a valuation of $500 million and has already raised three rounds of funding this year, benefiting from the surge in generative AI investment.

?? Controversies: Despite its rapid growth, Perplexity has faced accusations of content scraping and plagiarism from media outlets like the New York Times, though the company has denied these claims.

?? Why It Matters: Perplexity’s push to secure $500 million in new funding signals strong investor interest in AI-powered search engines, as the company seeks to disrupt Google’s market share. However, ongoing controversies may challenge its growth trajectory.

?? AI Startup Zip Secures $2.2B Valuation in Its Series D Funding Round

AI-powered business software company Zip has announced a $2.2 billion valuation following its Series D funding round, led by BOND . The company raised $190 million, which will fuel its research and development, expand its AI solutions, and drive global expansion, particularly in Europe, the Middle East, and Africa.

The Details:

?? Valuation Increase: Zip’s valuation surged from $1.5 billion in 2023 to $2.2 billion after raising $190 million in Series D funding.

?? Investor Interest: The funding round was led by BOND and saw participation from new investors DST Global, Adams Street, and Alkeon, as well as existing backers Y Combinator and CRV.

?? Platform and Customers: Zip’s AI-powered platform streamlines procurement processes and reduces risk, with clients including Arm, Canva, Coinbase, Lyft, and Reddit.

?? Global Expansion: Zip plans to allocate funding toward expanding its presence across Europe, the Middle East, and Africa, alongside further AI development.

?? Why It Matters: Zip’s rapid growth and successful funding round highlight the continued interest in AI-driven business solutions, even amidst a slowing venture capital market. With major enterprises already onboard, Zip is well-positioned to lead in the AI-powered procurement space.

?? AI Chip Startup Kneron Seeks $300M at $1B Valuation

Kneron Inc., a Taiwanese-American AI chip startup, is in talks to raise $300 million in its latest funding round, which could value the company at $1 billion. The funds will be used to drive product innovation and support Kneron’s global expansion, including plans to open a regional office and R&D lab in Saudi Arabia.

The Details:

?? Funding Amount: Kneron is seeking to raise $300 million, potentially valuing the company at $1 billion.

?? Strategic Expansion: Plans include opening a regional office and future R&D lab in Saudi Arabia, aligning with the country’s digital economy ambitions.

?? Key Investors: Past investors include Qualcomm Ventures , Sequoia Capital China (HongShan), Hon Hai Precision Industry Co., and Horizons Ventures .

?? IPO Plans: Kneron is preparing for a 2025 Nasdaq listing via Spark I Acquisition Corp., a special purpose acquisition company (SPAC).

?? Why It Matters: Kneron’s edge AI chips are designed to power devices like cameras, cars, and drones without needing cloud connections. As demand for edge AI grows, with projections that this market could surpass cloud-based AI by 2032, Kneron is positioning itself to be a leader in AI hardware innovation.

?? interface.ai Raises $30M to Help Banks Automate Customer Requests

Customer automation platform Interface.ai has raised $30 million in a funding round led by Avataar Venture Partners, with $20 million in equity and $10 million in debt. This marks the first external funding for the AI startup, which had previously been bootstrapped.

The Details:

?? Funding Amount: $30 million raised, with $20 million in equity and $10 million in debt. ?? Product Offering: Interface.ai provides voice- and text-based AI agents to handle basic customer service requests for banks, such as modifying mortgage payments or opening new accounts.

?? Target Market: Interface serves over 100 financial institutions across North America, including community and regional banks that may lack the resources to develop their own AI solutions.

?? Founders: Launched in 2019 by Srinivas Njay, formerly of Microsoft, and Bruce Kim, founder of Inovaware.

?? Why It Matters: With banks increasingly adopting AI to enhance customer service, Interface.ai aims to democratize access to AI for smaller financial institutions. Its pre-built AI solutions are tailored specifically for the banking industry, offering scalable automation without the need for large in-house teams. The fresh funding will help expand the company’s team and accelerate its go-to-market efforts across North America and India.

?? AI Startup Neysa Raises $30M in Funding Round Led by NTTVC

Mumbai-based AI cloud and platform-as-a-service startup Neysa has secured $30 million (about Rs 252 crore) in a funding round co-led by NTTVC , Z47 (formerly Matrix Partners India), and Nexus Venture Partners . The new funds will support the scaling of infrastructure and fuel research and development to democratize scalable AI solutions.

The Details:

?? Funding Amount: $30 million raised in the latest round, following a $20 million round in April 2024.

?? Investors: Co-led by NTTVC, Z47, and Nexus Venture Partners.

?? Platform: Neysa’s flagship platform, Velocis, launched in July, provides on-demand access to high-performance computing infrastructure for AI projects.

?? Clientele: Neysa currently serves 12 customers, including AI-first digital natives, media and entertainment companies, service providers, software vendors, and the public sector.

?? Why It Matters: Founded in 2023 by Sharad Sanghi and Anindya Das, Neysa is leveraging its cloud and AI platform to provide scalable, secure AI solutions. With the new funding, Neysa is well-positioned to expand its infrastructure and continue developing innovative AI tools for diverse industries.

?? AI Startup Attention Raises $14M to Bring NLP Tech to Sales Calls

New York-based AI startup Attention has raised $14 million in a Series A funding round led by Alven, with participation from Eniac, Frst, Liquid2 Ventures, 645 Ventures, and Aglae. Attention uses natural language processing (NLP) to streamline sales calls by automatically filling out CRM systems and generating actionable insights.

The Details:

?? Funding Amount: $14 million raised in Series A funding, following a $3.1 million seed round in January 2023.

?? Investors: Led by Alven , with participation from Eniac, Frst, Liquid2 Ventures, 645 Ventures , and Aglae.

?? AI-Powered CRM Updates: Attention’s platform transcribes sales calls, generates action items like follow-up emails, and updates Salesforce records automatically. It can also identify churn risks and provide performance scorecards.

?? Founders: Launched in 2021 by Anis Bennaceur and Matthias Wickenburg, the duo pivoted the company in 2022 to focus on AI-powered sales call transcription and CRM automation.

?? Why It Matters: By automating the tedious aspects of sales calls, Attention’s AI enables sales teams to focus on closing deals rather than administrative tasks. With new funding, the startup aims to further automate business processes, ultimately putting sales teams "on autopilot."

?? Helsinki-Based DataCrunch.io Raises $12.7M to Hyperscale AI Computing in Europe

DataCrunch.io, a European AI computing infrastructure provider, has raised $12.7 million (€12 million) in seed funding, led by byFounders. This investment will help the company scale its infrastructure as it aims to become Europe’s first AI hyperscaler, delivering enhanced computing capacity for AI companies.

The Details:

?? Funding Amount: $12.7 million (€12 million) raised in seed funding, bringing DataCrunch’s total capital raised to $18 million.

?? Key Investors: Led by byFounders, with participation from J12 Ventures, and angels like Oskari Saarenmaa (Aiven) and Tuomo Riekki (Smartly).

?? Expansion Goals: The funds will support infrastructure scaling, team expansion, and the launch of Nvidia H200 servers and GB200 NVL72 clusters.

?? Clientele: DataCrunch’s services are used by AI innovators, including developers at OpenAI, Sony, Freepik, and Nex.art.

?? Why It Matters: As AI models grow increasingly resource-intensive, DataCrunch addresses a critical need for cost-effective, high-performance computing. By offering scalable, on-demand infrastructure designed for AI workloads, the company is positioning itself as Europe’s go-to provider for AI infrastructure.

Sovereign AI

???? Biden Orders AI Push with New Security Safeguards

The White House has issued a new national security memorandum directing federal agencies to accelerate AI adoption while setting clear boundaries for its use in sensitive areas like defense and intelligence.

The Details:

?? Government Strategy: The memo outlines plans for leveraging AI across national security departments while addressing the risks associated with the technology.

?? Restrictions: AI is prohibited from making autonomous decisions on nuclear weapons, targeting systems without human oversight, and conducting privacy-violating activities.

?? AI Chip Infrastructure: Biden highlighted the need to improve the U.S. AI chip infrastructure and directed agencies to help protect AI labs from foreign espionage.

?? Private Sector Collaboration: OpenAI published a blog alongside the memo, outlining its stance on national security and its partnerships with government initiatives.

?? Why It Matters: As AI continues to evolve, the U.S. is taking a proactive stance to establish guardrails around its use, particularly in national security. The memo also emphasizes the importance of protecting private sector innovations, signaling a shift in recognizing commercial AI as vital to national security.