- Hyrise AI
- Posts
- 🤖 DeepMind Alumni Introduce Haiper
Welcome, AI Enthusiasts.
Two former DeepMind employees, Yishu Miao and Ziyu Wang, have launched Haiper, an AI-powered video generation tool, following the recent release of OpenAI's Sora model.
Amazon recently introduced Rufus, an AI-powered chatbot within the Amazon Shopping app, aimed at aiding product research and offering recommendations.
In today’s issue:
🤖 DeepMind Alumni Introduce Haiper
🦾 Assessing Amazon's Rufus Chatbot: Mediocre Performance Leaves Room for Improvement
🛠️ 3 New AI tools
💻 Custom prompts ChatGPT and DALL-E 3
🤖 3 Quick AI updates
Read time: 3.5 minutes.
LATEST HIGHLIGHTS
DEEP MIND
🤖 DeepMind Alumni Introduce Haiper
Image source: Haiper
To recap: Two former DeepMind employees, Yishu Miao and Ziyu Wang, have launched Haiper, an AI-powered video generation tool, following the recent release of OpenAI's Sora model. Haiper, developed by Miao and Wang, focuses on video generation after initially working on 3D reconstruction using neural networks. The tool has garnered $13.8 million in seed funding and offers features like animating images and repainting videos in different styles. While currently concentrating on consumer-facing services, Haiper aims to develop a core video-generation model for broader use. It faces competition from OpenAI's Sora and other players like Google and Nvidia-backed Runway. Investors foresee the need for significant advancements in AI video technology before widespread consumer adoption.
The details:
1. Founders' Background: Haiper was founded by Yishu Miao and Ziyu Wang, both of whom are former employees of DeepMind, a leading artificial intelligence research lab. Miao previously worked at TikTok, while Wang has experience as a research scientist at DeepMind and Google.
2. Funding Rounds: Haiper raised $13.8 million in a seed round led by Octopus Ventures, with participation from 5Y Capital. Prior to this, the company secured a $5.4 million pre-seed round in April 2022, with contributions from angel investors such as Phil Blunsom and Nando de Freitas.
3. Focus on Core Model Development: While currently emphasizing its consumer-facing website, Haiper intends to develop a core video-generation model that could be offered to other developers. The company has not disclosed details about this model but has reached out to developers to try its closed API, emphasizing the importance of feedback for rapid iteration.
Here is the key takeaway: Haiper, founded by former DeepMind employees Yishu Miao and Ziyu Wang, has entered the AI-powered video generation market with a focus on advancing video generation technology. Despite facing competition from players like OpenAI's Sora and Google-backed initiatives, Haiper has secured significant funding and aims to develop a core video-generation model for broader use. The company's emphasis on community building, free features, and responsiveness to developer feedback underscores its commitment to innovation and differentiation in a rapidly evolving market.
AMAZON
🦾 Assessing Amazon's Rufus Chatbot: Mediocre Performance Leaves Room for Improvement
Image source: Unspalsh
In Summary: Amazon recently introduced Rufus, an AI-powered chatbot within the Amazon Shopping app, aimed at aiding product research and offering recommendations. Rufus, accessed via swiping up or tapping on the search bar, provides basic chat functionality with limited export options and no conversation sharing capabilities. While it offers advice on product attributes and features, its responses sometimes lack nuance and relevance, displaying occasional stereotypical biases. Rufus also refrains from controversial topics and generally avoids favoritism towards Amazon products over competitors. However, its responses to non-shopping queries can be inconsistent and occasionally inaccurate. Despite its current limitations, Rufus is still in beta, with potential for improvement to enhance the shopping experience on Amazon's platform.
Key points:
1. Introduction of Rufus: Amazon recently launched Rufus, an AI-powered chatbot integrated into the Amazon Shopping app for product research and recommendations.
2. Features and Limitations: Rufus offers basic chat functionality but lacks advanced features like conversation sharing and export options. While it provides product advice, its responses sometimes lack nuance and relevance, displaying biases and struggling with controversial topics.
3. Impartiality and Competitiveness: Rufus generally avoids favoritism towards Amazon products and competitors but occasionally fumbles non-shopping queries and provides inconsistent or inaccurate responses.
4. Potential for Improvement: Rufus is still in beta, and Amazon promises improvements to enhance its functionality and address user concerns. Despite its current limitations, Rufus has the potential to become a valuable tool for shopping on Amazon's platform.
Our thoughts: On one hand, it's commendable that Amazon is exploring AI-driven solutions to enhance the shopping experience. Rufus has the potential to assist users with product research and recommendations, which could streamline the shopping process.
However, Rufus currently falls short in several areas. Its responses lack depth and nuance, often displaying biases and struggling with complex queries. Additionally, the limitations in conversation sharing and export options limit its usefulness.
Despite these shortcomings, Rufus is still in beta, and I'm optimistic about its potential for improvement. If Amazon can address its current issues and enhance its functionality, Rufus could become a valuable tool for Amazon shoppers. However, Amazon needs to prioritize transparency regarding data usage and implement safeguards to ensure unbiased and accurate responses.
TRENDING TECHS
📉 SingleStore Kai- 100x-1000x faster analytics for MongoDB applications
🧾 Pythia World-Free AI tools for founders, product leaders, and marketers
🤖 ChatGPT Master of Data- Use ChatGPT to become a Data Ninja
AI DOJO
Advisor:
Prompt: Write a persuasive email to convince your boss to implement a new company-wide initiative.
DALL-E 3
Visual creation:
Prompt: Create visual representations of famous landmarks around the world.
QUICK BYTES
Google has announced a significant search quality update aimed at improving the reliability of search results by targeting the prevalence of low-quality or spammy websites that often rank highly. The update will focus on downranking pages that prioritize search engines over user experience, including those with poor content quality and those designed to manipulate search rankings. Google's efforts will address content generated at scale, whether by humans or AI, as well as site reputation abuse where reputable sites host low-quality content to gain ranking benefits. This update responds to concerns raised by both users and search marketers regarding the declining quality of Google Search results, particularly in the face of increasing SEO spam. Publishers and niche review sites have expressed frustration over their content being overshadowed by less reputable sources in search rankings. Google aims to tackle these issues before enforcement begins on May 5, potentially reshaping how consumers perceive the usefulness of Google Search amidst advancements in AI and content summarization technologies.
Numbers Station, a startup leveraging large language models (LLMs) for data analytics, has launched its cloud-based product, Numbers Station Cloud, in early access. This service enables users within enterprises to analyze internal data using a chat interface. While similar tools translate natural language queries into SQL, Numbers Station focuses on overcoming limitations by tailoring its platform to individual companies through a semantic catalog. This catalog ensures alignment with a company's specific metrics and definitions, enhancing precision in query responses. Despite challenges in building this platform, Numbers Station aims to expand its AI platform for analytics beyond chat services, addressing various data problems for enterprises. The company has already garnered interest from Fortune 500 clients like Jones Lang LaSalle, indicating potential for impactful business outcomes through its innovative approach to data analytics.
Ema, a startup based in San Francisco, is unveiling its product of the same name, aiming to revolutionize how AI, particularly generative AI, transforms work processes. With a vision of creating a "universal AI employee," Ema targets automating mundane tasks within enterprises, allowing employees to focus on more strategic work. Backed by a $25 million funding round and a roster of notable customers, including Envoy Global and TrueLayer, Ema offers two main products: Generative Workflow Engine (GWE) and EmaFusion, designed to emulate human responses and evolve with feedback. Unlike traditional robotic process automation or AI acceleration tools, Ema leverages over 30 large language models and domain-specific models to address accuracy, hallucination, and data protection issues. The startup's impressive founding team, led by CEO Surojit Chatterjee and head of engineering Souvik Sen, brings extensive experience from Coinbase, Google, and Okta, bolstering confidence in Ema's ability to execute its ambitious goals. Investors, including Accel and Prosus Ventures, are drawn to Ema's ability to drum up business and its founders' track records. Ema's approach, cutting across different large language model silos, suggests a potential for diversification and broad applicability, addressing concerns around fragmentation and data access in enterprise AI solutions.
SPONSOR US
🦾 Get your product in front of AI enthusiasts
THAT’S A WRAP