- Human In The Loop
- Posts
- ChatGPT as your AI assistant, Claude outcodes rivals, ElevenLabs' real-time AI voices
ChatGPT as your AI assistant, Claude outcodes rivals, ElevenLabs' real-time AI voices
🛠️ Product Updates
Image
Endless Web3 Genesis Cloud is integrating Stability AI's Stable Diffusion 3.5 into its blockchain infrastructure, pushing decentralized AI beyond current market offerings. The partnership introduces a custom Sketch-to-Image workflow in Endless' Luffa app (200,000+ users), allowing anyone to transform rough sketches into polished visuals with built-in NFT capabilities. This strategic collaboration bridges the AI-blockchain gap, positioning Endless ahead of competitors with a unique mix of advanced image generation, simplified creative tools, and on-chain monetization options for both creators and mainstream users.
Text
Character.AI has expanded beyond text-based chatbots with three new multimedia features now available to all users. The platform's AvatarFX video-generation model—previously a subscriber exclusive—now allows free users to create five daily videos by uploading photos, selecting voices, and writing dialogue. Users can also craft Scenes with pre-populated storylines and will soon access Streams for character interactions, all shareable on an upcoming community feed. While implementing safeguards like blocking real person photos and watermarking videos, these features position Character.AI ahead of competitors by integrating video creation with social sharing—though concerns persist about potential misuse despite safety measures.
Anthropic's Claude 4 now offers five streamlined integration paths for developers, turning the AI assistant into a powerful coding ally. The latest Sonnet and Opus models can be leveraged through the Claude web app with GitHub repository support, Claude Code for direct project enhancement, automated GitHub workflows via "@claude" tagging, VSCode integration with diff viewing capabilities, and a Python SDK for custom applications. While Claude 4 Opus delivers advanced capabilities, its premium pricing positions it for enterprise users, with even the more accessible Sonnet model commanding higher rates than competing coding assistants.
Audio
ElevenLabs has launched Conversational AI 2.0, pushing the envelope on AI-driven voice interaction with sophisticated turn-taking, automatic multilingual support, and multimodal inputs. The platform now integrates Retrieval-Augmented Generation for real-time knowledge access with minimal latency—crucial for healthcare and customer service applications. Enterprise users will appreciate the added HIPAA compliance and seamless Twilio/SIP trunking integration. Coming just four months after their initial conversational release, this aggressive update signals ElevenLabs' determination to dominate the AI audio space while competitors struggle to match its comprehensive feature set.
ElevenLabs has supercharged its voice AI platform with Conversational AI 2.0, just four months after the original release. The upgrade introduces human-like turn-taking models, integrated knowledge retrieval with privacy safeguards, and automatic language detection that eliminates manual switching. Enterprise users gain HIPAA compliance, optional EU data residency, and expanded telephony features including outbound calling and SIP integration. This rapid evolution transforms what businesses can build—from multilingual customer service bots to healthcare-compliant voice agents—without sacrificing natural interaction.
Video
Microsoft brings OpenAI's Sora to the masses with Bing Video Creator, now available free on iOS and Android. Users can generate 5-second vertical videos from text prompts with surprising quality—something OpenAI hasn't even offered to its paying customers yet. The service smartly limits "Fast" generations to ten before requiring standard speeds or Microsoft Rewards points, balancing accessibility with server demands. While currently limited to mobile and lacking desktop support, this move leapfrogs competitors by democratizing AI video creation for everyday users.
💡 Insights
OpenAI's leaked strategy document reveals ambitious plans to transform ChatGPT into an "AI super assistant" that serves as users' primary "interface to the internet" by 2025. The roadmap shows a two-phase approach: first developing "T-shaped" capabilities combining broad daily skills with deep expertise in complex tasks, followed by monetization efforts. OpenAI appears particularly concerned about Meta's ability to embed competing AI functionality across its products, advocating for regulations that would allow users to choose ChatGPT as their default assistant. This strategic pivot underscores CEO Sam Altman's recent focus on massive data center investments to support the growing infrastructure demands.
📚 How-To
ScrapeGraph and Google's Gemini join forces in a new DIY competitive intelligence toolkit that automates market analysis end-to-end. The tutorial demonstrates how ScrapeGraph's SmartScraperTool extracts competitor data (products, pricing, tech stacks) while Gemini 1.5 Flash transforms raw information into actionable insights on market positioning and strategic gaps. The Python implementation, complete with a CompetitiveAnalyzer class, offers businesses a scalable pipeline for continuous competitor monitoring—particularly valuable for AI/SaaS companies and e-commerce brands needing real-time market intelligence without enterprise-level budgets.
⭐ Reviews
Google's Veo 3 AI video model brings automatic audio generation to Gemini Ultra subscribers and Vertex enterprise users—a first among competitors, though quality ranges from impressive to awkwardly literal. While showing fewer errors than its predecessor, Veo 3 frustrates with painfully low daily generation limits (just five videos), long processing times, and absent fine-tuning tools. The $250/month Ultra subscription (currently discounted to $125) feels steep for what's essentially an impressive but unfinished product better suited for AI enthusiasts than professional creators.
🎤 Takes
In a surprising twist, Anthropic's free Claude 4 Sonnet outperformed its paid counterpart, Claude 4 Opus ($20/month), in comprehensive coding tests. Sonnet aced all four challenges while Opus stumbled on two critical tests—failing to address security vulnerabilities in a WordPress plugin and producing less robust currency validation code. The only area where the premium model showed slight superiority was in script elegance for a complex Chrome-AppleScript integration. This unexpected performance gap raises questions about Anthropic's tiered pricing strategy and what exactly customers are paying for in the Pro plan.
Reply