Microsoft just kicked off its 2023 release wave 2 for Dynamics 365 and Power Platform, rolling out fresh Copilot capabilities across ERP, customer service, and low-code development. Teams can now generate Power Automate workflows with plain text and receive automated purchasing recommendations, with features rolling out through March 2024.
Researchers from ANU, Oxford, and BAAI just developed 3D-GPT, an AI system that translates text prompts into 3D environments. By deploying specialized agents to conceptualize scenes and write code directly for Blender, it automates procedural 3D modeling. It isn't photorealistic yet, but it signals a major shift toward text-driven asset creation.
Stack Overflow just laid off 28 percent of its staff—over 100 employees—primarily targeting its go-to-market and support teams. As developers increasingly turn to AI chatbots for code assistance, the veteran coding forum is feeling the heat just a year after doubling its headcount to chase profitability.
Portkey just revealed that GPT-4 latency has steadily dropped over the past three months, effectively cutting 99th-percentile wait times by more than half. While complex prompts still spike processing, median latencies now sit comfortably below 1ms per token. It is still pricier, but speed is no longer the bottleneck holding teams back.
Leaked screenshots reveal Google is quietly bringing its multimodal Gemini model to MakerSuite alongside an unannounced tool called "Stubbs." The secret feature lets creators generate, deploy, and share functional app prototypes with just a single text prompt and an optional reference image.
OpenAI is reportedly planning major developer updates at its upcoming November 6 conference to slash costs by up to 20 times. The upgrades include a memory-enabled stateful API to curb token usage and vision capabilities for analyzing images. Why it matters: OpenAI wants to cement its platform as the default infrastructure before startups defect to open-source alternatives.
Replit just dropped Replit Code V1.5 3B on Hugging Face, an open-source completion model trained on 1 trillion tokens across 30 programming languages. Despite its compact 3B size, fine-tuning lets it outperform larger models like CodeLlama 7B. Best of all, it is free for commercial fine-tuning without strict restrictions.
Zapier just launched Canvas, a visual flowchart tool with built-in AI that helps users plan end-to-end business workflows and convert them into automated zaps. Users can simply describe a problem to generate a process diagram. The company also made its automation-first database, Tables, generally available.
Amazon just officially launched Amazon Bedrock, opening up its fully managed platform that brings leading foundation models from Anthropic, Meta, and others into a single API. Alongside the rollout, AWS made its Titan Embeddings model available for enterprise search and RAG workflows, and previewed private codebase customization for CodeWhisperer.
Cloudflare just launched a suite of edge tools—Workers AI, Vectorize, and AI Gateway—to help developers deploy models on nearby GPUs with pay-as-you-go pricing. It also includes vector storage and traffic caching to keep costs under control. The move could make running open models much cheaper and faster without central cloud lock-in.
GitHub just opened the public beta of Copilot Chat to all individual subscribers, expanding the feature beyond enterprise teams. Developers can now debug code, get explanations, and generate security patches directly inside VS Code and Visual Studio using plain English. It brings conversational AI coding straight into the everyday workflow of solo devs.
OpenAI just announced its first global developer conference, OpenAI DevDay, scheduled for November 6. The one-day event will feature a keynote and breakout sessions previewing new developer tools, with parts streamed online. While over 2 million developers use its models, in-person attendance will be capped at just a few hundred.
Niantic just rolled out generative AI developer tools for its 8th Wall platform, integrating Inworld AI and OpenAI. Devs can now drop interactive AI characters directly into browser AR experiences and use DALL-E to generate custom textures for 3D objects on the fly, backed by two ready-to-test sample projects.
Google just expanded its preview of Duet AI across Google Cloud, infusing generative AI into coding, database migration, and security ops ahead of general availability later this year. It now assists with legacy code refactoring, PromQL queries, and BigQuery SQL generation directly inside IDEs and cloud consoles. Why it matters: Google is making sure developers rarely need to leave their flow state or switch tools to get AI assistance.
Google just dropped a massive update to Vertex AI at its Cloud Next conference. PaLM 2 gets a 32k-token context window alongside quality boosts to Imagen and Codey, while third-party heavyweights like Claude 2 and Llama 2 join its Model Garden. With new extensions and enterprise search tools, Google is making a direct bid to outflank AWS Bedrock.
Meta just dropped Code Llama, a specialized open-source coding model built on Llama 2 that scales up to 70B parameters. It handles code generation, debugging, and infilling across major languages, boasting a context window of up to 100k tokens. Why it matters: it gives developers a powerful, free alternative to proprietary coding assistants.
Scenario just launched a Unity plugin that lets game developers create style-consistent visuals inside the engine. Creators can now generate tilemaps, avatars, props, and environments without leaving their Unity workflow, eliminating the friction of bouncing between design apps and the game engine.
OpenAI just partnered with Scale to help enterprises customize its advanced models using proprietary data. The collaboration pairs OpenAI's fine-tuning capabilities with Scale's enterprise data engine and evaluation expertise, while ensuring customer data is never used to train other models.
OpenAI just launched fine-tuning for GPT-3.5 Turbo, allowing developers to customize models with their own proprietary data. Early testers managed to shrink prompt sizes by up to 90% by baking instructions directly into the model, speeding up API calls while slashing operational costs. Why it matters: Developers can now achieve consistent, tailor-made outputs without blowing their token budget.
Hugging Face and VMware just unveiled SafeCoder, a commercial solution designed as an on-premises alternative to GitHub Copilot. The platform lets companies fine-tune StarCoder models on their internal codebases inside their own VPC, ensuring proprietary code never touches third-party servers. Why it matters: Enterprise developers get tailored AI pair programming without leaking IP or running into licensing minefields.
