AI
5 min read
AI Notes July 2025

Q3 2025 Cutting through to the must-know AI tools
General AI tools
- GPT-5 — Released with advanced reasoning, multimodal support, and agent mode for autonomous workflows. https://chat.openai.com
- Claude — Claude Opus 4.1 boosts coding, agentic reasoning, and logic. Enterprise adds 500k-token chats, Artifacts; Claude Code’s /security-review scans code. https://claude.ai
- Perplexity — Research (formerly Deep Research), Comet AI browser, internal knowledge search, image generation, Labs. https://www.perplexity.ai
- Gemini 2.5 — Gemini now supports Storybook—create 10-page illustrated & narrated storybooks in 45+ languages. It also enables photo-to-video conversion (AI Ultra/Pro only) and image editing directly in the Gemini app. https://gemini.google.com
- Microsoft Copilot — Copilot now runs on GPT-5 with smart mode routing. Edge Copilot Mode adds multi-tab AI browsing and voice navigation. Copilot 3D turns images into 3D models. Copilot Appearance brings animated expression to the assistant. AI agents, memory, better image handling, call summaries, and admin tools enhance Microsoft 365 Copilot. https://copilot.microsoft.com
Creative AI tools
- Adobe Firefly — Structure Reference - Lock layout and vary creative content. Boards. https://blog.adobe.com
- Adobe Illustrator — Generative Shape Fill packs outlines with vector art. https://adobe.com/illustrator
- Midjourney V7 — Enhanced video generation modes, improved logos, new editor. https://midjourney.com
- Leonardo AI — Real-time canvas with style references and 3D mesh export. https://leonardo.ai
- Higgsfield — Consistent characters and multi-reference image generation for ad, fashion and editorial campaigns. https://higgsfield.com
- Freepik AI Suite — Scene-locked composition generator. https://freepik.ai
- ArchiVinci — Sketch-to-photo architectural renders. https://archivinci.com
- Flair — Drag-and-drop product shoots, now with fashion-model mode. https://flair.ai
- Morphic — Storyboarding to final cut film studio in browser. https://morphic.com
- Story Diffusion — Maintains comic character consistency across panels. https://storydiffusion.ai
- DALL·E 3 — Inpainting and regional editing within ChatGPT. https://labs.openai.com
- Khroma — AI color scheme generator learning user’s taste. https://khroma.co
- AutoDraw — Quick sketch helper for wireframes. https://autodraw.com
- Fontjoy — Font pairing generator. https://fontjoy.com
- Botika — AI fashion models for Shopify images. https://botika.ai
- Magnific — AI upscaler with prompt-guided detail (added). https://magnific.ai
- Presti — Drops realistic furniture into AI-generated rooms. https://presti.ai
- IC Light V2 — Text-guided portrait relighting tool. https://iclight.ai
- Krea 3D Objects — Image to textured 3D mesh generation. https://krea.ai
- Flora — Team collaboration chaining text, images, and videos. https://flora.so
- Stable Virtual Camera — Generate explorable 3D scenes from photos. https://stability.ai
Video AI tools
- Runway Aleph — Video edits via prompt, including object and lighting changes. https://runwayml.com
- Veo 3 — Veo 3 can be accessed via Leonardo. https://leonardo.ai/
- Kling — Fast, controllable 1080p generative video for social media. https://scenario.com
- Sora - From Open AI. Sora 2 is coming soon. It aims to fix shaky human motion, add synchronized audio, stretch video length, and boost quality and resolution. https://sora.chatgpt.com/explore
- Midjourney and Firefly have video too
- Scenario — Direct video generation control by sketching on frames. Veo 3 can be accessed via Leonardo. https://scenario.com
- Seedance 1.0 — Narratively consistent video creation, multi-shot and character cohesion. https://scenario.com
- Mirage (Decart) — Real-time world transformation and style transfer for video streams. https://mirage.decart.ai
- Moonvalley Marey — Hybrid 2D-to-3D filmmaking AI platform with camera motion control. https://moonvalley.ai
- Dream Machine (Ray 2) — Cinematic AI-generated clips with physics. https://lumalabs.ai
- Amazon Nova Reel — Budget-friendly text to video generation. https://aws.amazon.com
- Pika Pikas — AI actor/background swaps via prompts. https://pika.art
- Gemini Video Generator — Generates AI-produced animated scenes. https://ai.google.com/video
- Descript — Video/audio editor with overdub & eye contact fixes. https://descript.com
- Viggle — Motion transfer & greenscreen-ready characters. https://viggle.ai
- Synthesia — Avatar video creation with multilingual voiceovers. https://synthesia.io
- Facefusion — Open-source GPU face-swap tool. https://github.com/facefusion
- Deep Live Cam — Real-time deepfake streaming for VTubers. https://github.com/hacksider
- Revid.ai — Auto video summarization for social shorts. https://revid.ai
- Riverside — Lossless recording with text-based editing. https://riverside.fm
- Flux — Fast 1080p video generator with style control. https://flux-ai.com
- HeyGen — AI face swaps with multilingual voice control. https://heygen.com
- Arcads — Automated ad video creation platform. https://arcads.ai
- Filmora AI — Video editing with auto cut, style transfer, and filler removal. https://filmora.wondershare.com
- Keytake — Convert documents and URLs into explainer videos. https://keytake.ai
Music & Audio AI tools
- Suno V4.5 — Fast 2-track mixing with improved vocals. https://suno.com
- Udio — AI music generation for songs and skits. https://udio.com
- ElevenLabs — Text-to-speech and sound effects, newly added music generation. https://elevenlabs.io
- Play HT — High-quality, low-latency voice API. https://play.ht
- Artlist AI Voiceover — Multilingual voiceover with Premiere support. https://artlist.io
- Hume Octave — Emotion-aware TTS with controllable tone. https://hume.ai
- SoundHound Chat — Car assistant AI for ordering and commands. https://soundhound.com
- ElevenLabs Bark — AI speech designed for pet-tech toys. https://elevenlabs.io
- Amazon Nova Sonic — Expressive voice AI powered by AWS Bedrock. https://aws.amazon.com
- Synthflow — AI tool automating meeting bookings and CRM notes. https://synthflow.ai
- Shamaze — AI voices scribing bedtime stories in personalized voice. https://shamaze.com
UI design AI tools
- Figma “First Draft” — Relaunched AI layout generator. https://theverge.com
- Uizard — UI mockups from text prompts. https://uizard.io
- Galileo — Generates polished UI screens via AI. https://usegalileo.ai
- Dora — No-code 3D site builder with WebGL export. https://dora.run
- Webflow AI Site Builder — AI builds hosted sites from briefs. https://webflow.com
- Attention Insight — Heatmaps predicting user gaze for UX. https://attentioninsight.com
- Operative.sh — Automated UX testing with screenshots. https://operative.sh
- Scene 2.0 — Whiteboard to website creation platform. https://scene.so
- AI CSS Animations — Generate CSS animation from text. https://aicss.dev
- Same.dev — Clone websites into editable React code. https://same.dev
- Firebase Studio — Build full-stack apps from prompts. https://firebase.google.com
Marketing AI tools
- Jasper — Brand voice control and team permissions. https://jasper.ai
- Aha — AI influencer team management and ROI dashboard. https://aha.inc
- Virallyst — Caption rewriting, hook testing, and auto-posting. https://virallyst.com
- Warmy — Domain warm-up service with updated deliverability guides. https://warmy.io
- Happenstance — Plain language AI network search. https://happenstance.ai
- AiSDR — LinkedIn automated meeting scheduler. https://aisdr.com
- Reachy — Respectful LinkedIn outreach agent. https://reachy.ai
- Keak — AI platform for landing pages with A/B test features. https://keak.com
No Code Builders (Vibe Coding)
- Replit Agent — Prompt-driven app deployment. https://replit.com
- Lovable - Introduced Agent Mode (Beta): AI now thinks, plans, edits code autonomously, searches files, reads logs, fetches docs, edits images https://lovable.dev/
- V0 by Vercell - Design Mode introduced - Easily tweak UI elements—typography, layout, colors—and preview design changes inline. https://v0.dev/chat
- GitHub Copilot in VSCode - Integrates directly into VS Code’s sidebar and terminal, letting you ask natural-language questions, generate code, explain errors, and even run inline commands in your project context.
- Emergent — No-code mobile app builder powered by agentic AI. https://emergent.sh
Local Models and Tools to Run Them
Tools
- Ollama - Ollama is an open‑source tool that lets you run large language models (LLMs) locally on your own machine, no cloud required. It offers a command‑line interface (CLI) and optional API for model management. You can download and use models like Meta’s Llama 2 and 3, Mistral, and Google’s Gemma directly on Windows, macOS, or Linux, keeping your data private and offline. https://ollama.com/
- Chatbox - Chatbox AI is a cross-platform AI client and smart assistant you can use on Windows, macOS, Linux, Android, iOS, and the web. Keeping your data private as everything stores locally. https://chatboxai.app/
- Open Web UI - Open Web UI is a self‑hosted, extensible AI interface you run on your own system entirely offline. It supports both Ollama and OpenAI‑compatible APIs. You install it via Docker, pip, or Kubernetes and access it through a clean, web‑based UI. It packs advanced features: RAG (retrieval‑augmented generation), document import, Markdown and LaTeX support, mobile PWA experience, and user roles and permissions. https://openwebui.com/
- Comfy UI - Comfy UI is an open‑source, node‑based interface for generative AI that lets you build, remix, and control creative workflows exactly how you want. You connect functional nodes on a visual canvas to load models, write prompts, and process outputs in real time, all running locally, fast, free, and infinitely customizable. https://www.comfy.org/
- Automatic 1111 - Automatic1111 is a browser‑based interface for Stable Diffusion that runs locally, built using Gradio. It exposes advanced image generation tools like inpainting, outpainting, upscaling, prompt attention, and extensions through an accessible graphical UI without needing command‑line interaction. https://github.com/AUTOMATIC1111/stable-diffusion-webui
Models
- OpenAI GPT-OSS — Open-source models (120B & 20B), optimized for reasoning and coding, Apache 2.0 licensed. https://huggingface.co/openai
- DeepSeek R1-Omni — 671B open-weights, 200k context, free for research. https://huggingface.co/deepseek-ai/DeepSeek-R1
- Llama 3 (8B/70B) — Llama 3.1 adds a 405B-parameter model with 128K token context, multilingual coding plus safety via Llama Guard 3. Llama 3.2 brings vision-capable and edge-friendly models. https://ai.meta.com
- LG EXAONE Deep 32B — Laptop-friendly model scoring near GPT-4 on STEM tasks. https://huggingface.co/LGAI-EXAONE/EXAONE-Deep-32B
Ethical Models
- FLite - F Lite is an open‑source, 10 billion-parameter diffusion model trained exclusively on about 80 million high-quality, legally licensed, safe-for-work images from Freepik’s stock library. Available in two versions, Standard and Texture, it delivers strong performance in illustrative and vector styles, while photorealism, complex scenes, and short prompt handling remain areas to improve. It integrates with ComfyUI, Hugging Face, and the diffusers pipeline, and is released under a permissive CreativeML Open RAIL-M license. https://www.freepik.com/blog/f-lite-freepik-and-fal-ai-unveil-open-source-image-model-trained-on-licensed-data/
- Bria AI - Not a local model but can be configured. Bria 3.2 is a compact, open‑source text‑to‑image model (4B parameters), rivaling top models in quality. It’s faster to fine‑tune, better with text prompts, and built entirely on licensed data. https://bria.ai/
- Blunge - Blunge now champions ethical AI image generation. It protects artists’ rights through manual ownership checks, full copyright retention, private custom models, and secure hosting. You own your unique style—and control it entirely. https://www.blunge.ai/
Other AI tools
- Gradio -Gradio is an open-source Python library that lets you quickly build and share interactive web apps for machine learning models, all from a simple script. https://www.gradio.app/
- Wide Research by Manus — Tool for handling multiple research tasks concurrently. https://manus.ai
- Harvey — AI bot specializing in contract review and legal due diligence. https://harvey.ai
- CopyCat — Low-code browser automation from natural language instructions. https://www.runcopycat.com/
- Exa Search — Hybrid semantic+keyword search API for docs and e-commerce. https://exa.ai
- Lambda Inference API — Pay-per-token gateway to every major frontier model. https://lambdalabs.com/inference
- Zapier MCP — One prompt triggers 8,000+ SaaS actions; perfect for agents. https://zapier.com/mcp
- Kimi K2 — Chinese open-weight 1T-parameter LLM outperforming GPT-4 in coding/math tasks. https://kimi.moonshot.cn
- Payman — Agent-driven hiring with secure payment. https://payman.ai
- Pinokio — One-click local deployment of AI apps. https://pinokio.computer
- Cursor v1.3 — Adds shared terminal, faster context-aware coding chat. https://cursor.com
- Browser Use — Library for headless browser automation in agents. https://github.com/browser-use
- Databutton MCP — Drag-and-drop AI workflow builder. https://databutton.com/mcp
- Documenso — Open-source alternative to DocuSign. https://documenso.com
- Terra Security — AI-driven penetration testing platform. https://terrasecurity.ai
- Agent Simulate — Synthetic user load testing for UX research. https://agentsimulate.com
Education AI tools
- Google Gemini Guided Learning — Interactive stepwise tutoring with flashcards & video. https://gemini.google.com
- OpenAI Study Mode — Prompts users for reasoning, stepping away from direct answers. https://help.openai.com/study-mode
- Claude for Education — Large-context tutor and worksheet generator. https://claude.ai/edu
- OpenAI Academy — Courses on prompt engineering and AI safety. https://openai.com/academy
- AI Tutor by Roadmap.sh — Interactive study tool following coding/learning roadmaps. https://roadmap.sh/ai-tutor
- TurboLearn — Note-taking, flashcards, and quizzes from various media. https://turbo.ai
- NotebookLM Audio/Video Overviews — Podcast-style or visual learning summaries powered by AI. https://blog.google.com
- Nvidia free AI courses — Hands-on training for AI and ML fundamentals. https://nvidia.com
- Globe Explorer — AI-based interactive knowledge maps. https://explorer.so
- Class Central — Indexed catalogue of online AI courses. https://classcentral.com
- University of Illinois AI — MBA-level AI specialization online. https://coursera.org
- Microsoft Generative AI Beginner — Twelve-lesson introductory curriculum. https://microsoft.com
- Maven AI Bootcamps — Cohort courses on safety, prototyping, product. https://maven.com
Worth checking out
- Devin 2.0 — Autonomous developer agent. https://cognition.ai
- Convergence Parallel — Multi-agent orchestration framework. https://convergence.ai
- Mistral OCR API — High-speed multilingual OCR API. https://mistral.ai
- NotebookLM “Discover Sources” — Curated research companion. https://notebooklm.google.com
- Keytake — Converts documents into branded explainer videos. https://keytake.ai
Think pieces and resources
- Google Prompt Engineering Guide — Best practices for prompt design. https://ai.google.com/prompts
- State of AI 2025 — Concise trend graphs and insights. https://stateofai2025.com
- Agent Survey 2025 — 264-page review on autonomous agents. https://arxiv.org/abs/2504.12345
- World Economic Forum Future Jobs Report — Skills and wage impact analysis. https://weforum.org/reports/future-jobs
- Hugging Face SmolAgents Course — Build lightweight AI agents. https://huggingface.co
- 601 AI Income Ideas — Monetization guide for AI applications. https://hubspot.com
- IBM Building AI-Powered Chatbots — Vendor-neutral course. https://cognitiveclass.ai
- Elements of AI (Stanford/Harvard) — Free fundamentals of AI and ethics. https://elementsofai.com
July 2025 Trends and Market Highlights
- Agentic AI models like GPT-5, ChatGPT Agent Mode, Google Gemini Deep Think, and Chinese Kimi K2 push beyond simple question-answering to real task completion and multi-tool orchestration.
- AI advancements in video generation and editing (Runway Aleph, Scenario Veo 3, Seedance, Mirage, Moonvalley Marey) bring cinematic quality and control to prosumers and studios.
- AI coding platforms: Amazon Kiro, Windsurf’s acquisition by Cognition, Cursor upgrades, and application of agentic coding agents signify a maturing developer ecosystem.
- Emergence of Generative Engine Optimization (GEO) in place of traditional SEO with tools like Peec AI, SpyGlass, and exploration of AI agent search behaviors.
- Increasing regulatory divides: Google signs EU AI code while Meta rejects it; US-China tech tensions manifest in chip supply and model development.
- Medical AI achieves superhuman diagnosis accuracy and cost-efficiency, with Microsoft’s MAI-DxO leading enterprise adoption.
- AI privacy and safety developments: ChatGPT’s Study and Safety Modes, Claude’s honesty updates, and rising enterprise focus on secure, ethical AI usage.
- Major tech investments and acquisitions continue: Nvidia AI chips, massive funding rounds, talent wars highlighting shifting power dynamics.
(If you spot any missing links, please DM or comment!)

John Luba
Author & Content Creator