AI News You Can Use: Turbulent Week of Releases, Claude’s New Connectors, and the Push Toward AGI‑Like Assistants

Overview

This week has been described as the most turbulent in AI releases. Major announcements include a near‑AGI open‑source bot (originally Claudebot, now Maltbot), a huge upgrade to Claude’s connectors, new Google Gemini capabilities, OpenAI’s Prism for research, and several creative‑AI demos. The host, Igor Pagani, walks through each development, explains practical use‑cases, and reflects on the broader trend toward a single, highly capable digital assistant.

Claudebot → Maltbot

• What it is: An open‑source “ChatGPT on Crystal” that can control your computer, browse the web, send messages, and schedule actions.
• History: Developed and released as Claudebot, then renamed Maltbot after a cease‑and‑desist from Anthropic over the name similarity.
• Strengths: Direct access to Telegram, email, browsers, and the OS; can automate tasks that big companies are hesitant to expose for security reasons.
• Weaknesses: Very rough around the edges – frequent crashes, bans on Gmail/X, and a lot of manual troubleshooting.
• Takeaway: A reckless but eye‑opening proof‑of‑concept that shows what a fully integrated AI assistant could look like.

Claude Connectors & Gamma Presentation Builder

• Connector upgrade: Claude now lets users browse and enable external apps directly from the settings menu.
• Demo with Gamma (formerly GMA): After connecting, a user can ask Claude to create a presentation (e.g., “army of AI agents dressed as penguins”). The generated slides appear inside the chat and can be edited, though heavy edits are slower than using the native app.
• Workflow advantage: Start brainstorming in Claude, then hand off to the specialized app for polishing and final output.
• Other connectors: Slack, Excel, Chrome, etc., each with slightly different interaction patterns (some end the conversation, some continue).

Claude for Excel & Free‑Plan Enhancements

• Excel extension: Now available on the Pro plan ($20/month) and lets users manipulate spreadsheets via natural language, replacing many VBA tasks.
• Free‑plan perks: Users can create and download Excel sheets, PowerPoints, markdown, Word docs, and more without paying.
• Skills feature: Users can package a successful prompt sequence as a reusable “skill,” streamlining repetitive workflows.

Direction Toward an Integrated AGI‑Like System

• Recent releases (Maltbot, Claude connectors, Excel extension, Chrome integration) point to a single ecosystem that can:
• Control the browser and OS.
• Interact with specialized apps (presentation makers, spreadsheets, communication tools).
• Offer a conversational interface that feels like a personal assistant.
• The host argues this convergence is a step toward true AGI, where one system performs a wide range of tasks at human‑level quality.

Google Gemini Agentic Vision

• New capability: Gemini can analyze images using code, drawing bounding boxes and counting objects (e.g., a hand with six fingers).
• Potential uses: Detailed inspection of engineering diagrams, medical images, or any scenario requiring pixel‑level reasoning.
• Context: Similar functionality appeared in OpenAI’s GPT‑4 Vision but was later down‑scaled; Google appears to be re‑investing in it for future Gemini updates.

OpenAI Prism for Scientific Writing

• What it does: A customized ChatGPT workspace tailored for reading, annotating, and co‑authoring research papers.
• Availability: Free to use, aimed at researchers who need AI assistance in literature review and manuscript drafting.

Google SAT‑Prep App

• A single‑purpose Gemini app that quizzes users on SAT questions, providing instant feedback and adaptive practice.
• Illustrates the trend of building narrow, high‑impact AI tools for specific educational tasks.

LTX Studio Audio‑to‑Video

• Feature: Convert a short audio clip (generated with 11 Labs) into a synchronized video with lip‑sync.
• Performance: Works intermittently; when successful, the result is impressive.
• Implication: AI‑driven media generation is becoming more accessible, though reliability still varies.

11 Labs Album Review

• The AI‑generated album was described as “average” – competent but lacking soul or originality.
• Highlights a broader observation: AI can produce passable content quickly, but true artistic excellence remains elusive.

Creative AI in Animation (Google Director Case Study)

• A director used Gemini to transform hand‑drawn sketches into an animated short, achieving a unique visual style without the cost of fully hand‑drawn frames.
• Demonstrates how AI can augment, not replace, human creativity, enabling artists to focus on higher‑level design while AI handles repetitive rendering tasks.

Final Thoughts

• The week showcases a dual movement: specialized AI apps for narrow tasks and a push toward a universal assistant that can orchestrate those apps.
• As integration improves, the line between “tool” and “agent” blurs, bringing us closer to the AGI vision.

The rapid rollout of integrated connectors, open‑source bots, and specialized AI apps signals that we are converging on a single, versatile digital assistant capable of handling everything from spreadsheet automation to creative animation—bringing the promise of AGI‑level functionality ever closer to everyday users.