AI AssistNews
Latest updates and events from AI Assist
Google Cloud: 2026 is the year of AI agents
Google Cloud released its AI agent trends report: the key shift is from prompts to autonomous agents that run end-to-end business processes on their own.
Read more →
Business runs 12 AI agents, half work in isolation
Businesses run about 12 AI agents on average, yet roughly half of them work in isolation. The value is in connecting the agents, not in their number.
Read more →
Google ships Gemini Flash for production agents
In its July update Google introduced Gemini 3.6 Flash plus Flash-Lite and Cyber for production agents, and opened AlphaEvolve to all Cloud users.
Read more →
DeepSeek V4: AI models now ship like patches
DeepSeek-V4-Flash landed on July 31 and the trend is now set: AI models update like patches. Choose one by job fit and cost, not by leaderboard rank.
Read more →
NVIDIA releases open Isaac GR00T models for robots
The new open models let robots understand natural-language commands and perform complex, multi-step tasks.
Read more →
Anthropic releases Claude Sonnet 5 — its most agentic model yet
Claude Sonnet 5 became the default in Claude Code: a native 1M-token context window and the ability to run as an autonomous agent.
Read more →
xAI moves Grok 4.5 into private beta
Grok 4.5 is based on the 1.5T V9 foundation model and is roughly 50% larger in scale than Grok 4.4.
Read more →
Google pushes Gemini 3.5 Pro release to July
The release slipped from June: Google is refining the model over feedback about excessive token use in agentic tasks.
Read more →
Microsoft launches seven in-house MAI models
Microsoft AI introduced a family of seven new in-house models, strengthening its own model lineup.
Read more →
We launched an AI sales manager for the security agency Kaskad
For the security agency Kaskad (Kharkiv) we implemented an automated sales manager that handles inquiries around the clock and passes hot leads to the sales team.
Read more →
Microsoft and Google challenge OpenAI and Anthropic in AI coding
Microsoft and Google introduced their own coding models, intensifying competition with Anthropic and OpenAI.
Read more →
Anthropic releases Claude Opus 4.8, leader in independent benchmarks
Anthropic’s new model topped the Artificial Analysis Intelligence Index, surpassing previous leaders.
Read more →
Google released Gemini 3.1 with up to a 2M-token context
Google’s new model works natively with text, images, audio, and video and extends the context window to 2 million tokens.
Read more →
AI for automated tour selection for the company Ibis
For the travel company Ibis we built an AI that selects suitable tours on the customer’s request and advises in messengers.
Read more →
DeepSeek released its V4 model
DeepSeek V4 ranked among the strongest models of May 2026, alongside GPT-5.5, Claude, and Gemini.
Read more →
Over 500 AI models are now available on the market
The AI model race is accelerating: over 500 models are available across commercial APIs and open source.
Read more →
OpenAI released GPT-5.5
OpenAI introduced GPT-5.5 just six weeks after GPT-5.4, focusing on parallel reasoning and an updated Codex.
Read more →
AI secretary for the director of the East-West Foundation NGO
For the East-West Foundation NGO we built an AI secretary that manages the director’s calendar, reminds about tasks, and helps with correspondence.
Read more →
Google introduced the Gemma 4 open models
Google’s new Gemma 4 open-model series is built for advanced reasoning and agentic workflows.
Read more →
Startup funding hit a record on the AI boom: $300B in a quarter
In Q1 2026 investors poured around $300B into startups on the wave of the AI boom.
Read more →
The industry shifts toward “physical AI” and autonomous robots
The industry’s focus is moving from static automation to “physical AI” — systems capable of reasoning and perceiving their environment.
Read more →
A roundup of the biggest AI events of March–April 2026
A concise overview of the key AI releases, deals, and trends of March–April 2026.
Read more →
Anthropic made Claude’s 1M-token context generally available
A 1M-token context window for Claude became generally available with no long-context surcharge.
Read more →