AI is quickly moving from optional add-on to core business infrastructure. In just the past two weeks, we’ve seen major updates from OpenAI, Google, and Perplexity, new AI agents rolled out across tools like Salesforce and HubSpot, and fresh regulatory changes that could reshape how companies build and use AI. This blog breaks down what these developments mean in practical terms whether you’re leading strategy, managing data, or supporting teams across marketing, finance, HR, legal, or tech.
OpenAI Taps Google’s TPUs and Goes All-In on o3‑pro OpenAI is making big moves. Its new o3‑pro model now powers ChatGPT Pro and Team plans—offering stronger reasoning, better planning, and fewer major errors than GPT-4 Turbo. Behind the scenes, OpenAI has also started using Google Cloud’s TPUs to reduce dependence on Nvidia and lower compute costs. View Details
Google’s Gemini Startup Kit Pushes AI to the Edge of Innovation Google has launched its Gemini AI Kit for startups offering free Cloud credits, technical support, and access to Gemini’s latest multimodal models. It’s designed to help early stage founders embed AI from the start and keep them inside Google’s ecosystem as they scale. View Details
Perplexity Brings Real Time Stock Data Into AI Search Perplexity AI just added live financial data to its Labs tier including real-time stock prices and company level trading info. The update transforms the tool into a powerful assistant for finance teams, combining conversational search with hard market data. View Details
Meta’s Chatbots Want to Talk First Leaked docs reveal Meta is testing proactive chatbots, AI agents that initiate conversations with users to increase engagement. It’s part of a broader push to personalise AI tools and embed them more deeply into daily digital life. View Details / More on Meta’s chatbot plans
Autonomous agents arriving in your CRM
Salesforce Says AI Now Does 50% of Its Work Salesforce’s new Agentforce 3 platform includes a command centre, secure tool integrations, and human-agent collaboration features. Inside the company, digital agents already perform up to half the work across engineering, support, and sales ops. View Details
Sakana’s Multi-Agent AI Shows 30% Performance Boost Sakana AI has developed a system of collaborating models like a team of AIs that outperform single model systems by 30%. Inspired by natural swarms, their “Treequest” approach hints at the next evolution in LLM design. View Details
HubSpot Expands AI Agent Access Across All Core Hubs HubSpot’s Customer Agent is now available to all Pro and Enterprise users across Marketing, Sales, Service, Content, and Operations Hubs. This marks a major expansion in how AI can support customer interactions, helping teams respond faster and reduce manual effort across the entire customer journey. View Details
LinkedIn Launches AI-Assisted Talent Search to Speed Up Hiring LinkedIn has rolled out a new AI-powered candidate search tool aimed at making recruitment faster and more accurate. The advanced assistant refines job matches, surfaces top candidates based on skills and intent, and helps hiring teams reduce time-to-hire without sacrificing quality. View Details
Navigating the evolving AI rulebook
EU AI Act Takes Effect, Transparency Rules Now Apply The EU AI Act is now live, with general-purpose AI models subject to new obligations around safety, risk, and transparency. A public consultation is shaping which AI use cases will be labelled “high-risk” including many in HR, finance, and legal. View Details
UK Updates GDPR with New Data Use and Access Act The UK’s new Data Use and Access Act simplifies how AI systems can use data while strengthening privacy safeguards. HR, legal and compliance teams should review how data is sourced, stored, and audited across AI projects. View Details
EU’s AI Code of Practice Delayed, but Guidance Is Coming The voluntary Code of Practice for general-purpose AI has been delayed until later this summer, but regulators stress that documentation, governance, and risk mitigation should start now. View Details
Denmark Moves to Protect Identity Rights from Deepfakes
Denmark has announced plans to give individuals automatic copyright over their own faces, voices, and personal likeness in response to growing concerns around AI generated deepfakes. If passed, the law would make it illegal to use someone’s features in synthetic media without consent. View Details
Key takeaways for your team
- Marketing & Sales: AI agents in CRMs are here. Clean data and clear task boundaries will unlock the best results.
- Finance: Real-time data integrations, like those from Perplexity, are making AI tools more useful for daily decision-making.
- HR & Legal: Transparency in hiring, risk scoring, and audits is becoming a regulatory requirement. Prepare now.
- Tech & Ops: Platform shifts (like OpenAI using TPUs) signal changing infrastructure trends. It’s time to reassess your stack!
- Leadership: From defence contracts to data law, AI is strategic. Build internal AI policy and training before it’s required.
Further Reading & Sources
- OpenAI starts using Google TPUs – Reuters
- Inside o3-pro and tool use – OpenAI
- OpenAI’s government & defence contracts – The Verge
- Salesforce Agentforce rollout – CMSWire
- EU AI Act regulatory updates – Osborne Clarke
- UK Data Use and Access Act – National Law Review
GenFutures Lab helps organisations navigate AI adoption with clarity, compliance, and creativity. Need tailored training, benchmarking, or policy guidance? Contact us to get started: https://genfutures.co.uk/contact-us/
