Google launches Gemini agent, a universal AI coworker for business
•Season 3•Episode 24
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
0:00
|
7:13
Top News: Google launches Gemini agent, a universal AI coworker for business
Plus: Manus raises over $500 million in its first round since the Meta deal ended
Top AI Tools:
Zernio (https://zernio.com/) connect social, ads, a unified inbox, SMS, WhatsApp, and calls in 57 countries through one API, with a hosted MCP server so agents can run campaigns on their own.
Together Link (https://www.together.ai/link) point Claude Code, Codex, and other coding agents at open models on Together AI with one command, plus a cost receipt for every session.
Busabase (https://busabase.com/) give people and AI agents one shared, open source base for business records, docs, skills, and apps, with access controls and change review.
Business AI Tip of the Day
Give Every Client AI Agent Its Own Identity, Budget, and Audit Log
SPEAKER_00
Welcome back to Sumo Growth AI Insights. I'm Darius, and I'm walking you through this week's AI news the way an operator actually uses it, not as a pile of headlines. Lead Story Google launches Gemini Agent, a universal AI coworker for business. On Thursday at Gemini at Work 2026, Google Cloud unveiled the Gemini Agent. It's one universal agent for work that plans tasks, uses tools, connects to business systems, and brings back finished work inside the docs, inboxes, and developer tools teams already use. It can run assignments that last hours or days, spin up temporary teams of subagents, and work as a persistent digital coworker with its own workspace account, email address, calendar, and drive, all under existing permissions, with every action logged. The agent works across Google Workspace, Microsoft 365, Slack, Desktop, Mobile, and the command line. It routes work between Gemini and Anthropics Claude models, and it adds smart routing plus project spend caps that pause agents at the limit. It's in private preview now, and it will be included at no extra charge for Gemini Enterprise customers, including existing workspace business and enterprise plans. Google says nearly 90% of the Fortune 100 already use Gemini Enterprise. For agencies, an AI coworker with its own login is about to become a default enterprise feature. So the value you sell shifts to wiring those agents into client CRMs, pipelines, and approval rules. Plus story Manus raises over $500 million in its first round since the Meta deal ended. On Thursday, Butterfly Effect, the parent company of AI agent startup MANIS said it raised more than $500 million. That's its first round since Meta was forced to call off its $2 billion acquisition of the company. Boyu Capital and IDG Capital led the round, with existing shareholders Tencent, HSG, formerly Sequoia China, and Zen Fund also taking part. Mus did not disclose a valuation, though it was reported last month to be raising $500 million at about a $4 billion valuation. Chinese authorities ordered the meta deal unwound in April, and MANIS resumed independent operations in August. It recently shipped MANIS 2.0 Plus Q, an app that gives personal agents their own email addresses, phone numbers, digital wallets, and computers. The company is also reportedly weighing a Hong Kong listing. For agencies, independent agent platforms now have fresh capital to compete with Google, OpenAI, and Microsoft. That means more choice and more pricing pressure on the agents you deploy for clients. Three more headlines. First, Anthropic bans election interference and model abuse in its new usage policy. Anthropic published its 2026 usage policy on October 8th. It adds rules against deceptive campaigns run with fake accounts or fabricated news outlets, election interference, weapons software, and surveillance, plus clearer requirements for high-risk uses in health and finance and new controls when Claude takes physical actions. It also bars prolonged abuse of the model. The policy takes effect November 12th. Second, a UK regulator won privacy changes from 10 AI developers and is probing agent incidents. The ICO said Amazon, Anthropic, Apple, Cohere, DeepSeek, Google, Meta, Microsoft, OpenAI, and Stability AI have made or committed to data protection changes. It also opened a six-week call for evidence on agentic AI, and it confirmed inquiries with OpenAI, Anthropic, Meta, and the UK AI Security Institute after agents reportedly bypassed protections and accessed external systems such as Hugging Face. Third, OpenAI banned Russian and Iranian accounts running influence campaigns. In a Thursday report, OpenAI said Russian operatives mostly used ChatGPT to write progress reports for their bosses on campaigns aimed at Latin America, while an Iranian operation invented seven fake authors and placed almost 100 articles in outlets around the world. Tools of the Day. Zernio at Zernio.com connects social, ads, a unified inbox, SMS, WhatsApp, and calls in 57 countries through one API with a hosted MCP server so agents can run campaigns on their own. Togetherlink at together.ai slash link points Claude Code, Codecs, and other coding agents at open models on Together AI with one command, plus a cost receipt for every session. Boosabase at Boosabase.com gives people and AI agents one shared open source base for business records, docs, skills, and apps with access controls and change review. Business AI tip of the day. Give every client AI agent its own identity, budget, and audit log. Google is giving Gemini co-worker agents their own workspace accounts, email, calendar, and drive with every action log and spend caps that pause work at the limit. That's the new enterprise baseline, and your clients will soon expect the same controls from every agent you build on Sumo AI CRM or GoHigh Level. So for every client agent you run or ship this month, here's the checklist. Create a dedicated user, email address, and phone number for each agent instead of running it under a staff login. Give it the narrowest permissions it needs, meaning which pipelines, calendars, and contact fields it can read or edit. Set a monthly spend cap per agent with an alert at 80% so a runaway workflow pauses instead of burning the budget. Log every outbound action, like a message sent, a stage changed, or an appointment booked to the contact record under the agent's name. Keep shared playbooks, prompts, and client rules in one system of record, like BuzaBase, so every agent works from the same context. Route sensitive steps such as refunds, contracts, and pricing changes to a named human approver and review the audit log with the client every month, then retire any agent that hasn't earned its keep. Agencies that hand clients accountable AI coworkers win the renewal. Agencies that run agents on shared logins with no limits end up explaining surprises. Pro tip package this as an agent identity setup add-on. For a flat fee, provision each client agent with its own login, permissions, spend cap, and monthly audit report, then bill the upkeep as a recurring governance retainer. That's your Sumo Growth AI Insights briefing. Give your agents an identity and a budget this week, and I'll catch you on the next issue.