AI biweekly recap by Fellow – AI agents and security

In partnership with

Welcome to the second issue of the AI Biweekly Recap by Fellow. In this issue, we explore how AI agents are being integrated into companies, their impact on productivity, and the latest developments in the space. 

Spotlight: AI Agents 🔦

First, what is an AI agent?

Agentic AI – short for autonomous generative AI agents – is software designed to tackle complex tasks and hit targets with minimal (or even zero) human oversight. That sets it apart from today’s chatbots and co-pilots because those typically aren’t self-driven in the same way.

Deloitte forecasts that in 2025, 25% of companies that use gen AI will launch agentic AI pilots capable of handling complex tasks with little human oversight, growing to 50% of companies launching AI agents by 2027. 

AI agents are no longer just a concept; they are actively transforming the way businesses operate. From No-code Agent Builders, to Research Agents, Customer Service Agents, AI Meeting Assistants, and Go-To-Market Copilots, these systems are starting to make an impact in the productivity of companies that are fast and forward-thinking enough to adopt them. 

As we approach this shift, understanding AI agents, their potential impact, and the security risks involved will be crucial for tech leaders looking to stay ahead of the curve. 

3 steps to prepare your organization for AI agents ✍️

1. Identify suitable tasks: Consider which tasks and workflows in your organization are well-suited for agentic AI to execute.
   

2. Strengthen security measures: Prioritize security and compliance when implementing an agent, as they will probably have access to valuable and potentially sensitive enterprise data, as well as internal systems.
   

3. Pilot low-risk, supervised implementations: Start with low-risk use cases and non-critical data, and lots of human oversight. Remember that in multiagent systems, “hallucinations” can spread from one agent to another; they can persuade other agents to take the wrong steps and give incorrect answers.

Resource: AI meeting assistant security checklist 🔐

As an AI meeting assistant, we know firsthand the productivity boost these tools can bring, but we also understand the security challenges that come with them. To help organizations implement AI meeting assistants securely, we've put together a comprehensive AI Meeting Assistant Security Checklist for Executives, IT, and HR teams.

Use this resource to ensure your organization stays protected while unlocking the full potential of AI-powered meetings.

📥 Download the checklist to learn:

• Key security considerations when implementing an AI meeting assistant.

• Best practices for data protection and compliance.

• How to evaluate AI vendors to ensure your data isn’t shared with insecure meeting bots.

This week in AI 📰

1) Google Meet AI Note Taker Review

We won’t bury the lead — while Gemini does deliver a summary and transcript and works well with Google Meet, it simply doesn’t measure up to purpose-built AI note takers. From inaccuracies to sluggish emails to limited language support, it’s clear that Gemini is a basic tool that lacks features like meeting agendas, pre-meeting briefs, and integrations. Plus, if you’re concerned about security and privacy, Gemini may not be the pick for you.

Read this review to learn about how Google Gemini’s “Takes Notes for Me” feature works, its limitations, and what is a better choice for your team.

2) OpenAI releases Operator

OpenAI’s new Operator is an exciting step toward the AI-powered future we’ve all been anticipating. Designed to act as a fully autonomous AI agent, Operator can handle tasks with minimal human oversight. It is currently a research preview only available in the US, meaning it has limitations and will evolve based on user feedback.Stay tuned as we keep an eye on how it develops.

3) Perplexity AI Assistant

Perplexity has also entered the fray with its own assistant for Android, which Perplexity has access to the web. That allows the assistant to do things like creating calendar entries and use your phone’s camera to answer questions about what’s around you or on your screen.

4) DeepSeek

A Chinese AI startup, DeepSeek, has made headlines by introducing a new AI model that challenges the dominance of U.S. companies like OpenAI. This model reportedly operates at a fraction of the cost of its competitors, leading to a notable reaction in the stock market, particularly affecting Nvidia shares due to concerns over AI hardware demand. This development has been described as a "Sputnik moment" for the U.S. tech sector, indicating a shift in global AI development dynamics.

Learn AI in 5 minutes a day

This is the easiest way for a busy person wanting to learn AI in as little time as possible:

1. Sign up for The Rundown AI newsletter

2. They send you 5-minute email updates on the latest AI news and how to use it

3. You learn how to become 2x more productive by leveraging AI

Sign up to start learning.

Final thoughts 💭

As AI agents continue to evolve, their role in the workplace will become even more significant. For tech founders and executives, staying informed and implementing AI responsibly will be key to maintaining a competitive edge.

What are your thoughts on AI agents in business? Hit reply and share your insights!

Until next time,

Manuela and the Fellow team