The Lindahl Letter
The Lindahl Letter
Enabling automated agent actions
0:00
Current time: 0:00 / Total time: -3:52
-3:52

Enabling automated agent actions

We all thought you would be able to easily ask the house to turn on all the lights or command your television by voice to do things by walking into the room. Some of us thought the entire wall would be a television screen by now and that has not happened either. However, some of the new 100 inch TVs on the market are really large. Enabling automated actions is what is happening within the latest development kits related to the companies making and contributing to LLMs. We have seen Google teams introduce low stakes automated actions like making dinner reservations or screening calls. Those types of training activities help them build and reinforce automations without really doing anything particularly risky. It’s all about the ecosystems and when Google teams start to really deeply allow an assistant based agent to do things that are deeply integrated at that point things are going to rapidly change. That is when your agent will be empowered to the point of being able to really automate some things that will be impactful. Sam Altman of OpenAI has said that 2025 will be the year that we will see agents working effectively [1]. I’m guessing that Sam has spent some time thinking deeply about what these agents are going to be capable of doing. 

We will probably start to see Google calendar automations where meetings with unfulfilled action items automatically get scheduled or task follow ups by chat can come from the agent. This type of recursive review of things that happened where a transcript is recorded and checked against a project plan or calendar is certainly on the roadmap. It’s going to be about bringing the next set of low stakes actions to the business world and calling it revolutionary. A lot of hype is going to occur. Sure systems with robotic process automation or coded workflows have been able to automate things for people willing to invest in those automations. With the advent of agents that are able to schedule automated actions it changes the barrier to entry by fundamentally lowering it. People are probably going to be more willing to trust one of the known major brands with this technology considering that most smartphones have banking information saved and are logged into a myriad of other consequential accounts. Having practical limits on what agents are able to enable in terms of automation remains probably the most important process enablement gate to be considered. Apparently the teams at Google don’t expect to deploy any useful agents until 2025 at the earliest [2].

The push toward true agent-based automation is an ongoing journey. While current tools may seem like small, incremental advances—like handling calendar follow-ups or screening calls—they represent foundational steps toward a more integrated, intuitive digital ecosystem. As AI agents begin to bridge the gap between simple command-driven functions and context-aware actions, we’re stepping into an era where automation isn't just a convenience but a fundamental part of daily life. This gradual transformation will bring more impactful applications, positioning agents not as isolated tools but as active partners in our productivity.

Footnotes:

[1] https://www.tomsguide.com/ai/chatgpt/the-agents-are-coming-openai-confirms-ai-will-work-without-humans-in-2025 

[2] https://techcrunch.com/2024/10/29/google-says-its-next-gen-ai-agents-wont-launch-until-2025-at-the-earliest/ 

If you enjoyed this content, then please take a moment and share it with a friend. If you are new to The Lindahl Letter, then please consider subscribing. Make sure to stay curious, stay informed, and enjoy the week ahead! 

Discussion about this podcast

The Lindahl Letter
The Lindahl Letter
Thoughts about technology (AI/ML) in newsletter form every Friday