🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

This seems like a seminal moment in advertising. Part of the intrigue is its novelty, which will wear off. But it’s clear that in 5 years from now, the creative field will be transformed. Counterintuitively, this will also mean advertising grows, not shrinks.

We’re entering a new paradigm of software that is all about managing AI Agent execution. This is going to show up in many novel ways, but ultimately will mean tools designed to help you plan, orchestrate, review, and incorporate the work of AI Agents.

One novel way that AI Agents will change how we work is that they can execute many attempts at the same problem, and you can compare options. Before this, usually you’re stuck on a path given all the preceding work. Now with AI Agents, you’ll just test endless approaches.

The idea of writing software on your phone would have been comical 2 years ago. Now with AI Agents this is a reasonable form factor. Because the interaction model of AI Agents is heavily task oriented, you’ll be able to do almost any type of complex work from anywhere.

Multi-agent systems were found to use 15X more tokens than regular AI chat. This is Jevons paradox. Any efficiency improvement in AI models will quickly get eaten up by doing far more complex work with AI Agents. We’re only at the start of this curve right now.

Despite what makes the headlines, most companies I talk to are focused on using AI Agents to do more than they did before vs. doing the same and spending less. Usually it’s to reduce busy work, generate more revenue, build products faster, or serve customers better.

Getting AI Agents to work extremely well for complex enterprise use cases is non-trivial. If you’re building agents, your moat will directly correlate to the amount of software you have to build on top of the AI models to execute the task. The harder the problem the better.

The most important factor for AI Agents is to get them the context necessary to execute the task successfully. No matter how powerful AI models get, context will always be king. Data, workflows, tools, domain knowledge, and tuned instructions will all be critical.

Nailing the right AI Agent UX right now is critical. Requiring too much human interventions won’t deliver enough value. Handing off all the work without user validation doesn’t quite work yet. Build UX patterns that live in the middle, but that can evolve to full autonomy.

It’s possible to imagine a future where AI Agent pricing converges with the cost of compute. However, AI Agents require significant domain knowledge, customer data, workflows, and a tuned UX to make them effective. This then would lead to software-like margins. Will see.

AI Agents will likely follow a self driving trajectory, where you need a human in the loop for a long tail of tasks for a while. The big difference is we’ll get a growing number of autonomous agents along the way, where full self driving is an all or nothing proposition.

AI Agents create a major TAM expansion in software because you can bring automation to many industries for the first time. Verticals dominated by unstructured data and workflows like legal, healthcare, consulting, and major parts of finance can now actually go digital.

It’s crazy how much AI Agents are going to change the core premise of software. Since the beginning of the industry, software has been something that enables people to do their work. We’ve spent decades building features that make people incrementally more productive when using

Companies going AI-first should dedicate some talent that knows what AI is capable of to be in the trenches to design next gen workflows. AI moves so fast it’s hard to decentralize this knowledge yet. Huge opportunity for engineers and IT folks that jump on this early.

A big part of software in the future will be tools that maximize the user’s ability to manage AI Agents and their tasks. No matter how good AI Agents get, there’s going to be a long tail of work required to plan, review, orchestrate, and incorporate what they deliver.

The reason we’re so excited about AI at Box is because AI Agents thrive on unstructured data. Most companies are sitting on a goldmine of intelligence that they’ve never been able to fully tap into, and now can.

We just released major AI updates at Box: an enhanced data extraction AI Agent for better accuracy on complicated documents, the beta of our remote MCP server to integrate with any AI system, and major updates to the Box AI Studio for building custom agents in Box.

OpenAI is expanding the access to Box within ChatGPT. To get the promise of AI, we need interoperability between our systems so AI Agents can use any tools and data sources to execute their work. This is what the future will look like.

AI Agents having the right context to execute their tasks is the defining factor for agent success. Getting the deepest domain understanding, task instructions, tool use, and the right chunks of corporate knowledge are some of the most important problems in software now.