a16z "Three Major Predictions for AI Agents in 2026": The Disappearance of Input Boxes, Agent-First Usage, The Rise of Voice Agents

```

At the recent “Big Ideas for 2026” online seminar hosted by the well-known venture capital firm Andreessen Horowitz (A16z), their partner team drew a clear blueprint for the evolution of AI technology: Artificial intelligence is evolving from a chat tool waiting for instructions to an “agent” capable of proactively executing tasks.

At the same time, they proposed three major industry-altering “guesses”: user interface interactions will shift from “prompting” to “execution,” product design will shift from “human-centered” to “agent-first,” and voice agents will move from tech demos to large-scale deployment.

Guess One: The Disappearance of the Input Box.

A16z AI applications investment team partner Marc Andrusko boldly predicts, “By 2026, the input box as the main user interface for AI applications will disappear.” He believes that the next generation of AI applications will no longer require users to input complex instructions, but instead will observe user behavior, proactively intervene, and provide actionable plans for review.

Behind this shift is a huge leap in the business value of AI. Andrusko points out, “We used to focus on the global annual software spend of 300 to 400 billion US dollars, but now what excites us is the 13 trillion US dollars of labor spend in the US alone. This expands the software market opportunity by approximately 30 times.”

He compares future AI agents to the top “S-Class employees:” “The most proactive employees identify problems, diagnose root causes, implement solutions, and only then come to you: ‘Please approve the solution I found.’ This is the future of AI applications.”

Guess Two: “Agent-First Usage” and Machine Legibility.

A16z growth investment partner Stephanie Zhang proposed a disruptive design logic: software will no longer be designed for human eyes. She points out: “Things important for human consumption may no longer be so for consumption by agents. The new optimization direction is not visual hierarchy, but machine legibility.”

According to Stephanie, the “5W1H” principles or refined UIs we optimized for human attention in the past will face reconstruction in the agent era. She predicts: “We may see a large amount of ultra-personalized, high-frequency content generated for agent interests—which is like keyword stuffing in the age of agents.”

This shift will profoundly impact all aspects from content creation to software design.

Guess Three: The Rise of Voice Agents.

Meanwhile, A16z AI application investment team partner Olivia Moore observed, “AI voice agents are beginning to gain a foothold.” She stated that voice agents have evolved from sci-fi concepts to systems that real enterprises are now purchasing and deploying at scale. Especially in healthcare, banking/finance, and recruitment, voice agents are highly favored due to their high reliability, strong compliance, and ability to address staffing shortages.

She shared an interesting finding: “In banking and financial services, voice AI actually performs better because humans are very good at violating compliance rules, while voice AI can execute accurately every time.”

Moore emphasized, “AI won’t steal your job, but a person using AI will.” This signals profound changes for traditional call centers and BPO industries, with providers who leverage AI technology to offer lower prices or greater capacity gaining a competitive advantage.

Main points from the online seminar:

The end of UI: The era of the prompt box as the core interactive interface is coming to an end. AI will shift from “passive response” to “active observation and intervention.”30-fold market growth: AI’s target market is shifting from the $400 billion software spend to the $13 trillion labor spend, representing a fundamental commercial leap.S-Class Employee Model: The ideal AI should act like a highly proactive employee: discover problems, diagnose causes, provide solutions, and execute, leaving only the final confirmation to humans.Machine Legibility: Software design goals are shifting from “human-first” to “agent-first.” Visual hierarchical UI will no longer be central.Content creation divergence: Brand competition will shift from attracting human attention to “Generation Engine Optimization” (GEO), with large amounts of “high-frequency content” generated for AI crawling possibly emerging.Compliance advantage of voice AI: In high-threshold industries like finance, voice AI outperforms humans as it can 100% comply with regulations and provides traceable records.Healthcare and government entry: Voice agents are solving high turnover problems in healthcare and are expected to address public services pain points such as 911 calls and DMV in the future.Voice AI industrialization: Voice AI is developing into a full industry rather than a single market. There will be winners at every layer of the value chain—from foundational models to platform-level applications, all representing great opportunities.From tool to “AI employee”: AI is no longer just an auxiliary tool but a digital employee capable of independently handling complete business processes.

Full Transcript of the a16z AI Team Seminar (translated by AI tool):

Host 00:00 Welcome to “Big Ideas for 2026.” We’ll hear from Marc Andrusko on the evolution of AI user interfaces and a fundamental shift in how we interact with intelligent systems. Stephanie Zhang will discuss the implications of designing for agents rather than humans, a shift that’s reshaping product development. Olivia Moore will share her views on the rise of AI voice agents and their growing role in our daily lives. These are not just predictions—they are insights from those working directly with the founders and companies building the future world.

Marc Andrusko 00:31 I’m Marc Andrusko, a partner in our AI application investment team. My big idea for 2026 is the disappearance of the input box as the main user interface for AI applications. The next wave of applications will require far fewer prompts. They will watch what you’re doing and proactively intervene, providing action plans for your review.

Marc Andrusko 00:49 The opportunity we’re going after used to be the $300–$400 billion global software spend each year. Now, what excites us is the $13 trillion labor spend in the US alone. This makes the software market opportunity, or the potential total addressable market (TAM), about 30 times larger. If you start from there and think, okay, if we all want this software to work for us, ideally it needs to be as good as humans or even better, right? So I like to think: what do the best employees do? What do the best human employees do? I’ve been talking a lot lately about a diagram that’s been floating around Twitter. It’s a pyramid of five types of employees and why the most proactive ones are the best. If you start at the base of the pyramid, those are people who notice a problem and come to you for help, asking what to do. That’s the least proactive kind of employee.

But if you rise to the S-Class—the most proactive employee you can have—they will notice a problem, conduct necessary research to diagnose the source, study possible solutions, implement one, and then keep you informed, or come to you at the last moment to say, “Please approve this solution I’ve found.” I think that’s what the future of AI applications will look like. And I think this is what everyone wants. It’s the direction we’re all striving for. So I’m very confident we’re almost there. I think large language models (LLMs) are continuing to get better, faster, cheaper, and I think to some extent, user behavior will still require a human-in-the-loop for final approval, especially in high-risk scenarios. But I think the models are fully capable of coming up with very smart suggestions on your behalf, with you basically just needing to click approve.

Marc Andrusko 02:27 As you guys know, I’m fascinated by the concept of AI-native CRM. I think it’s a perfect example of what these proactive applications could look like. In today’s world, a salesperson might open their CRM, browse open opportunities, look at their calendar, and think, okay, what actions can I take now to make the biggest impact on my sales funnel and close rate? In the future, your AI agent or AI CRM should be continuously doing all these things for you—not just identifying the most obvious opportunities in your funnel, but also digging through your emails from the past two years to find a lead you let go cold that had potential, maybe suggesting we send them this email and pull them back into your pipeline, right? So I think, on drafting emails, organizing calendars, reviewing old call logs, the opportunities are endless.

Marc Andrusko 03:22 Ordinary users will still want that final mile of approval. In almost 100% of cases, they’ll want the “human in the loop” to be the final decision maker. That’s fine.

Marc Andrusko 03:33 I think that’s the way it’s naturally evolving. I can imagine a world where power users invest a lot of extra effort to train whatever AI applications they use to learn as much as possible about their behavior and work style. Those applications will take advantage of larger context windows, leverage memory functions already integrated into many LLMs, so that power users could really trust the app to complete 99.9% or even 100% of the work. They’ll be proud of the number of tasks handled with no human intervention at all.

Stephanie Zhang 04:09 Hi, I’m Stephanie Zhang, an investing partner on A16z’s growth team. My big idea for 2026 is: create for agents, not for humans. One thing I’m really excited about for 2026 is people have to start changing how they create. This covers everything from content creation to application design. People are starting to interact with networks or apps through agents as intermediaries. What’s important for human consumption isn’t the same for agent consumption.

Stephanie Zhang 04:41 When I was in high school, I took journalism. We learned the importance of starting news articles with the 5W1H (who, what, when, where, why, how) in the lede and feature stories with an anecdote at the beginning. Why? To grab human attention—humans might miss a deep and insightful statement buried deep on page 5, but agents won’t.

Stephanie Zhang 05:02 For years, we’ve optimized for predictable human behavior. You want to be among the top Google search results? You want to be one of the first products listed on Amazon. This optimization applies to the web and to software we design. Apps are designed for human eyes and clicks. Designers optimize for good UI and intuitive flows. But as agent usage rises, the importance of visual design for overall understanding will decline. Previously, when something happened, engineers would go into their Grafana dashboards and try to piece together what occurred. Now, AI SREs (site reliability engineers) take in telemetry data, analyze it, and directly message hypotheses and insights in Slack for humans to read.

Before, sales teams had to click through and browse Salesforce or other CRMs for information. Now, agents fetch and summarize the insights. We’re no longer designing for humans but for agents. The new optimization criterion isn’t visual hierarchy but machine legibility. This will change how we create and the tools we use. What are agents looking for? That’s a question we don’t know the answer to. But what we do know is that agents read the full text of an article much better than humans, who might only read the first few paragraphs. There are lots of tools and organizations out there using them to make sure that when consumers prompt ChatGPT and ask for the best company credit card or best shoes, they show up. So there’s lots of what’s called SEO (editor’s note: or similar concepts) tools being used, but everyone’s asking one question: what do AI agents want to see?

Stephanie Zhang 06:43 I love this question about when humans may fully leave the loop. We’ve already seen this happening in some cases. Our portfolio company Dekagon is autonomously answering many client questions. But in other cases, like security operations or incident resolution, we often see more humans in the loop, with AI agents first trying to figure out what’s wrong, analyze, and provide scenarios to humans. These tend to be cases with greater responsibility and complex analysis, so humans stay in the loop. And they may stay there longer until the models and tech reach extremely high accuracy.

Stephanie Zhang 07:33 I don’t know if agents will watch Instagram Reels. It’s really interesting—at least technically, optimizing for machine legibility, insight, and relevance becomes very important, especially contrasted with the past, which put more emphasis on flashy ways to grab attention. We’ve already seen many examples of massive, ultra-personalized content—for instance, maybe you’re not creating an extremely insightful article, but massive amounts of low-quality content for lots of different things you think agents might be looking for. It’s almost the agent era’s equivalent of keyword stuffing: content creation cost is nearly zero, and creating huge amounts of content becomes very easy. This is the potential risk of massive content generation for agents’ attention.

Olivia Moore 08:48 I’m Olivia Moore, a partner in our AI applications investment team. My big idea for 2026 is that AI voice agents will start to take a spot. In 2025, we’ve seen voice agents move from a sci-fi idea to systems that real businesses are buying and deploying at scale. I’m excited to see voice agent platforms expanding—working across platforms and modalities to handle complete tasks, getting us closer to the true AI employee vision. We’ve already seen enterprise customers in almost every vertical at least piloting, if not deploying, voice agents at considerable scale.

Olivia Moore 09:25 Healthcare might be the biggest area here. We see voice agents appear in nearly every part of the healthcare stack—from calls to insurance, pharmacies, providers, even, surprisingly, patient-facing calls. This could be basics like scheduling and reminders, but also sensitive calls like post-op follow-ups or initial psychiatry intakes, all handled by voice AI. Honestly, I think a key driver is current high staff attrition and recruiting difficulty in healthcare, making voice agents a pretty good solution for reliably completing tasks. Another similar category is banking and financial services. You might guess compliance and regulation are too high for voice AI to succeed, but it turns out voice AI actually performs better there, because humans are very good at violating compliance and processes, and voice AI gets it right every time. Plus, you can track voice AI’s performance over time. Finally, another breakthrough area is recruitment—from retail front-line jobs to entry-level engineering and even mid-level consultancy, you can create a candidate experience where they can interview any time, instantly, and be routed into the rest of the human recruiting flow.

Olivia Moore 10:46 As base models improve, we’ve seen huge improvements this year in accuracy and latency. In fact, in some cases, I’ve heard voice agent companies deliberately slow agents down or add background noise to make them sound more human. For BPOs and call centers, I think some will have smoother transitions; others may face a steep cliff with the threat from AI, especially voice AI. It’s like they say: AI won’t take your job, but someone using AI will.

Olivia Moore 11:16 We’re seeing that many end customers may still want to buy a solution, not a technology they must implement themselves, so in the short to mid term, they’ll still use call centers or BPOs, but may choose those who can offer lower prices or higher capacity via AI. Interestingly, in some places, on a per-employee basis, humans are still cheaper than top-notch voice AI. So as models get better, will costs drop and will call centers in those markets face more threat than they do now—it’ll be interesting to watch.

Olivia Moore 11:50 AI is actually very good at multilingual conversation and strong accents. Often, in meetings, I may miss a word or phrase, but my (Granola) transcript captures it perfectly. That’s a great example of what ASR or voice-to-text vendors can do.

Olivia Moore 12:08 I hope to see more use cases next year—anything government-related. We’re investors in Prepared 911—they handle non-emergency calls—but if you can handle 911 calls with voice AI, you should be able to handle DMV or any government service line calls, which today are very frustrating for both customers and staff on the other end.

Olivia Moore 12:32 I’d also love to see more consumer voice AI. So far, it’s mostly B2B, since replacing or augmenting phone agents with much cheaper AI is so clear-cut. In consumer voice, I’m excited about things in health and wellness. We’ve seen voice companions rising in assisted living, nursing homes, serving as both companions and trackers of various health metrics over time. We view voice AI as an industry, not just a market, so there will be winners at every layer of the tech stack. If you’re interested in voice AI or want to start a company in that field, I’d suggest checking out the models—there are great platforms like 11 Labs where you can test creating your own voice and voice agents. You’ll get a good sense of what’s possible and what’s coming next.

Note: This translation cannot guarantee 100% accuracy.

Risk Warning and DisclaimerThe market has risks; investments should be cautious. This article does not constitute individual investment advice nor does it consider the special investment goals, financial situations, or needs of particular users. Users should consider whether any opinions, views, or conclusions in this article fit their specific situations. If you invest based on this, you do so at your own risk. ```