Thinking like a manager
David Peterson

Forget technical skills; in the age of AI, management skills are the new superpower.

I was speaking with a friend recently about their (somewhat painful) transition from individual contributor to manager, and all the little things they needed to improve on to be a better manager. The conversation sparked an interesting observation: in the AI era, the skills that make you a great manager will make you a fantastic individual contributor.

Here’s what I mean. When I was starting out at Google, a request from a client for some sort of deep-dive analysis on their advertising performance would send me down a day-long spiral of data cleaning, Excel pivot-ing, chart formatting and memo writing. Now? I’m pretty sure I could spend five minutes instructing AI to process the data, another ten reviewing and providing feedback on the output, and then the balance on thoughtfully analyzing the results and drafting a memo.

This shift “up the value chain” has happened many times before. Architects used to spend countless hours at the drafting table creating detailed manual drawings. When CAD software arrived in the 1980s, it didn't just speed up drawing – it fundamentally changed the scale and scope of what architects could achieve. Suddenly, they could rapidly explore multiple design iterations, simulate structural loads, and manage vastly more complex projects. The architect's value shifted from manual precision to creative vision and system-level thinking. The spreadsheet triggered a similar transformation in finance, enabling accountants to shift from crunching numbers to providing strategic insights.

But the AI revolution feels different, and I think it's because of three things:

First, it's universal. Unlike previous tools that transformed specific industries, AI is hitting every knowledge worker simultaneously. We’re all having our "spreadsheet moment" at once.

Second, it's cognitive. Previous tools mainly automated manual or computational tasks. AI can handle tasks we once thought were uniquely human – writing, analysis, problem-solving, even creative work.

Third, it's adaptive. A spreadsheet does what it's programmed to do. AI tools learn and improve based on feedback. The relationship between human and tool is more dynamic, more collaborative than ever before.

And that brings us to my conversation with my newly promoted friend…the skills that make someone an effective manager of people are becoming crucial for individual success with AI. A good manager doesn't do all the work themselves (even if they could) – they break down work, delegate effectively, and quality-check output. Similarly, effective AI users don't just throw vague prompts at ChatGPT and hope for the best. They break problems into manageable chunks, provide clear context, give feedback, and carefully evaluate results.

The immediate implications seem obvious enough. There's a clear need for tools that help individuals manage and coordinate multiple AI agents, as well as quality-control systems that can help individuals verify outputs and refine inputs at scale. And, indeed, there are countless “agent orchestration platforms” out there seeking to solve exactly these sorts of problems.

But the second-order effects are, to my mind, more intriguing. An individual armed with AI tools will soon be able to orchestrate dozens of simultaneous workflows, each operating at machine speed. Imagine a world where everyone commands a team of tireless digital agents, capable of executing tasks with superhuman efficiency? In this reality, the bottleneck won't be technical skills or domain expertise, but the distinctly human ability to think strategically, design complex systems, and manage the interplay between human and artificial intelligence.

This raises profound questions about the future of work and the skills we'll need to thrive. And probably suggests some big opportunities for entrepreneurs as well. I’ve got some ideas of my own (like how the ability to decompose work into AI-manageable chunks will become a critical skill, and can maybe also be done by AI?), but if this is an area you’re working on, I’d love to hear from you!

David

Am I Thinking About AI the Right Way?
Gil shares the four AI themes and questions he's thinking about.

The Venture Apocalypse
The venture world is deeply debating its future, but core principles remain unchanged.

No Sleepwalk to Success
Engineering success in a technical startup.

Revenue Durability in the LLM World
Everything about LLMs seems to make revenue durability more challenging than ever.

More blogs

ENTERPRISE/TECH NEWS

Search wars and agent wars. There were two major developments in the AI world this weekend. First, OpenAI launched a search product designed to compete with Google search. Second, both Anthropic (Claude) and Google announced agents that will allow users to automate agentic tasks on computers in a way that mimics human interaction with the computer. Claude’s product is called “computer use” and is in beta. “Claude has the ability to “see” a screen via screenshots, adapt to different tasks and move across workflows and software programs. It can also navigate between multiple screens, apps and tabs, open applications, move cursors, tap buttons and type text.” Google’s product, codenamed Jarvis, is expected in December. According to the Information, “Anthropic and Google are trying to take the agent concept a step further with software that interacts directly with a person’s computer or browser. OpenAI also has been developing similar software for the better part of the year. Google’s agent, similar to the one Anthropic launched, responds to a person’s commands by capturing frequent screenshots of what’s on their computer screen, and interpreting the shots before taking actions like clicking on a button or typing into a text field, two of the people said. There are key differences between the two companies’ agents. Anthropic has said its product can operate different applications installed on a person’s computer, while Jarvis can only operate a web browser and has been tailored to Google’s Chrome browser, the two people said.”

Towards systems of agents. Jay Gupta of Foundation Capital wrote on X and in a great blog post about “systems of agents” as the next great organizing principle of software. “Enter Systems of Agents: AI-powered entities that don't just assist - they act. They parse emails, understand calls, process documents, and most importantly, take autonomous action. They turn unstructured business communication into structured, actionable intelligence.“

Apply researchers call the capabilities LLMs into question. Ignacio de Gregorio reported that “a group of Apple researchers has published a paper claiming that large language models (LLMs), the backbone of some of AI's most popular products today, like ChatGPT or Llama, can’t genuinely reason, meaning their intelligence claims are highly overstated.” He continues: “I take issue with the fact that instead of LLM proponents claiming to be building AGI and having to prove that is the case, it seems like AI academia is being forced to prove otherwise. Shouldn’t it be the wrong way around? Shouldn’t we be of the opinion that, unless proven otherwise, LLMs are not the solution? Clearly, the reason for this is none other than the abhorrent amounts of capital being invested into this vision being true. However, as we’ve seen today, there are many more reasons to be skeptical about LLM’s reasoning capabilities than simply believing that scale will do the trick.”

Gartner’s top seven trends for CIOs. Gartner listed Agentic AI and AI governance as two of the top seven trends CIOs should be watching for in 2025. It was also interesting to see disinformation security (perhaps a nod to deepfake social engineering?) and post-quantum cryptography make the list as well.

The unstructured data stack. Our friend Astasia Myers of Felicis outlined the unstructured data stack, which we feel is a sophisticated and more future-proof way of framing the evolving data landscape. She writes that “as enthusiasm for GenAI has increased, so too has the understanding that its success relies on strategically utilizing an organization’s unstructured data. Unstructured data powers training and fine-tuning GenAI models, Retrieval-Augmented Generation (RAG) search and AI agents, and contextual analytics. With the rise of GenAI use cases and the explosion of unstructured data itself, we believe unstructured data is entering a golden age and will be used considerably more in the future. Unstructured data has become a veritable gold mine of information and a significant focus for organizations seeking to leverage it for strategic advantage. Unstructured data’s attributes means it doesn’t fit well within the traditional data infrastructure stacks. A new unstructured data stack is emerging and consists of three crucial components: data extraction and ingestion, data processing, and data management. Each part plays a vital role in deriving value from unstructured data in the age of AI.‍” (Kudo’s to FalkorDB for making the list!)

HOW TO STARTUP

The dangers of overcapitalization. The BG2 pod is essential listening pretty much every time it comes out, but the last episode contains a detailed discussion of why overcapitalization is so dangerous for founders. The key discussion starts about twelve minutes in, and is a must-listen for any founder (or VC) at any stage.

Pricing the work. Kyle Poyar, top SaaS pricing expert, wrote about some of the emerging (and disruptive) pricing models for AI-first software businesses. The key point is that “We're moving away from charging for access to software and toward a model of charging for the work delivered by a combination of software and AI agents.” In short, it’s going to get really complicated. “Folks are increasingly selling units of work completed rather than selling access to the software (seat licenses) or consumption of the software (usage). With the software and the service associated with the software now bundled together, there’s the potential for a lower total cost of ownership (TCO) for the customer along with greater pricing power for the vendor…Shifting to these newer value-based pricing models isn't a simple pricing change you can just announce in a press release. It's a business model evolution that looks a lot like the shift from on-prem to SaaS in the first place. Things are about to get interesting. These new AI pricing models might mean greater volatility in both usage and spend. Variable margin profiles across products and customers. Seasonal revenue fluctuations. The potential for project-based, non-recurring use cases.Put simply, annual recurring revenue (ARR) continues to get dethroned. The new “ARR” has an extra “R”: annual revenue run-rate or ARRR.”

The Wiz Story. Sequoia published a history of Wiz’s rapid ascent. The story highlights the importance of team over idea: “The founders started hiring, getting even more of the old gang back together from Adallom and Microsoft to become their earliest “Wizards.” Like a reverse “Field of Dreams,” they were again less concerned with building a specific thing than with surrounding themselves with the right team with whom to build.” Another, related, theme is the importance of building culture through intense in-person work: “Space constraints—the team was operating out of a two-room office—meant that everyone, regardless of job function or title, worked out of the larger room, reserving the smaller room for ongoing CISO conversations and sales calls. This put product people right next to their engineers, gave the engineers insight into the sales process, allowed sales to easily rope technical specialists in on client calls, and put leadership in dialogue with their newest hires. Transparency and interdepartmental context was built into the nascent company’s DNA.” Finally, the significance of the company’s agentless technology for their rapid growth can not be overstated: “Best of all—at least from a sales perspective—when potential customers provided Wiz with read-only access to their cloud infrastructure, Wiz’s product had the ability to crawl their infrastructure in real time, providing a window into security exposures and misconfigurations and how they were all connected. “In 15 minutes, it basically started providing value for customers, showing them results that would kind of light up like a Christmas tree, making very clear all the things that are wrong in their cloud infrastructure,” says Balkansky. “This made for a very sexy demoable product.” A sexy, demoable product coupled with pioneering technology and an indefatigable team put Wiz on a surefooted path to success.”

HOW TO VENTURE

The Cyberstarts story. Forbes published a detailed look at how Cyberstarts’ CISO “sunrise” program operated and an investigation of claims of quid-pro-quo kickbacks for helping portfolio startups.

Back to basics. Patty Wexler of Avila VC wrote an excellent updated manifesto on why she believe venture capital should favor the smaller more artisanal funds. “The power law is real, but without discipline it will not deliver sustainable returns. And with a giant bucket of 2% fees, the impetus to work insatiably for the 20% might be muted.” We wholeheartedly agree.

Portchain announced Montreal Gateway Terminals will join the Portchain Connect network. Montreal Gateway Terminals will use Portchain Connect to increase the quality and speed of their berth alignment with customers through digital handshakes and secure data sharing.

Viably and Airwallex have partnered on a solution that enables eCommerce businesses to simplify cross-border transactions and manage multiple currencies in a single account.

Aquant
Senior Sales Engineer (Boston)

Groundcover
Developer Marketing Manager (Tel Aviv)

Planable
SEO Specialist (Bucharest)

Tensorleap
Algorithm Developer (Tel Aviv)

More jobs