Claude Code's Creator Says We're Overcomplicating AI Prompts

October 12, 2026


For years, people have been told that getting better results from artificial intelligence requires mastering the art of prompt engineering. Detailed instructions, carefully structured requests and elaborate step-by-step directions have become common recommendations for anyone trying to use AI effectively.


But Boris Cherny, the creator of Claude Code at Anthropic, is suggesting that the approach may be getting unnecessarily complicated.


In advice reported by Business Insider on October 7, 2026, Cherny argued that as AI models become more capable, users should spend less time micromanaging how they complete tasks and more time explaining what they actually want to accomplish.


His recommendation comes down to three principles: tell AI what you want it to do, tell it how much effort to spend, and tell it how to verify its work.


For businesses increasingly incorporating AI into everyday operations, that distinction could change how employees approach everything from research and reporting to marketing and financial analysis.


Why More Instructions Aren't Always Better


Early AI models often needed substantial guidance to complete complicated tasks reliably. Users learned to break work into smaller steps, specify exactly how responses should be structured and provide detailed instructions to keep the model on track.


Those techniques can still be useful. But Cherny's argument is that increasingly capable models can make more of those decisions themselves.


Consider the difference between asking an employee to achieve a result and telling them precisely how to perform every action along the way.


An experienced employee can often work more effectively when given a clear objective, relevant context and enough freedom to choose an appropriate approach. Cherny believes a similar principle increasingly applies to advanced AI models.


That doesn't mean providing less information about the task. It means distinguishing between information the model genuinely needs and instructions that unnecessarily restrict how it works.


Rule 1: Tell AI What You Want It to Do


The first principle is straightforward: clearly describe the outcome you're trying to achieve.

For a business owner, that might mean asking AI to identify opportunities to improve profitability, evaluate customer feedback or develop a plan for entering a new market.


The important part is explaining what a successful result should look like.

For example, "analyze our sales" leaves too much open to interpretation. It doesn't explain the business objective, the period being examined or what decisions the analysis should support.


A more useful request would be:


"Analyze our last 90 days of sales and identify the three biggest opportunities to improve profitability. Explain your recommendations in plain English."


This provides a clear direction without dictating every calculation, comparison or analytical step.


The model has an objective, but it also has room to determine how best to approach the work.


Rule 2: Tell AI How Much Effort to Spend


Not every task deserves the same level of investigation.


Rewriting a short email may require only a quick response. Evaluating a company's financial performance, researching competitors or comparing expansion opportunities demands considerably more attention.


Cherny recommends making that expectation clear.


Instead of assuming the AI knows how deeply to investigate, users can explicitly request a quick answer, a thorough analysis or deeper research and reasoning.


For example, a business owner might ask for a brief summary of customer feedback before a meeting. But when evaluating a significant investment, that same person could ask AI to investigate multiple options, compare their advantages and disadvantages, and identify risks before reaching a recommendation.


The distinction matters because a response can appear polished without being sufficiently researched.


By communicating the required level of effort, users give AI a better understanding of the importance and complexity of the task.


Rule 3: Tell AI How to Verify Its Work


The third principle may be the most important for businesses.


An AI-generated answer can sound convincing even when it contains incorrect calculations, outdated information or unsupported conclusions.


Cherny's recommendation is to explain how the result should be checked.


For financial analysis, that could mean verifying calculations against the original spreadsheet. For market research, it could mean checking claims against credible sources.

For an operational recommendation, it might involve identifying assumptions and highlighting information that is still missing.


A practical request could include:


"Check all calculations against the source data, identify any assumptions you make and flag conclusions that cannot be verified."


This doesn't guarantee accuracy, and AI reviewing its own work isn't a substitute for independent verification when the stakes are high.


But defining verification requirements gives the task a clearer standard for completion.


For businesses, the goal should be to receive work that can be examined and trusted, rather than simply accepting the first confident-sounding response.


What Better Prompting Looks Like in Practice


Imagine a business owner wants to understand why profitability has declined over the last quarter.


A vague request might be:

"Analyze our sales."


At the opposite extreme, the owner might write a lengthy prompt dictating every calculation, requiring twelve specific charts and prescribing the exact sequence of the analysis.


Neither approach is necessarily ideal.


The first doesn't explain what the business needs. The second may spend too much effort controlling the process instead of communicating the objective.


A stronger alternative would be:


"Analyze our last 90 days of sales. Identify the three biggest opportunities to improve profitability. Investigate thoroughly, verify your calculations against the original data and explain your recommendations clearly."


This request covers all three of Cherny's principles.


It establishes the outcome, specifies the level of effort and explains how the result should be verified.


The examples are illustrative rather than prompts quoted from Cherny, but they demonstrate how his advice can be applied to everyday business decisions.


Anthropic Has Been Simplifying Its Own Instructions


Cherny's October comments weren't the first time he made this argument.

In a July 2026 discussion, he described how Anthropic had removed approximately 80% of

Claude Code's system prompt as its models improved.


System prompts are the underlying instructions that guide how an AI application behaves.

They can define its role, expectations and operating rules.


The reduction reflected a broader idea: instructions that were helpful for less capable models may become unnecessary, or even restrictive, as models improve.


This doesn't mean AI applications no longer need guardrails or carefully designed workflows. Rather, it suggests that some of the complexity previously required to guide a model can be replaced by the model's improved ability to understand and execute a task.


The same thinking underlies Cherny's advice to everyday users.


Simple Prompts Still Need Good Context


There is an important distinction between simplifying a prompt and making it vague.


A business owner asking AI to develop a marketing strategy still needs to explain the company, its customers, its goals and any meaningful budget or operational constraints.


Someone requesting financial analysis still needs to supply accurate data and explain which decisions the analysis should support.


And repeatable business processes may still require detailed instructions to ensure consistency, compliance and predictable outputs.


In legal, financial and other high-stakes situations, human review and independent checks remain essential.


Cherny's advice isn't an argument against precision. It's an argument against unnecessary micromanagement.


The most effective prompt may contain substantial business context while leaving the model flexibility to determine the best approach.


Better AI Results Start With Better Objectives


The growing accessibility of AI has created an unusual challenge for businesses. Employees can now delegate increasingly complex tasks to software, but many still approach those tasks as though they need to describe every individual action.


Cherny's advice suggests a simpler approach.


Define the result you need. Explain how thoroughly the work should be performed. Establish how the result should be checked.


Then allow the model to determine an appropriate way to complete the task.


As AI becomes more capable, the advantage may shift away from knowing complicated prompting techniques and toward knowing how to define business problems clearly.


The skill isn't necessarily writing longer prompts. It's knowing what outcome matters and how to recognize a good result.



Sources

Business Insider, October 7, 2026: Claude Code creator Boris Cherny's AI prompting advice

Boris Cherny, July 2026: We Cut 80% of Claude Code's Prompt

Your Next Marketer Might Need to Code

October 26, 2026
Marketing is becoming more technical. As businesses adopt AI tools, automation and increasingly complex software, a new kind of specialist is emerging: the marketing engineer. For years, businesses have built marketing teams around familiar responsibilities: advertising, content creation, search engine optimization, social media and lead generation. Those responsibilities haven't disappeared. But the technology behind them has become increasingly complicated. Businesses now rely on advertising platforms, customer relationship management systems, email automation, website analytics, e-commerce software and AI tools. Making those systems work together is becoming an important part of marketing performance. That's where the marketing engineer comes in. What Is a Marketing Engineer? A marketing engineer combines marketing knowledge with technical skills. Traditional marketing specialists might develop a campaign strategy, create content or manage advertising budgets. A marketing engineer can help build the technical systems that make those activities more effective. That might involve connecting a CRM to advertising platforms, automating reporting, improving conversion tracking, integrating customer data or developing workflows using APIs and AI tools. The role connects marketing strategy with technical execution. Rather than relying entirely on separate developers, analysts and software specialists, businesses can use technical marketing expertise to bridge those functions. The Problem With Disconnected Marketing Systems Many businesses already have the technology they need. The challenge is getting it to work together. A company might use Google Ads to attract customers, Shopify to process orders, Klaviyo for email marketing and a CRM to manage relationships. Each platform collects information, but that doesn't necessarily mean the information is connected. Employees may still need to download reports, transfer leads between systems, update spreadsheets or manually compare campaign results against actual sales. Those disconnected processes can make marketing slower and more difficult to measure. Marketing engineering focuses on improving the connections between systems. When information moves reliably between platforms, businesses can reduce repetitive work and make decisions using more complete data. AI Search Is Changing How Businesses Get Discovered Technical marketing is also becoming more relevant as AI-powered search develops. Traditional search engine optimization focuses on helping websites appear in organic search results. But customers increasingly use AI-powered tools to research products, compare services and find answers. That has brought greater attention to two related concepts. Answer Engine Optimization (AEO) focuses on making information clear, accessible and useful for answer-based discovery. Generative Engine Optimization (GEO) focuses on improving how content can be discovered, understood and potentially referenced by generative AI systems. These approaches overlap with established SEO practices, including technically accessible websites, useful content and clear information architecture. They do not guarantee inclusion in AI-generated answers. For businesses, the practical lesson is that website content and technical structure need to support more than traditional search rankings. Automation Can Eliminate Repetitive Marketing Work Another opportunity is workflow automation. Marketing departments often spend considerable time on tasks that don't require much creative judgment. Downloading reports, updating spreadsheets, transferring customer information, preparing campaign summaries and moving data between applications are common examples. Many of these activities can be automated when systems are properly connected. That doesn't eliminate the need for marketers. It can give them more time to focus on strategy, creative development, customer relationships and improving campaign performance. The objective isn't to automate everything. It's to identify which tasks can be handled more reliably and efficiently through technology. Better Data Can Lead to Better Marketing Decisions One of the biggest challenges in marketing is understanding which activities actually contribute to revenue. A campaign might generate thousands of impressions, hundreds of clicks and dozens of leads without producing profitable customers. Marketing engineers can help businesses connect advertising activity with customer and sales information. That includes improving conversion tracking, integrating data sources and identifying where potential customers drop out of the buying process. The goal is to move beyond reporting what happened on an advertising platform and develop a clearer understanding of what happened across the business. Does Your Business Need a Marketing Engineer? Not every company needs to hire a full-time technical marketing specialist. For smaller businesses, the first step is identifying where disconnected systems, manual processes or unreliable data are creating problems. Consider whether your advertising and sales platforms share information, whether campaign performance can be connected to revenue and whether repetitive reporting tasks could be automated. It is also worth reviewing whether your website is technically accessible, well structured and capable of answering the questions potential customers are asking. Depending on the complexity of the business, these improvements might be handled by an internal employee, an outside specialist or a combination of both. The important decision isn't necessarily who to hire. It's determining which capabilities the business actually needs. The Future of Marketing Is More Connected Creative thinking, compelling messaging and customer relationships remain essential to effective marketing. But those strengths increasingly depend on the technology supporting them. Businesses need systems that communicate, data that can be trusted and processes that allow teams to work efficiently. Marketing engineering represents one approach to bringing those capabilities together. The next competitive advantage may not come from producing more content or launching more campaigns. It may come from building a marketing operation where everything works together.

P&G Rebuilt Its Operations and Saved $600 Million. Here's How.

October 23, 2026
Procter & Gamble didn't simply outsource its support operations. It first rebuilt how they worked, then brought in outside partners. By 2008, its business-services transformation had generated approximately $600 million in cumulative savings. In the late 1990s, Procter & Gamble faced a problem familiar to businesses of almost every size: too many teams performing similar work. Finance, human resources and other support functions were spread across different parts of the company. That meant duplicated activities, inconsistent processes and unnecessary costs. Instead of continuing to expand those departments independently, P&G decided to fundamentally change how its business operated. First, P&G Fixed Its Internal Operations In 1999, P&G began bringing its support functions into a single global organization. Over the following three years, the company consolidated and standardized more than 70 services. It also established shared-service centers to support operations around the world. The goal was to replace fragmented systems with a more consistent and efficient operating structure. Importantly, P&G didn't begin by handing all these functions to outside companies. It first reorganized how the work was performed internally. That foundation would make the next phase possible. Then Came $4.2 Billion in Outsourcing Agreements In 2003, P&G entered into outsourcing partnerships valued at approximately $4.2 billion. Outside specialists took responsibility for selected operations, including IT infrastructure, payroll support, accounts payable and facilities management. Rather than maintaining every capability internally, P&G could rely on partners with specialized expertise, established systems and the scale to manage these activities. But it didn't outsource everything. The company retained control of functions it considered strategically important, including procurement, logistics, IT architecture and technology-driven innovation. This distinction was central to the strategy. P&G wasn't trying to eliminate internal operations. It was deciding which activities made sense to manage itself and which could be handled more effectively by partners. The Benefits Went Beyond Cost Savings One of the clearest examples came during P&G's integration of Gillette. According to P&G's business-services leadership, its established systems and outsourcing partners helped the company mobilize approximately 750 additional people within weeks. The business-services integration was completed in 15 months, a process its leadership believed could otherwise have taken three to four years. That flexibility mattered. P&G could draw on existing capabilities and outside resources without having to build every part of the operation from scratch. By 2008, the company's broader business-services transformation had produced approximately $600 million in cumulative savings. Those savings came from a combination of internal consolidation, standardized processes and selective outsourcing, not outsourcing alone. What Smaller Businesses Can Learn Most businesses aren't managing P&G's scale or negotiating multibillion-dollar agreements. But the underlying decision is surprisingly relevant. As companies grow, they often add employees and departments without reconsidering how work gets done. Over time, the same problems can emerge: duplicated tasks, fragmented responsibilities, rising administrative costs and employees spending too much time on activities outside their core expertise. The first step isn't necessarily outsourcing. It's understanding the operation. Which processes are inefficient? Which functions require direct internal control? Which activities could a specialist handle more effectively? Customer support, bookkeeping, administrative work, creative production and certain technology operations may be candidates for outside support, depending on the business. But outsourcing a poorly defined process can simply transfer the problem to someone else. P&G's experience illustrates a more disciplined approach: improve the system, identify which capabilities are strategically important and use external partners where they offer a genuine advantage. You Don't Have to Build Everything In-House The most important lesson from P&G's transformation isn't that every business should outsource more. It's that growth doesn't require owning and managing every operational function. By combining stronger internal processes with carefully selected partners, businesses can reduce unnecessary complexity and concentrate their resources where they create the most value. Sometimes the smartest way to grow isn't to build another department. It's to rethink how the work gets done. Sources McKinsey & Company, From internal service provider to strategic partner: An interview with the head of Global Business Services at P&G (2008). Procter & Gamble, 2008 Annual Report.