How to Transform Inputs for Improved Outputs

Most people chase better tools, hoping for better results—but tools only amplify the quality of the thinking behind them. A mediocre prompt given to the best AI will still produce average outcomes, while a well-structured, intentional prompt can turn even a simple tool into something powerful. Because prompting is thinking in disguise—it’s the ability to break down ideas, communicate clearly, and direct intelligence with precision. When you get better at prompting, you’re not just improving outputs; you’re upgrading how you reason, decide, and solve problems. And once your thinking reaches that level, results stop being accidental—they become predictable. If you’re using AI regularly, this can completely change how you interact with tools like ChatGPT, Claude, or Gemini. Let me break it down in a way that actually sticks 👇 Level 1: Beginner — “Just tell it what to do” This is where most people start. You give a simple instruction and hope for magic. Example: “Give me 10 video ideas on productivity.” It works… but the output is generic because your thinking is generic. Level 2: Skilled — Add Context Now you guide the AI. You don’t just say what you want—you explain who it’s for and why it matters. Example: “List 10 productivity video ideas for college students with short attention spans.” Now the output starts becoming relevant. Level 3: Advanced — Define the Output This is where clarity becomes power. You tell AI: What to do Who it’s for AND how to present it Example: “List 10 productivity ideas for beginners with busy schedules. Format as a table with idea + one-line description.” Now you're not just getting answers… you're getting structured thinking. Level 4: Specialist — Assign a Role Here’s where things get interesting. You tell AI who it should become. Example: “Act as a content strategist…” Now the AI responds with depth, perspective, and intent—not just information. Level 5: Expert — Add Constraints Most people skip this—and that’s why their outputs feel bloated. You define limits: How many outputs What to avoid When to stop Example: “Give exactly 10 ideas. No extra explanation.” This is how you turn AI into a precision tool. Level 6: Elite — Add Reasoning & Quality Control This is the top 1%. You’re not just prompting… you’re engineering thinking. You ensure: Accuracy Uniqueness Value Example: “Ensure ideas are unique, actionable, and relevant. Stop at exactly 10.” Now AI is no longer assisting you. It’s collaborating with you at a high level. ✅️ Here’s the real insight most people miss: AI doesn’t reward intelligence. It rewards clarity. The gap between average users and power users isn’t the tool… It’s how they think before they type. 🔹️If you're a teacher, creator, or professional trying to leverage AI: 👉 Don’t just ask better questions 👉 Design better prompts 👉 Structure your thinking Because in the AI era… your prompt is your new skillset. 😎 Image Credit: Adam Biddlecombe

Stop winging it with AI prompts. 7 proven techniques to transform your results: Most people treat AI like Google - throw in some keywords and hope for the best. But AI responds to structure, context, and clarity. Here are 7 simple techniques that transformed my results: 1. RTF framework (request, task, format) Perfect for quick, clear outputs. Tell AI what you want, why you want it, and how to deliver the output. 2. Meta prompting Ask AI to help craft your prompt. "Help me refine and improve the following prompt ..." AI is good at many things: including writing great prompts. 3. Use delimiters XML tags and markdown structure help AI parse complex requests. Useful for large contexts where examples/documents could overlap. 4. RISEN framework Role, Instructions, Steps, End Goal, Narrowing. My go-to for complex projects that need expert-level thinking. 5. Chain-of-thought prompting Add "Think step by step" and watch accuracy improve. Especially powerful for analysis. 6. Role-based prompting "Act as a [specific expert] with [experience]" filters responses through the right lens every time. 7. Few-shot prompting Show AI 2-3 examples of what you want. It learns the pattern and applies it perfectly to new inputs. Pro tip: Layer these techniques. Role-based + Chain-of-thought + Few-shot e.t.c. Remember that AI is only as good as the questions you ask and the context you give it. What's your favourite prompting technique? 👇 --- Enjoy this? ♻️ Repost it to your network and follow Owain Lewis for more.

One of the most common mistakes that companies make is to focus too narrowly on a small set of metrics while overlooking the broader ecosystem of inputs that drive results. At Amazon, we rejected the conventional wisdom that Executives should focus on just a few high-level metrics. Instead, we spent years developing mechanisms to measure, analyze, and improve thousands of input metrics (actions) based on the impact they have on output metrics (business results). The key lessons from this approach are: 1. Do not limit the number of metrics you monitor – Track a broad set of metrics, add, delete, and edit them over time based on observed results. 2. Review both controllable inputs (leading indicators) and output metrics (results) – They must be reviewed in tandem to understand cause-and-effect relationships. 3. Regularly review, analyze, and adjust metrics – The metrics that Amazon tracks are continuously improved to more accurately represent the speed, quality, and cost of every customer-facing process. 4. Implement new product and process improvements designed to deliver improvements for your input metrics. If you have selected the right inputs, then improvements to your outputs will follow. 5. Control your processes by continuously reviewing all relevant input metrics to ensure they stay within desired tolerances as internal and external factors change over time. Following these steps uses the Six Sigma technique known as DMAIC - Define, Measure, Analyze, Improve, and Control. The typical approach is to focus deeply on metrics like sales and gross margin while spending little or no time measuring or managing elements of the customer experience. At Amazon, this focus is reversed.

Monday Technical Deep Dive: Prompting for Precision You've probably heard about AI everywhere, but are you prompting it right to get the best results? Getting useful output from models like Gemini or ChatGPT isn't magic; it's a skill called Prompt Engineering. If your prompt is weak, your output will be too. I recently attended Google’s Generative AI Leader Program and solidified a core principle: Better Inputs = Better Outputs. Here are three simple techniques to immediately improve your results: 1. Zero-Shot Prompting (The Baseline) This is the simplest approach. You give the model no examples, just the instruction. Example: "Explain the concept of API idempotency." When to use it: For basic questions, definitions, or tasks where the model already has extensive knowledge. It's your starting point. 2. Few-Shot Prompting (The Teacher) This is where you give the model a few examples of the desired input/output format before asking your actual question. You are essentially teaching it your style. Example: "Here are three examples of how I write a professional email closing: [Example 1], [Example 2], [Example 3]. Now, write an email to a recruiter following this style." When to use it: When the output needs to match a specific format, tone, or structure (e.g., code functions, marketing copy, or technical documentation). 3. Chain-of-Thought (CoT) Prompting (The Analyst) This is the most powerful technique for complex tasks. You instruct the model to explain its reasoning step-by-step before providing the final answer. Example: "Before giving the final answer, first list and explain the security risks associated with deploying this new cloud function. Then, suggest three mitigation strategies." When to use it: For complex analysis, multi-step problem-solving, or debugging. For me, this is essential when working on AI and Security concepts, as you need verifiable reasoning. Prompting is a skill that will only grow in importance. Which of these techniques are you going to test today? Let me know your results! #GenerativeAI #PromptEngineering #TechnicalDeepDive #SoftwareEngineering #AI

Teams obsess over AI's design outputs. Yet the best results come from obsessing over the inputs. I'm giving a talk next week on how to design stronger AI inputs, and here's the framework I keep coming back to, grounded in Anthropic's prompting guides and Sam Altman's assertion that AI can generate its own training data. The quality of an AI response is shaped long before the model generates an answer. → PROBLEM- What are we trying to solve? Every AI interaction should start with a design problem, whether that's improving onboarding, building trust, or reducing abandonment. Clearly naming the user need and business challenge gives AI something real to work with. → KNOWLEDGE- What should AI understand? Good design decisions need evidence. Background research, customer feedback, UX metrics, and past project learnings help AI understand users and behaviors, not just surface plausible-sounding ideas. The more good context, the more grounded the output. → CONSTRAINTS- How should AI behave? Design never operates in a vacuum! Brand standards, accessibility requirements, technical limits, and business goals tell AI what "acceptable" actually means. Without them, AI will confidently recommend things that look right but fall down when trying to implement them. → DIRECTION- What should AI do? This is where the work gets specific. The task names the challenge (improve navigation, increase conversion). The situation adds context: users, market conditions, business priorities, current performance. Together, they focus AI on the right opportunity. → JUDGMENT- Is this actually a good answer? Great design requires evaluation, not just generation. AI can produce dozens of options, but someone still has to judge them against user needs, business goals, and real UX metrics like trust, comprehension, and completion. Better inputs make judging the outputs easier, because you've already defined what "better" means. The dashed loop in my diagram is the part many people miss… outputs become inputs for the next decision. That feedback cycle is where design and AI actually get good together. It's also why we built Glare. The strongest AI systems are built on better inputs. How are you seeing this?

Finance leaders: You can't control revenue by focusing solely on outputs. You have to manage inputs.   Instead of "we need more revenue this quarter," try "let's book 20% more qualified meetings."   The first one is just hope. But the second one is action. So stop telling your team to "hit the number." and instead, tell them which levers to pull. Once they know that, they can take the right steps forward to improve revenue.   Because it's only an output. Revenue happens after a multi-week sales process, which happens after outbound campaigns, which happen after marketing efforts. If you're just tracking the end result, you're not helping anyone understand how to improve it.   Of course you'll measure revenue—that happens by default. But the metrics that actually drive improvement are the inputs you can directly control: → Number of outreach attempts → Meeting conversion rates → Pipeline velocity → Demo quality scores   The main principle: Only measure numbers that you know how to tell people to move.   As we've evolved at Equals, focusing more on inputs, the outputs have taken care of themselves.   But most leaders are still focused on the wrong end of the equation—demanding better results instead of coaching the specific behaviors that create them.

Most sales leaders manage to outputs. Andy Grove taught me to manage to inputs, and it changed how I have led ever since. High-Output Management, Andy's book, teaches you to treat a sales organization as a production system. You can't manage what you can't see. His prescription: pick a small number of leading indicators — metrics that predict future output, not just report past performance — and obsess over them weekly. The companion concept is catching problems at the lowest value stage. In manufacturing, raw material gets more expensive to scrap the further it moves through production. In sales, a rep 9 months in with no pipeline is infinitely more expensive to fix than catching a ramp issue in week 4. The two numbers I've built every weekly operating rhythm around: → First meetings sat per week, per rep → Visible pipeline per rep (by segment and ramp stage) Everything else, close rates, ASP, cycle time, matter, but they are downstream. Get the inputs right, and the outputs follow. The thing Grove hammers that most people miss: pair your indicators. Pipeline volume without pipeline quality is theater. Every leading metric needs a counter-metric or you'll optimize off a cliff. Doug May said something similar on a recent Hunters and Unicorns podcast: the productivity function exists to optimize every rep's horsepower, not just count who hit quota. He cut ramp time in half at Harness — not by hiring better, but by engineering the system around the seller. That's the whole game. What are the two leading indicators you're actually managing to right now?

𝟔 𝐏𝐫𝐨𝐯𝐞𝐧 𝐖𝐚𝐲𝐬 𝐭𝐨 𝐈𝐦𝐩𝐫𝐨𝐯𝐞 𝐂𝐡𝐚𝐭𝐆𝐏𝐓 𝐎𝐮𝐭𝐩𝐮𝐭𝐬 Getting mediocre results from LLMs? The problem is not the model. It is the prompt. Here are six techniques that immediately improve output quality. 𝟏. 𝐅𝐄𝐖-𝐒𝐇𝐎𝐓 𝐏𝐑𝐎𝐌𝐏𝐓𝐈𝐍𝐆 • Include a few input-to-output examples before your actual query.  • The model learns the pattern from your examples and applies it to the new input.  • Three well-chosen examples are usually enough to dramatically improve consistency and format. 𝟐. 𝐙𝐄𝐑𝐎-𝐒𝐇𝐎𝐓 𝐏𝐑𝐎𝐌𝐏𝐓𝐈𝐍𝐆 • Give precise instructions with no examples.  • Works best when your task is straightforward and you describe exactly what you want.  • No examples needed just clear, specific instructions that leave no room for interpretation. 𝟑. 𝐂𝐇𝐀𝐈𝐍-𝐎𝐅-𝐓𝐇𝐎𝐔𝐆𝐇𝐓 𝐏𝐑𝐎𝐌𝐏𝐓𝐈𝐍𝐆 • Ask for step-by-step reasoning.  • Instead of jumping straight to an answer, the model walks through Input to Thought to Thought to Output.  • This reduces errors on complex problems like math, logic, and multi-step analysis.  • Add "think step by step" to your prompt and watch accuracy improve. 𝟒. 𝐏𝐑𝐎𝐌𝐏𝐓 𝐇𝐈𝐄𝐑𝐀𝐑𝐂𝐇𝐘 Define different levels of authority within your prompt: • System Prompt: Sets the foundational behavior and constraints • Development Prompt: Adds task-specific instructions and context • User Prompt: The actual query on top Layering prompts this way gives you precise control over how the LLM responds. 𝟓. 𝐑𝐎𝐋𝐄-𝐒𝐏𝐄𝐂𝐈𝐅𝐈𝐂 𝐏𝐑𝐎𝐌𝐏𝐓𝐈𝐍𝐆 Assign a persona to shape the output style and depth. The same user query produces completely different outputs depending on the role: • "As a Research Scientist": technical, detailed, citation-heavy • "As a Technical Writer": clear, structured, documentation-style • "As a Product Manager": strategic, user-focused, business-oriented 𝟔. 𝐍𝐄𝐆𝐀𝐓𝐈𝐕𝐄 𝐏𝐑𝐎𝐌𝐏𝐓𝐈𝐍𝐆 Specify what the LLM should avoid by including "do not" constraints. Example: "Do not use any marketing language or exaggeration" combined with "Write a product description." This eliminates the fluff and forces the model to stay grounded. 𝐖𝐇𝐄𝐍 𝐓𝐎 𝐔𝐒𝐄 𝐄𝐀𝐂𝐇 Simple, clear tasks: Zero-shot. Need consistent formatting: Few-shot. Complex reasoning or math: Chain-of-thought. Production applications: Prompt hierarchy. Domain-specific outputs: Role-specific. Removing unwanted patterns: Negative prompting. 𝐓𝐇𝐄 𝐏𝐑𝐈𝐍𝐂𝐈𝐏𝐋𝐄 Better prompts do not mean longer prompts. They mean more deliberate prompts. Pick the right technique for your task and the model does the rest. 𝐖𝐡𝐢𝐜𝐡 𝐩𝐫𝐨𝐦𝐩𝐭𝐢𝐧𝐠 𝐭𝐞𝐜𝐡𝐧𝐢𝐪𝐮𝐞 𝐡𝐚𝐬 𝐦𝐚𝐝𝐞 𝐭𝐡𝐞 𝐛𝐢𝐠𝐠𝐞𝐬𝐭 𝐝𝐢𝐟𝐟𝐞𝐫𝐞𝐧𝐜𝐞 𝐢𝐧 𝐲𝐨𝐮𝐫 𝐰𝐨𝐫𝐤𝐟𝐥𝐨𝐰? ♻️ Repost this to help your network get started ➕ Follow Sathish for more #GenAI #AIAgents #AgenticAI

The most valuable habit I've picked up in the last few weeks is using AI Agent Management rules to manage my own life. Agentic Principles like Loss Function & Backpropagation can help you convert ambitious goals into specific outcomes, outputs and inputs. When you build an AI agent, you're forced to define a system goal, the tools required, and the metrics to track your agent's progress. If this works so well for managing AI brains, why not try it to manage our brains as well? Outcomes are the final goal (getting 1,000 customers/earning 5 Lacs a month/building a better body, getting 1M followers) Outcomes are not in our direct control. To reach outcomes, you need to generate a certain volume of outputs. It's similar to defining the system prompt for your agents. Outputs are the things you do and create to reach your outcomes. Then, you need a measure of "how good" or "how often" you will create these outputs. If I need 1M followers, I need to release content consistently. I need better conversion on our landing page. I need to optimise our website. When planning outputs, you need to be specific about the volume, frequency and minimum quality. Once you've locked outputs, it's time to do the actual work. This is when we define inputs If I need to release content consistently, I need a routine of reading, writing and analysing patterns 1 hour every day. If I need better conversion on our landing page, I need to spend 1 week analysing our competition and website heatmaps from a tool like Microsoft Clarity. When you write your to-do list for today, ask yourself how many of these are inputs vs outputs vs outcomes. Once you've split your list into three parts, prioritise items within each part, doing first those that give the greatest benefit. The first thing that comes to your mind will rarely be the best thing to do. Every 10 days, ask yourself if your inputs/outputs need to change to reach the outcome. In AI, we often talk about the Loss Function and Backpropagation. Backpropagation is similar to a quarterly business review, where you figure out why you missed your targets and how to fix your strategy for next month. The Loss Function tells you: "How far off are we from where we wanted to be?" If the number is high, your current model is poorly tuned. Your outcomes will help you calculate the loss function. Improving your inputs after feedback is you practising backpropagation. I'm not sure if these agentic principles were made by MBAs or Engineers because it is such a beautiful map on how to manage any kind of intelligence (AI or Human)