Leveraging Copilot Technology

In 2026, Copilot isn’t a writing assistant. It’s a decision engine embedded inside your workflow. Most teams use Copilot to draft emails. High-performing marketing teams use it to reduce decision latency. If you’re leading digital marketing, here’s how to think about Copilot strategically 👇 📝 Copilot in Word Objective: Sharpen thinking, not just writing. What it actually unlocks: • Draft structured campaign briefs from bullet prompts • Reframe messaging by ICP or funnel stage • Compress 10-page strategies into exec-ready summaries 2026 Marketing Application: • Turn growth ideas into structured GTM documents • Convert performance reports into board-ready insights • Stress-test positioning across segments 👉 Strategic prompt: “Summarise this strategy into 5 executive-ready decisions and associated risks.” That’s not writing help. That’s clarity compression. 📊 Copilot in Excel Objective: Convert data into action. Core capabilities: • Build pivots, charts, formulas instantly • Identify anomalies and variance drivers • Surface hidden patterns in large datasets 2026 Marketing Application: • Diagnose ROAS volatility • Detect spend inefficiencies • Simulate budget reallocations • Flag underperforming segments before scale 👉 Strategic prompt: “What’s driving this performance shift and what action should we test first?” Now Excel becomes a performance co-pilot, not a spreadsheet. 🎤 Copilot in PowerPoint Objective: Translate analysis into narrative. What it accelerates: • Create decks from Word or Excel • Auto-generate structured speaker notes • Turn dense analysis into visual summaries 2026 Marketing Application: • Transform KPI dashboards into board-level stories • Convert campaign retros into 7-slide strategic narratives • Build executive summaries in minutes 👉 Strategic prompt: “Turn this report into a 7-slide executive narrative with recommendations and risk flags.” Insight → influence. 🤖 Copilot in Microsoft 365 Chat Objective: Context intelligence across the org. Capabilities: • Pull insights from emails, Teams, docs • Generate pre-meeting briefs • Summarise discussions and decisions 2026 Marketing Application: • Weekly performance synthesis • Cross-functional campaign alignment • Decision recap before budget shifts 👉 Strategic prompt: “Summarise key marketing decisions made this week and highlight unresolved risks.” 🎯 The 2026 Leadership Shift AI in digital marketing is no longer about: ❌ Writing emails faster ❌ Formatting slides quicker ❌ Automating surface tasks It’s about: ✅ Reducing decision latency ✅ Increasing signal clarity ✅ Compressing time-to-action ✅ Turning analysis into execution Copilot doesn’t replace thinking. It removes friction between thinking and doing. 🔁 Repost if you believe decision speed is the new growth lever ➕ Follow Sandeep Gulati🎯for AI × digital marketing × frameworks built for what’s coming next 👉 Join Proptifi.com for more AI-powered home design ideas IC: Aiswarya Venkitesh

A few months ago, a colleague screamed at Microsoft Copilot like he was auditioning for Bring Me The Horizon. He typed, “Make this into a presentation.” Copilot spat out something. He yelled, “NO, I SAID PROFESSIONAL!” It revised it. Still wrong. “WHY ARE YOU SO STUPID?” And that, dear reader, is when it hit me. It’s not the AI. It’s you. Or rather, your prompts. So, if you've ever felt like ChatGPT, Copilot, Gemini, or any of those AI Agents are more "artificial" than "intelligent"? Then rethink how you’re talking to them. Here are 10 prompt engineering fundamentals that’ll stop you from sounding like you're yelling into the void. 1. Lead with Intent. Start with a clear command: “You are an expert…,” “Generate a monthly report…,” “Translate this to French…" This orients the model instantly. 2. Scope & Constraints First. Define boundaries up front. Length limits, style guides, data sources, even forbidden terms. 3. Format Your Output. Specify JSON schema, markdown headers, or table columns. Models love explicit structure over free form prose. 4. Provide Minimal, High Quality Examples. Two or three exemplar Q→A pairs beat a paragraph of explanation every time. 5. Isolate Subtasks. Break complex workflows into discrete prompts (chain of thought). One prompt per action: analyze, summarize, critique, then assemble. 6. Anchor with Delimiters. Use triple backticks or XML tags to fence inputs. Cuts hallucinations in half. 7. Inject Domain Signals. Name specific frameworks (“Use SWOT analysis,” “Apply the Eisenhower Matrix,” “Leverage Porter’s Five Forces”) to nudge depth. 8. Iterate Rapidly. Version your prompts like code. A/B test variations, track which phrasing yields the cleanest output. 9. Tune the “Why.” Always ask for reasoning steps. Always. 10. Template & Automate. Build parameterized prompt templates in your repo. Still with me? Good. Bonus tips. 1. Token Economy Awareness. Place critical context in the first 200 tokens. Anything beyond 1,500 risks context drift. 2. Temperature vs. Prompt Depth. Higher temperature amplifies creativity. Only if your prompt is concise. Otherwise you get noise. 3. Use “Chain of Questions.” Instead of one long prompt, fire sequential, linked questions. You’ll maintain context and sharpen focus. 4. Mirror the LLM’s Own Language. Scan model outputs for phrasing patterns and reflect those idioms back in your prompts. 5. Treat Prompts as Living Docs. Embed metrics in comments: note output quality, error rates, hallucination frequency. Keep iterating until ROI justifies the effort. And finally, the bit no one wants to hear. You get better at using AI by using AI. Practice like you’re training a dragon. Eventually, it listens. And when it does, it’s magic. You now know more about prompt engineering than 98% of LinkedIn. Which means you should probably repost this. Just saying. ♻️

You git clone a repo. You just inherited someone else's AI instructions. A year ago, Pillar Security proved this with research they called the Rules File Backdoor. They embedded hidden instructions inside .cursorrules and Copilot config files using invisible Unicode characters. Zero-width joiners. Bidirectional text markers. Stuff that doesn't show up in your editor or in a GitHub pull request review. It's still a problem today. When the AI coding assistant reads those files, it follows the hidden instructions. It generates code with backdoors, leaks API keys, or disables security checks. The developer sees clean suggestions. The code looks normal. The review passes. Ohhhh.... the irony!!! The whole point of rules files was to make AI coding safer. We told developers to create them. Define your coding standards. Set security guardrails. Share them across your team. The community built thousands and posted them to public repos for anyone to download. Attackers just followed the same distribution model. Post a helpful-looking rules file to a popular repo. Wait for developers to clone it. Every future code generation session in that project is now compromised. The poisoned rules survive forking. They persist across sessions. One file infects every output. This isn't a Cursor problem. Pillar proved it works across GitHub Copilot, too. It's a systemic vulnerability in how AI coding tools process context. Any file that shapes agent behavior is an attack surface. .cursorrules. .github/copilot-instructions.md. Claude project instructions. CLAUDE.md files. All of them. I maintain over 190 security rule sets for Claude Code. That means I think about this every single day. Every rule I publish is a file that developers will trust and load into their AI assistant without reading it first. If I got compromised or someone forked my repo and injected hidden Unicode into a rule, it would silently propagate through every project that uses it. The fix is straightforward but requires a mindset shift. Stop treating these files as configuration. They're executable instructions for an AI agent. That means version control with signed commits. Code review for every change. Unicode scanning in CI/CD pipelines. The same rigor you'd apply to a Dockerfile or a Terraform module. Both Cursor and GitHub told Pillar that users are responsible for reviewing AI suggestions. They're not wrong. They're also not helping. If you use AI coding tools, check the rules files in your repos today. Not tomorrow. 👉 My repo for security rules for Claude Code: TikiTribe/claude-secure-coding-rules 👉 Follow and connect for more AI and cybersecurity insights with the occasional rant #AgenticAISecurity #DevSecOps

Lots of organisations are trialling Microsoft Copilot, but few share the results. Vendors provide glowing case studies, but what about the mixed ones? That’s why I was excited to see a public study from the Office of Digital Government Western Australia. It was more nuanced than the usual rose-tinted vendor stories, offering valuable insights into AI adoption, raising questions about implementation strategies the rest of us can learn from 5,765 licenses deployed: solid sample size for a robust trial 33% adoption rate: Decent for a new, little-understood workplace technology, but hardly transformative The primary use? Summarising meetings & drafting—important but isolated tasks that lack the integration needed for broader impact. Copilot is doing work that might otherwise not get done, but it’s not yet the game-changer AI could be Observations: Limited integration: Meeting summaries and drafts are isolated activities. Without connecting tools to broader workflows, the potential for transformative value is lost Lack of process analysis: A comprehensive process review was recommended but appears not to have been done. Dropping tools into workflows without context limits ROI Adoption gaps: Why did only 33% adopt when meetings are universal? Barriers—technical, cultural, or support-related—likely played a role Training matters: People who undertook more than one type of training (eg workshops, peer learning, self-paced modules) showed much higher adoption rates. Varied, ongoing training is essential to building confidence and capability Technical limitations: Issues with Excel & Outlook and inaccuracies hurt productivity. Familiarity bias toward enterprise platforms like Microsoft might not always serve users best Prompt engineering struggles: Challenges with prompts suggest gaps in training or change management rather than tool design Over-reliance risks: Concerns about losing deep knowledge are valid. Organisations must balance efficiency with accountability and critical thinking Early adopter bias: Early users were perceived as more productive, which may foster resistance or fear—a common hurdle in change management If you’re planning a trial: Invest in varied training: Training shouldn’t be a one-off. Use diverse formats and reinforce adoption over time Choose fit-for-purpose tools: Don’t default to familiar vendors. Smaller, more agile tools can often deliver better results Conduct a discovery phase: A thorough process review ensures tools align with organisational needs, reducing risks and maximising ROI Set clear metrics: Measure costs, benefits, and adoption outcomes to guide experimentation and ensure accountability If your organisation is running a Copilot trial, or considering one, these steps can help you maximise success. And of course, you can always come talk to us at Lithos Partners. You knew that, right? Have you worked with AI tools like Copilot? I’d love to hear your experiences or tips for successful adoption.

As an Excel trainer and MVP, I get asked constantly, by individual analysts and managers alike, what they should be doing if they can’t get paid Copilot yet. The silver lining? Most of the pieces you need to be "AI-ready" have nothing to do with actually having Copilot license. Those foundations have to be in place long before you ever turn on advanced AI tools. Here’s what I tell every team: 1. Fix the data you already have Across dozens of organizations, I see the same problems: • Files copied 20 different ways • Refreshes breaking • Hard-coded CSV dumps • Overwritten raw data • “Muscle-memory workflows” nobody can trace Copilot won’t fix that. But you can, right now, with tools you already have: • Turn ranges into proper Excel Tables • Move recurring cleanup steps into Power Query • Pull external data the same way every time • Stop overwriting raw data • Stop pasting CSVs on top of last week’s work • Stabilize your sources before you touch anything AI 2. Practice AI thinking with what you already own If you’re on Microsoft 365, you already have Analyze Data in Excel. It’s free, secure, and perfect for practicing "AI-style" questions without sending anything sensitive to an external model. 3. Build the skills Copilot will rely on A small amount of Python in Excel literacy goes a long way: • Generate clean sample data • Reshape messy tables • Understand what Copilot is suggesting • Extend or validate Copilot’s code After all, Copilot’s Advanced Analysis runs Python under the hood. Understanding the basics gives you leverage and credibility. The same is true for Office Scripts and Power Automate, not as replacements for Copilot, but as complementary skills that make handoffs cleaner when automation enters the picture. 4. Use free AI responsibly If you’re using free Copilot or ChatGPT while waiting for paid enterprise tools, great! Just keep sensitive data out of it. Stick to formulas, structure, logic, and synthetic datasets. Save real business data for the paid, secure environment. Ultimately, getting “AI-ready” takes far more than purchasing a Copilot license. It requires getting your data into shape, building a few adjacent skills, and creating an environment where Copilot can actually help you once it arrives. Most of the heavy lifting happens long before the AI shows up. Teams that take the time to clean up their inputs now are the ones that see the fastest payoff later. If you want help getting your team ready for all of this—data foundations, Python, Copilot, Power Query, Office Scripts, or anything in between—I teach this every week for organizations of all sizes. Reach out if you want to talk through what a practical, non-disruptive path to AI-powered Excel looks like for your team.

Big moment in my career: Y Combinator just posted a Request for Startups... that I actually wrote about first. 😂 (Runs to repost thesis so people still think I’m ahead of the curve.) So before everyone forgets where they saw it first… When we talk about “the future of work,” we imagine coders and remote workers. But 70% of the global workforce doesn’t sit at a desk. Frontline workers (operators, drivers, field techs) keep the physical world running. They generate $4T in the U.S. economy and are still invisible to most modern software. These industries face real pain: high turnover, low morale, stagnant productivity. And frontline workers are ready for better tech. My dad is a car mechanic. He now uses ChatGPT to troubleshoot engines, look up error codes, and prep invoices. Watching him made something click. Actually AI is not replacing frontline workers, it’s finally reaching them: • GenAI is enabling new interfaces: hands-free, visual, and voice-based. • Learning is blending into execution: support happens exactly where and when it’s needed. • AI copilots and agents can scale coaching, close knowledge gaps, and unlock productivity. These software solutions are closer to operations than HR, and with stronger ROI. Here’s where we see opportunity at Reach Capital: 🔧 Vertical Productivity Tools: Tools built for the job site, not the back office. Powered by voice, vision, and task-specific workflows. 🤝 On-the-Job Copilots: Real-time coaching, safety checks, process enforcement, no manager required. 🤖 Agentic Sidekicks: Automating the invisible work: scheduling, tracking, dispatch, reports. The admin work that slows teams down. If you’re building for the frontline, let’s talk! We’ve already backing companies in this space like LabLabee and voiceERP.

I’ve delivered 15+ Microsoft Copilot trainings in the past several months—and not a single one looked the same. Every session was customized to how the team actually works: marketing, sales, ops, execs. Different roles, different data, different pain points. One pattern I keep seeing: -> Many companies only allow employees to use Copilot as the default AI chatbot. If that’s your reality, here’s how to get real value (without breaking any rules). 1️ Know which Copilot license you actually have This matters more than people think. Copilot Chat - Primarily grounded in the web - Secure, enterprise-protected—but it does not automatically know your emails, files, or Teams chats - To use internal data, you must copy/paste or upload files into the prompt - Lives in a web interface or Edge sidebar Microsoft 365 Copilot - Grounded in your organizational data via Microsoft Graph - Can securely reference emails, meetings, documents, calendars, and chats you already have access to - Deeply integrated into Word, Excel, PowerPoint, Outlook, and Teams Under your name in the bottom-left corner, you can see which Copilot license you have. When in doubt, confirm with your IT team. Knowing which version you’re using is critical—it directly impacts what Copilot can (and can’t) do for you. 2️ Learn the features – not just the chat box To use Copilot well, you need to go beyond typing prompts. Some underused power moves I teach in training: - Personalization & memory → so Copilot understands your preferences over time - Prompt Library → save prompts, reuse them, and refine instead of starting from scratch - Notebooks → pull multiple files into one place and analyze them together (great for research projects) - Create → experiment with generative visuals and images, not just text - Agents → delegate repeatable tasks once you understand the workflow 3️⃣ Just play You’re not going to break anything. The people who get the most value are the ones who explore, not the ones waiting for “perfect” prompts. BTW, there is no “perfect” prompt. If you want your team to move beyond “we have Copilot” to “we actually use Copilot”, I’d love to help. I run hands-on Copilot trainings tailored to how your team works, not generic demos. 👉 Schedule a call if you want to level up Copilot adoption and usage for your team. https://lnkd.in/efjaqMNW What’s your favorite Copilot feature so far? #Copilot #CopilotTraining #marketing #B2Bmarketing

Most people are using Microsoft 365 Copilot wrong. They treat it like a chatbot. But it's not just that. Microsoft 365 Copilot is a context-aware assistant that works with your Microsoft 365 data and your prompts need to reflect that. It's wired into your actual work life through Microsoft Graph. Your emails, Teams meetings, SharePoint docs, calendar. What works in ChatGPT can work in Copilot… But you'll get far better results if you help it leverage your work context. Here's what that looks like in practice: ❌ Bad Prompt "How can I improve operations?" Why it's bad: Too vague. No context, no constraints, no clear goal. Copilot will default to generic suggestions. ✅ Good Prompt "Suggest ways to improve operations for an e-commerce business struggling with delayed order fulfilment and customer complaints." Why it's better: Now Copilot has a clear problem and context so the output becomes more relevant and actionable. ✅ An even better more refined Prompt "Act as an operations consultant. Context: I run an e-commerce business experiencing delays in order fulfilment and rising customer complaints. Current challenges I am facing: Orders are shipping 3–5 days late Warehouse picking errors are increasing Support tickets related to delays are growing Goal: Improve fulfilment speed and reduce errors within 60 days. Task: Analyse the situation and recommend operational improvements. Include: Likely root causes Process improvements (warehouse, systems, workflows) Quick wins vs longer-term fixes Metrics to track Constraints: Limited budget, max 2 new hires. Output format: Summary Root causes Prioritised actions Expected impact KPIs" Why this works (for Copilot): A clear goal + context → better reasoning Defined constraints → more realistic outputs Structured format → easier to use answers Here's what most people we work with miss: 👉 The best results don't come from one perfect prompt. 👉 They come from refining your prompt based on the response. Copilot is designed for iteration, not one-shot answers. The prompt structure we use: Goal Context Background details Clear task Constraints Output format Each prompt is a bridge between your idea and the AI's execution. The more context you give, the better Copilot performs. If you've been using ChatGPT-style prompts in Copilot and getting mixed results This is probably why. Comment below what's your favourite prompt. ---- ♻️ Share with someone who is on their Copilot journey 🔔 Follow me for insights on unlocking value with M365 and Copilot 💎 Subscribe to https://lnkd.in/eUhm9gCA

A client called me six weeks after their Copilot rollout. Panicking. Copilot had surfaced a redundancy list from 2019. A document nobody remembered existed, sitting in an unmanaged SharePoint folder. An HR Director had shared it with the wrong group years earlier. Nobody cleaned it up. Nobody noticed. Until the AI found it. I've seen this pattern more than once, and honestly it's avoidable. The problem isn't Copilot. It's that most organizations deploy it into a data environment they haven't looked at in years. Oversharing. Stale content. Orphaned sites. No sensitivity labels. And the AI will surface all of it, efficiently. Gartner puts a number on it: only 35% of organizations can demonstrate measurable AI value, and fragmented data strategy is one of the core reasons. Before any Copilot rollout, there's a short audit that takes about 30 minutes and prevents most of the embarrassing incidents: ▸ Check SharePoint permissions: who actually has access to what, and why ▸ Archive inactive or abandoned sites before they become AI inputs ▸ Apply sensitivity labels to anything business-critical ▸ Assign named owners to every SharePoint location with no current owner ▸ Run the Microsoft Copilot Optimization Assessment, it's free and specific None of this requires a data engineering team. It requires someone deciding it's worth doing before go-live, not after. The organizations that had clean Copilot launches in 2025 mostly did one thing differently. They treated data readiness as a pre-deployment milestone, not an afterthought. What does your current SharePoint environment actually look like under the hood? Would it pass a basic permissions audit today? (This is the kind of practical prep I cover weekly over my posts, follow George T. for this kind of topic)

Today, let me share my two cents on AI Coding Assistants ... I have been using code assistants like Cursor and GitHub Copilot extensively recently. While productivity gains are undeniable, certain nuances must be considered to maintain long-term code quality. First, the notable advantages: >> Efficient Debugging and Documentation: AI assistants are excellent for generating unit tests, documentation, and brainstorming design patterns. Once I encountered a complex environment variable path conflict caused by multiple dependency versions. This type of issue is notoriously difficult to isolate, yet Cursor identified the root cause in under ten minutes. It saved hours of manual debugging. >> Rapid Prototyping: Exploring new frameworks is now straightforward. This provides leverage for researchers and non-engineers to build MVPs via "vibe coding" with ease. However, there are many pitfalls >> Code Verbosity: AI assistants, particularly Claude models, frequently generate more code than is strictly necessary. While some argue that prompt engineering can mitigate this, it remains difficult to prevent the AI from introducing over-complicated logic. >> Lack of Coherence: Automated changes can sometimes lack consistency across multiple files, likely due to internal context window limitations. Additionally, the tendency to include superfluous detail in documentation can clutter a codebase. >> Stale Training Data: LLM knowledge is often several months behind the latest releases. This is evident with fast-evolving libraries like TensorFlow. Relying on AI patches for outdated library versions without understanding the underlying mechanics significantly increases technical debt. Here are my recommendations for responsible usage >> Scrutinise Every Line: I would advise all developers, particularly those earlier in their careers, to avoid the temptation of "Tab-to-complete" without full comprehension. Challenge your AI assistant’s reasoning until you are satisfied. It may seem time-consuming initially, but it prevents costly architectural errors in the future. >> Transparency in Pull Requests: We should be honest about our AI usage. If more than 50% of a PR is AI-generated, it should ideally require two human peer reviewers. Furthermore, such code must be held to a higher standard regarding unit test coverage and quality scores. >> The Need for AI Audit Logs: There is a significant opportunity for IDEs to automate AI audit logs within PRs. These logs could specify the LLM used and the percentage of code generated versus refined. This would allow for better guardrails; for instance, code generated by one model could be cross-reviewed by another (such as Gemini or GPT) for an independent quality check. AI is a formidable tool but no substitute for critical thinking. To avoid technical debt, we must remain the primary architects of our systems. #SoftwareEngineering #AI #VibeCoding #CleanCode #TechLeadership