https://aditya-marin-gasga.pages.dev/ Data, guides and lessons on using AI in marketing. en-gb https://aditya-marin-gasga.pages.dev/job-application-response-rate/https://aditya-marin-gasga.pages.dev/job-application-response-rate/Wed, 04 Nov 2026 13:00:00 GMTNine in ten applications got no reply. My own data on response rates, rejections and ghosting, set against the 2026 hiring research. https://aditya-marin-gasga.pages.dev/apply-with-claude/https://aditya-marin-gasga.pages.dev/apply-with-claude/Wed, 04 Nov 2026 13:00:00 GMTThe job-search system I built with Claude: nine steps, a prompt to start with, and what went wrong. https://aditya-marin-gasga.pages.dev/writing/ai-made-posting-free-trust-is-scarce/https://aditya-marin-gasga.pages.dev/writing/ai-made-posting-free-trust-is-scarce/Wed, 04 Nov 2026 12:00:00 GMTWhen anyone can post at no cost, volume stops being an edge. What platforms, surveys and one study say about AI content, and why I'd publish less. https://aditya-marin-gasga.pages.dev/writing/fix-the-page-before-you-buy-more-traffic/https://aditya-marin-gasga.pages.dev/writing/fix-the-page-before-you-buy-more-traffic/Wed, 04 Nov 2026 12:00:00 GMTAt Gaba Healthcare, 1.8% of visitors booked. I rebuilt the pages before spending more, and the rate reached 5.0%. The steps, the limits of the proof, and where AI helps. https://aditya-marin-gasga.pages.dev/writing/ai-transformation-social-team-workflows-first/https://aditya-marin-gasga.pages.dev/writing/ai-transformation-social-team-workflows-first/Wed, 04 Nov 2026 12:00:00 GMTMcKinsey, BCG and Gartner point the same way: the tools aren't what holds teams back. A practical order of steps for a social team, with the sources and my caveats. https://aditya-marin-gasga.pages.dev/writing/how-i-cut-weekly-research-from-4-hours-to-1/https://aditya-marin-gasga.pages.dev/writing/how-i-cut-weekly-research-from-4-hours-to-1/Wed, 04 Nov 2026 12:00:00 GMTI wrote the steps down first, then let Claude, n8n and Make do the reading. Weekly research fell from four hours to one. What worked and what broke. https://aditya-marin-gasga.pages.dev/writing/listen-before-you-generate-social-ai/https://aditya-marin-gasga.pages.dev/writing/listen-before-you-generate-social-ai/Wed, 04 Nov 2026 12:00:00 GMTMost social teams start AI on writing posts. I'd start on listening, because it feeds every other decision. A practical case and the rules. https://aditya-marin-gasga.pages.dev/writing/performance-marketing-ai-defaults-audit/https://aditya-marin-gasga.pages.dev/writing/performance-marketing-ai-defaults-audit/Wed, 04 Nov 2026 12:00:00 GMTGoogle and Meta now automate targeting, copy and budgets by default. A monthly audit of what's switched on, what you feed it, and proof it works. https://aditya-marin-gasga.pages.dev/writing/social-posts-are-search-results-now/https://aditya-marin-gasga.pages.dev/writing/social-posts-are-search-results-now/Wed, 04 Nov 2026 12:00:00 GMTSocial, search and AI assistants are merging into one place where people find things. What the evidence shows, what it doesn't, and what social teams should change. https://aditya-marin-gasga.pages.dev/writing/test-120-ad-variations-without-wasting-budget/https://aditya-marin-gasga.pages.dev/writing/test-120-ad-variations-without-wasting-budget/Wed, 04 Nov 2026 12:00:00 GMTA few hooks, many rewrites of each, and budget that follows sales. How a team tested 120 ad copy variations at The Cirqle, what went wrong, and where AI helps now. https://aditya-marin-gasga.pages.dev/writing/when-your-brands-bot-answers-you-answered/https://aditya-marin-gasga.pages.dev/writing/when-your-brands-bot-answers-you-answered/Wed, 04 Nov 2026 12:00:00 GMTAI can handle many social replies, but not the blame. A tribunal ruling, the EU chatbot rule and customer data point to a three-tier policy. https://aditya-marin-gasga.pages.dev/writing/your-first-automation-a-weekly-research-brief/https://aditya-marin-gasga.pages.dev/writing/your-first-automation-a-weekly-research-brief/Wed, 04 Nov 2026 12:00:00 GMTWrite the steps down, then automate them in n8n or Make. The same weekly brief, side by side in both tools, with the checks that keep it honest. https://aditya-marin-gasga.pages.dev/writing/lead-scoring-fit-vs-engagement/https://aditya-marin-gasga.pages.dev/writing/lead-scoring-fit-vs-engagement/Tue, 07 Jul 2026 18:39:00 GMTCompanies score leads on engagement (opens, clicks, page views), then wonder why the high scores don't close. Engagement shows interest. It doesn't show fit. https://aditya-marin-gasga.pages.dev/writing/claude-code-cost/https://aditya-marin-gasga.pages.dev/writing/claude-code-cost/Mon, 06 Jul 2026 13:06:00 GMTClaude Code can cost $20 or over $1,200 a month for the same work. The price on the pricing page is the wrong number. Here's what actually sets your Claude Code cost. https://aditya-marin-gasga.pages.dev/writing/removing-the-first-job/https://aditya-marin-gasga.pages.dev/writing/removing-the-first-job/Fri, 26 Jun 2026 20:05:00 GMTThe fear is mass layoffs. The data shows something narrower: a 13 percent drop in entry-level employment in exposed fields. And the AI screening those candidates runs under a bias law almost no one enforces. https://aditya-marin-gasga.pages.dev/writing/how-rag-works-and-fails/https://aditya-marin-gasga.pages.dev/writing/how-rag-works-and-fails/Mon, 22 Jun 2026 15:20:00 GMTThe follow-up to the RAG explainer. The four ways RAG breaks in production (chunking, retrieval, faithfulness, a stale index), the fixes that hold, and the testing that shows them. https://aditya-marin-gasga.pages.dev/writing/what-is-rag-really/https://aditya-marin-gasga.pages.dev/writing/what-is-rag-really/Mon, 22 Jun 2026 14:16:00 GMTRAG from scratch: how chunk, embed, store, retrieve, augment and generate fit together, why it cuts hallucination without ending it, and where it quietly breaks. About 13 minutes. https://aditya-marin-gasga.pages.dev/writing/what-is-a-context-window-really/https://aditya-marin-gasga.pages.dev/writing/what-is-a-context-window-really/Mon, 22 Jun 2026 14:14:00 GMTA plain explainer of the context window: the shared token budget a model reads at once, why long context costs more and recalls less, and when retrieval beats a bigger window. About 12 minutes. https://aditya-marin-gasga.pages.dev/writing/agents-can-pay-now-governance-cannot/https://aditya-marin-gasga.pages.dev/writing/agents-can-pay-now-governance-cannot/Sat, 20 Jun 2026 13:00:00 GMTAWS gave AI agents a way to spend money on their own. The payment was never the hard part. Approving, capping and auditing a non-human spender is, and most companies haven't built for it. https://aditya-marin-gasga.pages.dev/writing/ai-note-taker-teardown/https://aditya-marin-gasga.pages.dev/writing/ai-note-taker-teardown/Thu, 18 Jun 2026 19:00:00 GMTTranscription accuracy is a given now. What matters is whether a bot announces itself in your meeting and where the recording lives. Here's how to pick, and where each one stops being enough. https://aditya-marin-gasga.pages.dev/writing/agent-eval-playbook/https://aditya-marin-gasga.pages.dev/writing/agent-eval-playbook/Thu, 18 Jun 2026 14:06:00 GMTTest an AI agent in layers: rule-based tool-call checks, step scoring, a dataset from real work, a calibrated LLM judge, and expert review on the high-stakes cases. Five steps and a rubric. https://aditya-marin-gasga.pages.dev/writing/1400-fake-ai-cases/https://aditya-marin-gasga.pages.dev/writing/1400-fake-ai-cases/Wed, 17 Jun 2026 13:00:00 GMTThe same tool that does real work in contract review is getting lawyers sanctioned for citing cases that never existed. One habit separates the two, and it shows where AI fits in legal work. https://aditya-marin-gasga.pages.dev/writing/the-full-bill-for-one-agent/https://aditya-marin-gasga.pages.dev/writing/the-full-bill-for-one-agent/Tue, 16 Jun 2026 19:00:00 GMTThe price on the vendor quote is a fraction of what a production agent costs. Here's the full bill, layer by layer, the unit that should replace cost-per-token, and where the maths breaks. https://aditya-marin-gasga.pages.dev/writing/agent-roi-pilot-production/https://aditya-marin-gasga.pages.dev/writing/agent-roi-pilot-production/Tue, 16 Jun 2026 15:14:00 GMTMost AI agent pilots stall before production: no owned metric, no workflow redesign, no kill criterion. McKinsey shows 62% experimenting but 23% scaling. Here are five questions to ask first. https://aditya-marin-gasga.pages.dev/writing/ai-roadmaps-political-risk/https://aditya-marin-gasga.pages.dev/writing/ai-roadmaps-political-risk/Mon, 15 Jun 2026 15:50:00 GMTA model went from launch to government-ordered shutdown in three days. Everyone measures capability. Few price governability: whether a model can legally run for your users tomorrow. https://aditya-marin-gasga.pages.dev/writing/the-training-data-reckoning/https://aditya-marin-gasga.pages.dev/writing/the-training-data-reckoning/Mon, 15 Jun 2026 12:40:00 GMTThe law on training data is settling into a rule. Training can be fair use, but how you got the material, and whether the output competes with it, can turn the same act into liability. Provenance is now the question. https://aditya-marin-gasga.pages.dev/writing/support-triage-agent-playbook/https://aditya-marin-gasga.pages.dev/writing/support-triage-agent-playbook/Sun, 14 Jun 2026 18:08:00 GMTSet up a support-triage agent the safe way: one narrow ticket category, grounded in your help content, an honest meaning of 'resolved,' and a human review gate on every action. Six steps. https://aditya-marin-gasga.pages.dev/writing/what-is-agentic-ai/https://aditya-marin-gasga.pages.dev/writing/what-is-agentic-ai/Fri, 12 Jun 2026 16:53:00 GMTAgentic AI plans and carries out multi-step tasks on its own. It decides what to do, uses tools, and keeps going until the goal is met. A clear guide to what really ships as an agent in 2026. https://aditya-marin-gasga.pages.dev/writing/ai-cold-email-playbook/https://aditya-marin-gasga.pages.dev/writing/ai-cold-email-playbook/Thu, 04 Jun 2026 00:00:00 GMTAI made cold email easier to write and harder to land. Inbox providers now run AI filters that catch templated outreach. What works uses AI for research and timing, not volume. https://aditya-marin-gasga.pages.dev/writing/ai-sdr-tools-honestly-compared/https://aditya-marin-gasga.pages.dev/writing/ai-sdr-tools-honestly-compared/Thu, 04 Jun 2026 00:00:00 GMTThe AI SDR market is full of bold promises, hidden prices and at least one real scandal. What works, what the categories mean, and how to tell if you need one at all. https://aditya-marin-gasga.pages.dev/writing/best-ai-lead-generation-tools/https://aditya-marin-gasga.pages.dev/writing/best-ai-lead-generation-tools/Thu, 04 Jun 2026 00:00:00 GMTAI lead-gen has split into dozens of tools that each solve one piece. Here's what works, what it costs, and the one thing (data accuracy) that decides whether it pays off. https://aditya-marin-gasga.pages.dev/writing/clay-vs-apollo-vs-zoominfo/https://aditya-marin-gasga.pages.dev/writing/clay-vs-apollo-vs-zoominfo/Thu, 04 Jun 2026 00:00:00 GMTThese three tools get compared as rivals, but they do different jobs. For most mid-market teams the answer is a combination. Here's the breakdown by team size and budget. https://aditya-marin-gasga.pages.dev/writing/honest-ai-sales-stack/https://aditya-marin-gasga.pages.dev/writing/honest-ai-sales-stack/Thu, 04 Jun 2026 00:00:00 GMTThe average B2B team runs 12 sales tools. The best run 6. Here's a lean stack, layer by layer, that covers a real sales motion, and the categories you can skip. https://aditya-marin-gasga.pages.dev/writing/what-is-an-sdr/https://aditya-marin-gasga.pages.dev/writing/what-is-an-sdr/Thu, 04 Jun 2026 00:00:00 GMTAn SDR fills the pipeline. They find and qualify prospects so account executives can close. What the role is, how it differs from a BDR, what it pays, and how AI is changing the job without ending it. https://aditya-marin-gasga.pages.dev/writing/what-is-conversation-intelligence/https://aditya-marin-gasga.pages.dev/writing/what-is-conversation-intelligence/Thu, 04 Jun 2026 00:00:00 GMTConversation intelligence records, transcribes and analyses your sales calls. What it does, how Gong and Chorus compare, and the cheaper tier the big vendors don't mention. https://aditya-marin-gasga.pages.dev/writing/ai-agents-compared-2026/https://aditya-marin-gasga.pages.dev/writing/ai-agents-compared-2026/Sun, 31 May 2026 00:00:00 GMTSeven AI agents to test: Manus, Devin, Operator, GPT Agent Mode, Gemini Workspace, Replit and Claude Computer Use. A shortlist by use case, not test results. https://aditya-marin-gasga.pages.dev/writing/ai-coding-assistants-2026-compared/https://aditya-marin-gasga.pages.dev/writing/ai-coding-assistants-2026-compared/Sun, 31 May 2026 00:00:00 GMTSix AI coding tools and one honest finding: there's no single winner, and the best developers pair two. What each is for, what it costs, and which pair fits how you work. https://aditya-marin-gasga.pages.dev/writing/ai-contract-review-playbook/https://aditya-marin-gasga.pages.dev/writing/ai-contract-review-playbook/Sun, 31 May 2026 00:00:00 GMTAI can catch risky clauses, missing protections and odd language in a contract in minutes. But only if you use it as a first-pass reviewer, not a stand-in for judgement. Here's the workflow. https://aditya-marin-gasga.pages.dev/writing/what-is-an-embedding-really/https://aditya-marin-gasga.pages.dev/writing/what-is-an-embedding-really/Sun, 31 May 2026 00:00:00 GMTA from-scratch guide to what embeddings are, how they're compared, why they make modern search work, and which model to pick in 2026. About 15 minutes. https://aditya-marin-gasga.pages.dev/writing/what-is-mcp-protocol/https://aditya-marin-gasga.pages.dev/writing/what-is-mcp-protocol/Sun, 31 May 2026 00:00:00 GMTThe Model Context Protocol went from a quiet Anthropic release to industry-standard plumbing in under two years. Here's what it is, why every major lab adopted it, and where the hype got ahead of the facts. https://aditya-marin-gasga.pages.dev/writing/best-ai-writing-tools-2026/https://aditya-marin-gasga.pages.dev/writing/best-ai-writing-tools-2026/Tue, 26 May 2026 00:00:00 GMTA shortlist of seven AI writing tools to try, with prices and what each is for. These are my picks to test, not test results. https://aditya-marin-gasga.pages.dev/writing/what-is-an-llm-really/https://aditya-marin-gasga.pages.dev/writing/what-is-an-llm-really/Fri, 15 May 2026 00:00:00 GMTA from-scratch guide to how large language models work: tokens, attention, the prediction loop, and what to make of it all. 12 minutes. https://aditya-marin-gasga.pages.dev/writing/cut-writing-time-70-percent/https://aditya-marin-gasga.pages.dev/writing/cut-writing-time-70-percent/Fri, 08 May 2026 00:00:00 GMTA four-prompt workflow for first drafts that avoids the usual AI tells. The prompts, and one before-and-after example.