A look back at my notes, doodles and conversations with clients and associates. How is it half-way through 2026 already?
- The recurring challenge: Practitioners are overwhelmed by shifts in search, AI copy fatigue, and pressure to deploy autonomous AI tools without clear processes.
- The collective lesson: Sustainable visibility comes from building authoritative website assets, applying strict briefing discipline, and separating true generative tasks from basic rule-based automation.
- The practical first step: Run a mystery shopper audit on your website using structured prompts, then strip out synthetic jargon from your copy.
Looking back across all the mentoring sessions, workshops, and catch-ups I hosted this month, it does feel as though the early honeymoon with artificial intelligence is well and truly over, and people are starting to ask much harder, much more practical questions.
How do we stay visible when Google turns search into a direct answer engine and keeps people on the results page?
How do we automate routine admin without losing our sanity or alienating our clients?
And how on earth do we keep our marketing human when the internet is being flooded with AI content that everyone can spot a mile off?
When I did my very first website project back in 1994, I didn’t even know anyone else with an email address, so I literally emailed myself to see if the thing worked. Thirty-two years later, the technology is infinitely more complex, but our core task as practitioners hasn’t changed one bit. We are still in the business of building trust, serving people, and making complex ideas accessible without losing our human voice.
Master the assistant before you release the agent
Lately, I’ve had a lot of conversations about AI agents. Business owners tell me they are frustrated by repetitive tasks and want an autonomous agent to handle client follow-ups or manage renewals. I get the appeal, I really do. But I fear that lot of people are trying to jump straight to autonomous agents without ever paying their dues and mastering the basics of Large Language Models or LLMs.
Think of ChatGPT or Copilot as a village full of specialists. If you hire a real person to join your team, you don’t hand them the keys to the front door on their very first morning and walk away. You give them a clear brief, you observe how they work, and you check their output. A digital assistant is no different. In fact, it needs far more precision because it cannot fill in logical gaps the way a human colleague would.
Before you let any platform communicate with the outside world, you need to master the discipline of briefing. That’s why I always teach the ABCD framework: define who the tool is acting as, break down the request step by step with numbers, specify the exact content format, and supply detailed context.
When I was working through workflow ideas with an IT support provider, we looked at Microsoft Copilot Office Agents and Copilot Notebooks. My advice was simple: keep it in a closed environment first. Map your manual process out on paper, test the prompt logic internally, and check the output yourself.
Pro Tip: Start an internal AI Brief Library. Save your best-performing briefs and prompts in a shared folder so your whole team works from the same tested logic.
Pascal-ism: Do not give a digital agent keys to the front door until it has proven its reliability as a quiet assistant in the back room.
Implementation Roadmap:
- Map your entire workflow manually on a piece of paper, marking every decision point.
- Test the brief inside a private workspace like Copilot Notebooks or NotebookLM.
- Run test data through the workflow and review the results personally.
- Release the automation only after the logic consistently passes your manual review.
Key Takeaway: Treat every AI tool like a new team member; establish strict briefing standards before granting any operational autonomy.
Surviving zero-click search by building a trust engine
Google held its annual conference recently, and I spent hours watching and sifting through the updates. The direction is clear: search is being completely turned upside down, with AI Overviews and more conversational experiences on Google Maps and YouTube becoming new gateways to the world’d information. And we have the future arrival of shopping agents allowing consumers to complete a purchase right from the AI summary pane.
This creates a strange situation for website managers. Your web traffic numbers might drop, yet your phone keeps ringing, you online sales or footfall stays steady because your website content fed the AI summary that gave the user their answer. Your website is no longer just a destination for human browsers; it is an authoritative reference manual for search crawlers.
To remain visible, your site needs to satisfy four distinct visitor personas:
- The Investigator persona needs a clear explanation of your exact process and methods.
- The Relationship Builder persona wants to see the real people behind the business, a rich About section, and clear values.
- The Dreamer persona looks for end outcomes, case stories, and visual transformation.
- The Negotiator persona looks for trade memberships, accreditations, and guarantees to reduce their feeling of risk.
If you take the trouble to build out comprehensive FAQ sections and expand your About page with real stories, you strengthen your Experience, Expertise, Authoritativeness, and Trustworthiness (EEAT) score. Search engines favour sites that provide depth over those that offer brief, shallow summaries.
Key Takeaway: Restructure your website content to act as an authoritative reference source that will allow AI Overviews to quote and cite extracts.
The 4 Ms content framework and the social post upgrader
In a mentoring session with a trade specialist and later with a retail group, the same complaint came up: “Pascal, I am working so hard on social media, but I don’t have time for my website.” People spend hours creating posts that disappear down a social feed in forty-eight hours, while their main website sits untouched for six months.
I always tell people that I love the four Ms of content marketing because they give you an instant construct for inspiration:
- Marketing content directly promotes your core offers, case studies, and testimonials.
- Mentor content answers practical client questions, FAQs, and how-to guides.
- Messenger content shares your reaction to industry news, regulatory changes, or trends.
- Merchant content offers commentary on techniques, physical tools, or digital technologies in your field.
When a short post resonates on social media, don’t let it die there. Run it through what I call the Social Post Upgrader. Take that short LinkedIn or Facebook post, bring it into Claude, and instruct the platform to expand it into a structured, long-form website article. Then, record a quick sixty-second “Too Long; Didn’t Read” (TL;DR) video overview to place at the top of the page.
Key Takeaway: Repurpose short-lived social updates into permanent website assets to feed search crawlers and reassure site visitors.
Putting AI to work as your mystery shopper
When you own a business, it is almost impossible to look at your own website objectively. You’ve been staring at those pages for years, agonising over every word. You are the least qualified person to judge what a first-time visitor experiences.
In the past, I used to hire market research teams or set up video mystery shopping, all rather complex and can cost a fortune. Today, you can turn platforms like Gemini Notebook (previously NotebookLM) or even ChatGPT into your own objective mystery shopper.
Upload your web pages, client profiles, and service overviews into Gemini Notebook. Set the persona of a specific customer type and instruct the tool to answer three practical questions:
- What is working well on this page?
- What is unclear or frustrating?
- What essential information is missing altogether?
During an open Q&A session with local retailers, a business owner ran his site through this exercise and immediately discovered a broken contact link that was sending client inquiries to a phone number instead of an email. Once the platform gives you its initial review, push it further by asking: “Is that the best advice you can provide? What critical questions have you forgotten to ask me?”
Key Takeaway: Use AI to simulate client feedback so you can identify website friction before real prospects leave your site.
Stripping out AI writing signals to preserve human voice
We need to have an honest conversation about the state of AI writing, and by extension AI content. Readers have developed an incredibly sharp filter for AI or synthetic copy. The moment someone sees words like “unlock”, “elevate”, “delve”, “tapestry”, or “meticulous”, and don’t get me starting with em-dashes, they know an AI generated the text, and they tune out.
Real human communication relies on conversational rhythm, natural pauses, and what I like to call “elegant clumsiness”. When we talk to clients, we don’t speak in perfectly polished corporate dissertations. We tell stories, we pause, we ask questions, and we vary our rhythm.
To keep your writing authentic, apply an anti-AI editing sweep to every draft:
- Strip out banned vocabulary such as elevate, unlock, unleash, delve, robust, pivotal, and meticulous.
- Eliminate long em dashes and replace them with standard commas, full stops, or plain parentheses.
- Vary your sentence lengths naturally and avoid rigid paragraph structures.
- Read your copy out loud to verify it sounds like a real professional talking to a colleague.
Key Takeaway: Remove generic AI buzzwords from your drafts and create a library of your own content you’ve compiled without AI.
Disentangling AI from basic automation
During a 1-1 session with a storage facility owner, we spent time exploring a common misunderstanding. He wanted to use generative AI to track client expiry dates in a spreadsheet and issue renewal reminders.
I had to explain that using a generative language model for fixed, rule-based admin is over-engineering the problem. If a task relies on predictable triggers, such as a date approaching in an Excel column, standard software tools like mail merges, automated inbox rules, or simple CRM workflows are far cheaper, faster, and more reliable.
Digital tools can do a perfect job of making things more complex and nowhere near as productive if we pick the wrong option. Use generative AI for unstructured, creative work: brainstorming, summarising meeting calls, drafting complex responses, or acting as a sparring partner. Keep your operational admin firmly anchored in clean software rules.
Key Takeaway: Use standard software rules for predictable administration, saving generative AI for tasks that require creative drafting or research synthesis.
Final Thoughts and Notes
Reflecting on June’s field notebook entries, my biggest takeaway is that successful marketing is returning to fundamental human habits. The frantic rush to grab every shiny tool is slowing down, and practitioners are realising that authenticity, clear processes, and genuine customer care are what actually drive long-term business.
Generative tools are extraordinary when used as quiet assistants for research, editing, and deep reflection. But they cannot replace the personal touch, the local context, and the real-world experience that you bring to your clients.
As you plan your activities for the months ahead, look at your existing content and workflows. Where can you strip away AI jargon, simplify your admin, and let your natural practitioner voice shine through?
I would love to know what your own field reflections are revealing this month. Feel free to get in touch to share your experiences or discuss how you are getting the balance right with AI.
To your success!
Pascal
