The word for 2026? "Headless"

Hello,

This is Simon with the latest edition of The Weekly. In these updates, I share key AI related stories from this week's news, list upcoming events, and share any longer form articles posted on the website.

I made a small claim on LinkedIn at the end of last week, in that the word for the rest of 2026 will be “headless”. You might have already heard Salesforce CEO, Mark Benioff, use this term when referring to some of their recent product developments, but what does it mean? Headless refers to working with a software platform without using its user interface.  It’s been possible to work in this way for many years through using an API system, but these days, AI systems can help anyone do this. I’ve mentioned MCP systems quite a few times in this newsletter, and it’s these tools that allow an AI chat interface to ‘talk’ to our everyday systems. As more and more pieces of software add their own MCP capabilities, it means someone could spend a good part of their working day just in a AI tools like Claude or ChatGPT, and using normal language to ‘ask’ a system to do something. This is also where agents come into play. As of today, it’s entirely possible to ask an AI chatbot to go and do your grocery shopping, without ever having to use their website.

This week, I had a small taste of this myself. Rather than opening Notion to find a particular newsletter idea I'd jotted down a few weeks ago, I simply asked Claude to pull it up for me. A few seconds later, there it was, summarised back to me in the chat window. No tabs, no scrolling, no searching. It's a small thing, but it's the sort of small thing that starts to add up over a working week.

So why am I confident that 2026 is the year "headless" goes mainstream? A few reasons. The big enterprise players — Salesforce, Microsoft, Google — are now actively building for this way of working, rather than treating it as an experimental side project. The MCP standard that lets these systems talk to AI tools is maturing quickly, and more software providers are adopting it every month. And perhaps most importantly, the AI chat interfaces themselves are finally good enough that asking in plain English actually works most of the time.

The interesting question for the rest of us isn't really whether this happens, but what it means for how we work. If the dashboards and interfaces we've spent years learning start to fade into the background, the skill shifts from knowing where the buttons are to knowing how to ask clearly — and how to spot when something's gone wrong. That's a topic I suspect I'll be coming back to plenty over the coming months.

Last month, 48% of visitors to documentation sites across Mintlify were AI agents, not humans.

Claude Code, Cursor, and other coding agents are becoming the actual customers reading your docs. And they read everything.

This changes what good documentation means. Humans skim and forgive gaps. Agents methodically check every endpoint, read every guide, and compare you against alternatives with zero fatigue.

Your docs aren't just helping users anymore. They're your product's first interview with the machines deciding whether to recommend you.

That means: clear schema markup so agents can parse your content, real benchmarks instead of marketing fluff, open endpoints agents can actually test, and honest comparisons that emphasize strengths without hype.

Mintlify powers documentation for over 20,000 companies, reaching 100M+ people every year. We just raised a $45M Series B led by @a16z and @SalesforceVC to build the knowledge layer for the agent era.

Make Your Docs Agent-Ready

In this section, I’m going to bring to you a real world example of AI use. This week, we look at easyJet’s Airbus Skywise system.

On the same day last week, OpenAI and Anthropic both announced they were launching dedicated enterprise deployment businesses to compete directly with the likes of McKinsey, Accenture, and IBM. OpenAI finalised a $10 billion joint venture with 19 investment firms — led by TPG, with Bain Capital and Brookfield as co-lead partners — valued at $10 billion, according to Bloomberg. Hours earlier, Anthropic announced a separate $1.5 billion venture backed by Goldman Sachs, Blackstone, and Hellman & Friedman, specifically targeting mid-sized companies owned by private equity. Fortune described Anthropic's move as "taking a direct shot at the consulting industry."

Why it matters: The major AI labs are no longer content to sell API access and let consultancies capture the implementation margin. They are building their own professional services arms — with capital, distribution, and investor relationships to match. For organisations currently relying on traditional consultancies to guide their AI strategy, the competitive landscape just got more complicated.

Companies that have moved quickly to cut staff on the expectation that AI will absorb the workload are discovering it isn't that simple, according to The Register (6 May). Analysts and researchers are warning that organisations restructuring before AI systems are actually production-ready are creating operational gaps that are proving expensive to close. The pattern is consistent: leadership pressure to reduce headcount meets vendor enthusiasm about AI capability, resulting in commitments made before systems have been properly tested at scale. The 80%+ enterprise AI project failure rate reported by RAND — and a GenAI pilot abandonment rate now running at 95% — provides the broader context for why these restructuring bets are going wrong.

Why it matters: The liability here is real. Cutting experienced staff ahead of AI deployment, and then finding the AI isn't ready, leaves organisations exposed on both sides — reduced human capacity and underperforming technology. The cautionary message for business leaders: treat AI-driven headcount decisions as a business risk, not just an efficiency opportunity.

The EU AI Act's most significant compliance deadline hits on 2 August 2026, when requirements for high-risk AI systems become enforceable. The rules apply to any organisation — regardless of where it is headquartered — if its AI systems affect EU residents. High-risk categories include AI used in hiring decisions, credit scoring, healthcare, and critical infrastructure. Non-compliance carries fines of up to €35 million or 7% of global annual turnover. Law firm Holland & Knight noted in April that many US companies are significantly under-prepared, and while some deadline extensions have been proposed under the EU's Digital Omnibus package, legal advisers are warning against relying on them materialising.

Why it matters: This is no longer a future compliance problem — it's a twelve-week deadline. Organisations using AI in any process that touches EU customers or employees should be reviewing their systems now. The categories most at risk (hiring, lending, HR scoring) are exactly the areas where enterprise AI adoption has been moving fastest.

Thanks for reading, and see you next Thursday.

Simon,

Was this email shared with you? If so subscribe here to get your own edition every Thursday.

Enjoying Plain AI? Share it and get a free gift!

If you find this newsletter useful, why not share it with a couple of friends or colleagues who would also benefit? As a special thank you, when two people subscribe using your unique referral link, I’ll send you my "Exclusive Guide: Supercharge Your Work with NotebookLM." It’s a practical, no-nonsense guide to help you turn information overload into your secret weapon.

You can also follow me on social media: