🔮Into 2026: Memia state of the nation

Putting the periscope up in preparation for another year of exponential AI

Kia ora and I hope you’re enjoying 2026 so far!

I did my usual Wil-E-Coyote thing, running off a cliff with legs spinning at the end of last year… the quantity and pace of news on my beat just keeps on increasing and — despite my ongoing efforts to automate the sensemaking of it all — it’s getting harder and harder to keep up¹. (Case in point: the number of new benchmark-breaking frontier AI models that came out in December set a new record that I’m sure will be broken within only a few months…)

A few weeks off drinking from the firehose has put a some welcome perspective on things… I thought I’d gather some thoughts on the current moment and then look forward to where things are headed with Memia in 2026.

🤯Memia Newsletter, Year 7

Amazingly the weekly Memia newsletter on Substack is now in its 7th year! The first post Memia 2020.01 went out on 15th January 2020. That post included only about 10 links compared to the >150 most weeks now.

As I always say, it’s a privilege that >3000 readers let me into your inbox every week to receive my thinking-out-loud about tech and other topics. Thank you for reading!

There’s been a lot of evolution and growth in the newsletter over the last 6 years. Some things that have changed:

• Quantity of signals scanned I’m now regularly scanning my way through over 2000 news articles per week from >30 tech / science RSS news feeds, plus whatever my LinkedIn, X, BlueSky, Farcaster, Google News algorithms see fit to feed me. Each newsletter now contains maybe >150 summarised stories with links… it’s an unwieldy beast but people continue to value receiving it each week and I know that there’s a hardcore group of you go through the whole thing in <30mins every week!🫡

• Taxonomy / story categorisation the structure of the newsletter continues to evolve as tech trends ebb and flow — in particular riding the AI wave for the last few years, much of the newsletter is now given over to AI industry news / releases / research … while the [Weak] signals section covers pretty much every popular category of emerging technology. Plus the Zeitgeist section keeps me honest about significant real-world events that *somehow* don’t turn up in mainstream media. As of the end of 2025, each story is now autocategorised into 11 sections and 90 subsections.

• Auto-summarisation it’s taken over a year to get there but the Sensorium toolset I’ve built progressively over 2025 now parses, analyses and auto-summarises stories for me and doubles up as a helpful research tool on strategy projects. Obviously there’s a balance here on quality and a hallucination risk - but the >10X productivity boost is clear in the metrics above.

• Editorial depth over the last year the main compromise has been going wide and shallow across a larger scope of content. My bandwidth for Sunday “Mind Expanding” / “Strategy Note” deeper thinking / posting has been usurped by Sensorium vibe-coding sessions to build a more automated workflow.

• Time taken as the newsletter scope/ length expanded over the last 2 years, the amount of time it took to compile kept expanding. At one point I was putting in >16 hours a week(!) … all completely worthwhile from a learning perspective, but financially unsustainable as I still need to do my day job too! It also meant I didn’t have time to write on deeper topics. At the end of 2025 this is back down to around 5-6 hours with Sensorium / AI automation and I think I can bring it down further while keeping the signal/noise quality.

• Competitive landscape when I started out, it was really only stalwarts such as Azeem Azhar (Exponential View), Benedict Evans and Ben Thompson (Stratechery) who were pumping out weekly high-signal tech/AI-related newsletters. Since then, the number of new AI industry / tech newsletters has grown considerably— now I receive at least 50 AI email newsletters per week from various sources. Many of these are now using AI summarisation in their own house style as well. Memia is by far the most comprehensive in breadth and length as far as I can tell — however, striking the right balance between signal / noise is always the constant challenge. Also the Substack platform’s limitations are starting to hit and I’ve noticed new sign-ups plateauing as achieving cut-through gets harder.

  • I have an instinct for Blue Ocean strategy… so the increase in competitive offerings is pulling me elsewhere…

💭What’s Memia for?

• The newsletter has grown out of me “learning out loud” to develop my own situational awareness on what’s happening at the frontier of AI and tech. Writing the newsletter each week has become my process to identify, analyse and absorb key signals of what’s going on — and then share that thinking with readers. Being right up to date also feeds into my public speaking and AI strategy engagements — and generates very stimulating contacts / conversations which often leads onwards to other interesting work opportunities.

  • Over the last few years my own personal “objective function” has shifted away from “make money” towards “maximise incremental learning”. (Bizarrely, through a mechanism I don’t yet understand, this also helps the “make money” objective… 😅). These days, if I could afford to spend every waking hour learning new stuff, I would.

• If I was to condense the Memia newsletter’s value proposition to readers into one sentence, I like to think it provides a particular refined “taste” to stimulate intelligent, curious minds across a regular set of narratives / news beats:

  • AI/Tech Signals:

    • Frontier AI and technology advances which are out of the lab but not yet in mainstream awareness

    • Technological sovereignty - personal, firm and nation-state

    • Trends where new technology challenges society / establishment norms

    • Industry and investment trends that signal who the major players will be and how markets are restructuring

    • Frontier developments in non-AI emerging technologies: robotics, drones, crypto, 3D printing, nanotech, biotech, quantum tech…

  • Mind Expanding “Philosophy of AI”:

    A particular pet interest of mine…

    • Thinking about future society with AGI, superintelligence and pathways to/beyond transhumanism.

    • Mind augmentation / enhancement technologies

  • Zeitgeist The “real” world:

    • Climate change, biodiversity collapse, demographic change, geopolitics that’s not covered in mainstream editorial policies.

📰Editorial stance

• Independent Above all, I fiercely value thinking and writing freely and not self-censoring to align with commercial / establishment interests. As such, Memia is an ad-free, reader-supported publication — thanks to those of you who are paid subscribers and help me to maintain my independent voice. (And stay caffeinated, natch.☕)

• Outsider perspective I’m physically based in Aotearoa New Zealand where there’s *very little* discussion in mainstream (/any) media about these topics. I often describe myself as a “bungee” stretched between NZ and Silicon Valley / China as AI accelerates away… but here they just keep pouring more fertiliser onto grass to increase the number of [environmentally-catastrophic] cows. So it’s an outsider view from the South Pacific, based on what I can glean from the public domain internet from far away. Luckily, there are lots of interested outsiders around and hopefully I provide an objectivity missing from other more *patriotic* sources…

• Neutrality

  • Tech neutrality: most raw technologies can be viewed as neutral tools which can be used for good or malign outcomes. Judge the user, not the tech.

  • Descriptive ethics, not prescriptive: on most issues I prefer asking rhetorical questions rather than taking hard ethical stances.

  • Explicitly taking an independent, neutral position on bipolar tech ecosystems. tech is tech wherever it was developed. *Obviously* US values are embedded in closed US AI models and Chinese values are embedded in “open” Chinese models. (By definition, the only way to achieve neutrality is to go fully local and open-source… but this path is more challenging and still some way behind the frontier.)

• Sourcing Nearly all references are sourced from the public domain - I try to go to the original source of stories (particularly scientific papers) — but over time I have found that there are trustworthy, reliable popular sci/tech newswire services out there such as Phys.org, Interesting Engineering, TechXPlore, Nanowerk, New Atlas, The Register, Rest of World (plus obviously TechCrunch, VentureBeat, The Verge provide in-depth coverage of Silicon Valley — I’m still sorting through reliable sources on Chinese and EU tech. Plus I’m wired into X, LinkedIn, Bluesky, Farcaster and Google News algorithms which after many years of training — unnvervingly — often know my taste better than I do.

  • I’m acutely aware that I’m operating at an interesting moment in the “news” industry: legacy news business models are long dead, who knows what will replace them? There’s an expanding commercial grey area between [using AI to automate] reading/summarising someone else’s journalism designed for ad-supported Web 2 business model. Meanwhile the large commercial AI model companies are (with or without permission) absorbing every token that’s ever published into their next training run. I’m participating at the edge of this particular wave and have a particular position to frame “news” as part of a set of Epistemic Technologies which support a global sensemaking commons. Keeping a watching brief on how this develops in 2026…

🔮Memia in 2026

So here’s a list of initiatives / projects I’m planning to work on with Memia this year:

• Surfing the AI frontier wave fundamentally the only way you learn new tools is by using them to get stuff done — increasingly my happy place is exploring the latest releases, AI models, tooling and applications every chance I get and building new skills right out at the edge (which in turn informs my analysis.)

  • AI capabilities just stepped up another noticeable notch at the end of last year with the release of Claude Opus 4.5 and GPT-5.2 Codex (and Gemini CLI v. next is coming soon I hear…) So if you’re not already using Claude Code or similar to vibe code your own app prototypes / PKM / move your own todo list backlog forward, then get to it! These will be in-demand skills in 2026, if nothing else.

• Weekly newsletter I’m going to lean into AI-generated summarisation for the weekly scan email. I’ll keep honing the format and include “Picks of the week” which will be human-written — but I think that the scanning service is now “table stakes” and as observed above Memia is operating in an increasingly competitive environment. (I know some readers are subscribing to other newsletters and/or experimenting with simple scheduled AI prompts, this will only get better):

  “Give me a newsletter format summary of the main developments in AI and emerging tech in the last 24 hours”

  (Simple, except: you have no idea what the agenda is of your closed-source AI “feed editor”.)

• Broader signal-scanning I think with the AI tooling now in place I should be able to at least 10X the number of sources / signals scanned throughout 2026 — likely these will be available to subscribers through the Memia website rather than the newsletter. Maintaining the signal-to-noise ratio will be key.

• Weekend Mind Expanding / Strategy Note posts on Substack I plan to get back into these, lots of things to discuss and develop my thinking further. Hopefully I can get into a better rhythm than last year as the weekly newsletter automation frees me up.

• Personal technological sovereignty 2025 was the year when I re-learned to code, installed Linux on my old PCs, became more familiar with the Linux command line and started replacing many of my SaaS / Cloud subscriptions with personalised (vibe-coded) apps that I host myself. The capabilities and opportunities are only getting larger now… by end 2026 I aim to be running far more of my tech on an open-source, self-developed, self-hosted, self-sovereign stack.

• Personal algorithmic sovereignty In particular, my personal Sensorium project has grown arms and legs in 2026. (It all makes sense inside my head but will likely appear as a mish-mash of features to anyone else!) Its key purpose remains, however: to provide a personally controlled editorial algorithm which filters and optimises what information reaches my mind, rather than leaving my brain wide open to closed-source black-box algos owned by oligarch-controlled tech companies trying to soft-hack my beliefs with opaque agendas.

  • I have limited support bandwidth but I’m keen to collaborate with other individuals who are interested in testing / building out from this. Reach out.

  • To that end, in 2026 I’ll open-source the code once it’s tidied up enough to share… (even though AI wrote almost all of it…!) If so inclined, you will be able to run it locally on your own machine and curate your own research database / newsletters…

  • It also increasingly supports meta-analysis across all topics covered by Memia since the outset. (For example AI Narratives, AI Scenarios…) Watch this space to see how powerful this may be.

• AI strategy consulting I continue to work with clients using Memia’s simple, effective AI strategy framework for enterprises, based upon three axes of AI Automation, Augmentation and Diffusion. Reach out if you want to talk more.

  

  Routes to Mt. Diffusion

• AI portfolio management toolset Alongside the framework, there’s now an AI portfolio management toolset to visualise, plan, prioritise and manage Enterprise AI investment. Deceptively simple:

  

• Public speaking Get the most up-to-date AI briefing presentation or keynote for your event. Details of how to book me here.

• Fast Forward Aotearoa monthly newsletter - I plan to revive the monthly signal-scanning newsletter for New Zealand-based tech news in 2026. It’s election year, perhaps tech will be on the policy agenda?!? Sign up here if interested.

• Another book? While nomadding around Europe last year I bravely bought the domain ffwd.world … not sure I’ve got the stomach for this right now, though.😅

• Something else? Reach out if you’ve got an interesting project you think I could contribute to!

Thanks for reading and your ongoing engagement. Onwards, into the breach…

Ben

1. One casualty of last year’s workload was that I didn’t manage to put out my annual “Year In Review” post — just not enough gas left in the tank, let’s see what we can do in 2026.