Our writing on AI in regulated work.
Posts on AI, expertise and how regulated operations decide, newest first. Every post we’ve published is in the archive below.
Our latest posts.
We publish most working days: what we learn in client work, in our sectors and in our own research.
What agents need to surviveA geologist and seven companions spent five weeks driving across the Arabian desert.
Does your team know enough to check AI?In 1962, a woman at a desk in Virginia spent a day and a half doing sums a computer had already done.
Can the model forget?There’s a tree on an island near Seattle that has swallowed a bicycle. A boy left the bike in the woods in 1954 because he didn’t like it, and never gave it another thought.
Decision fatigue in AI workflowsEvery so often, an airport scanner deliberately shows the screener a gun that isn’t in the bag.
Start with the posts for your role.
-
Operations
Where to begin on the floor
-
Quality
What quality will ask
-
Technical and digital
How it fits your estate
-
The person who signs
What it is worth
All the AI news that matters each week.
SNI is our weekly newsletter, covering tech, biopharma, medtech, advanced manufacturing and insurance.
Everything we’ve published.
Every post, by year and month, newest first, at the address it has always had.
2026
November
October
- SNI: Week 44
- The knowledge that isn’t in the manual
- Managing personal AI at work
- Human in the loop
- LBI Awards 2026News
- The leap of faith
- SNI: Week 43
- Applied AI and knowledge sovereignty
- Building careers in SMEs
- Training AI when experts disagree
- SNI: Week 42
- The International AI Summit 2026
- The AAIF guide to CHAPNews
- Melissa Proxenos and Corey Duffy nominated at the AI AwardsNews
- The need for speed
- SNI: Week 41
- The 2027 Generative AI for Medtech playbook is hereNews
- Meissner shortlisted for Pharma Industry AwardsNews
- Generative AI for Medtech playbook launch: Dublin, 7 OctoberNews
- How do you govern a million AI agent decisions?
- SNI: Week 40
September
- Meet us at EBDVF 2026News
- Are your beams on?
- SNI: Week 39
- Who owns the agent’s memory?
- Can AI duplicate your expertise?
- When your AI no longer needs you
- SNI: Week 38
- Dr Arsalan Shahid at AGNTCon and MCPConNews
- New partners, old habits
- SNI: Week 37
- EBDVF 2026: AI, data and human responsibilityNews
- Meissner makes the LBI Awards shortlistNews
- Meissner: case study
- Can an AI feel time?
- SNI: Week 36
- Shane Owens at ISPE Ireland’s GAMP SeminarNews
- Dr Arsalan Shahid at EBDVF 2026News
- Open-weight models come home
August
- SNI: Week 35
- When AI doesn’t know what it doesn’t know
- Join our introduction to human-agent collaborationNews
- SNI: Week 34
- We need a new name for AI
- Capturing tacit knowledge for AI agents
- Syncing intelligences
- SNI: Week 33
- What does your best operator know?
- AI can solve more. But what should it solve?
- The model is no longer the unit
- AI agents are starting to cooperate
- Not everything worth doing is a task
- SNI: Week 32
- CHAP gets its moment on the main stageNews
- Teaching AI agents where the line is
- The strange similarity between memory and AI
- SNI: Week 31
July
- Open-weight AI is entering a new era
- What really makes an AI agent work
- Can AI be wise?
- SNI: Week 30
- Why AI agents need better guardrails
- AI safety after the Hugging Face incident
- Medicine between appointments
- The trait that predicts AI success
- Genius of the wasted trip
- SNI: Week 29
- How to invent a new colour
- Join our CHAP webinar August 11News
- What the record doesn’t teach
- The secret to successful AI agents isn’t the code
- What exists when nobody is working?
- SNI: Week 28
- The next AI battleground: model routing
- SNI: Week 27
- AI reskilling starts with curiosity
- Learning loops & intelligence capital
- CHAP now speaks MCPNews
June
- The illegible parts of thought
- OpenAI’s 5.6 enters the spotlight
- SNI: Week 26
- Who creates demand for AI?
- AI-native companies don’t add layers
- SNI: Week 25
- A guide to why the integration of digital intelligence is becoming ever-more vital
- Brightbeam on Newstalk’s Breakfast BusinessNews
- Including the responsible human
- Kieran Ivers at Transformation Summit 2026News
- SNI: Week 24
- Thank you, CorkNews
- Your turn
- Where to start AI in manufacturing
- Could AI ‘solve’ viruses before maths?
- SNI: Week 23
- Join us at Transformation SummitNews
- Brightbeam joins the BDVANews
- Brightbeam hiring day in CorkNews
- AI, dreams, and the future of learning
May
- SNI: Week 22
- The bigger AI risk for Ireland isn’t jobs – it’s tax
- The real AI job problem isn’t what you think
- SNI: Week 21
- Google’s quiet AI gamble
- The new AI divide
- Cork engineering recruitment day 9th JuneNews
- Why AI still isn’t general intelligence (yet)
- SNI: Week 20
- Insurance and AI: the slow start is over
- Ireland’s AI shift is bigger than the headlines
- The AI context window arms race is missing the point
- Computation has a body
- SNI: Week 19
- Who decides when AI is confident enough to act?
- SNI: Week 18
April
- Join us at BioPharmaChem Impact 2026News
- AI exposes the next bottleneck
- What if AI is declared conscious?
- SNI: Week 17
- Who is the model working for?
- Are skills & sense-making ebbing away?
- Key insights from Manufacturing the FutureNews
- SNI: Week 16
- Rory Dunne on digital transformation & AI in manufacturingNews
- Is cyber risk still insurable in the AI era?
- Why enterprises might not capture the value
- Is AI investment optimising the wrong path?
- SNI: Week 15
- Why enterprise security must evolve for agents
- Why slow AI adoption can be a strength
- Why today’s AI rules will define the future
- SNI: Week 14
- Glen Keane on the problem of agentic driftNews
March
- Why organisational speed now defines value
- SNI: Week 13
- The hidden layer powering AI progress
- When AI gets the right answer for the wrong reason
- AI exposure does not mean job loss
- SNI: Week 12
- The real path to enterprise AI adoption
- The map has the wrong scale
- Why analogue thinking still matters in the AI era
- SNI: Week 11
- AI just solved another scientific domain
- How enterprises adopt AI: Part 3
- Women in TechNews
- SNI: Week 10
- When smarter AI becomes less predictable
- How enterprises adopt AI: Part 2
- Why most companies already have the data they need
February
- SNI: Week 09
- Two exponentials driving the next AI wave
- The emotional side of AI adoption
- How enterprises adopt AI: Part 1
- Digital Business Ireland AwardsNews
- SNI: Week 08
- Part two
- Even the peanut looks expensive
- How AI shows us psychological safety isn’t enough
- SNI: Week 07
- AI isn’t reducing work, it’s intensifying it
- The capability’s here. Is enterprise ready?
- Good luck Melissa ProxenosNews
- Learning time: a taxonomy
- The three skills of the agentic era
January
- Climbing to AGI
- Critical depth: when more layers unlock new capabilities
- The AGI stare-off
- Organisational speciation
- What’s it like to work at BrightbeamNews
- The organisational phenotype
- Closing the AI capability overhang
- The choices ahead
- The real bottleneck in AI-powered teams
- The core tension
- The new power shift in product creation
- The Productivity Clock
- You’re not aloneNews
- What will this year be, really?
- AI in 2026: what the experts predict
2025
December
- The end of interfaces: AI beyond screens
- Ground truth by example
- What the hypegeist missed: #1
- Why GPT-5.2 feels different
- The glass slipper effect
- Is OpenAI losing ground to Google?
- AI implementation: when foundation models outperform fine-tuning
- Is ambient AI the winning play?
- Why speed and iteration define AI success
November
- Three AI tools worth testing this weekend
- Medtech Rising 2025News
- AI: no need to clean your dirty data
- Why we still trust humans more than machines
- What Nvidia’s earnings mean for AI markets
- Plan for strategic integration
- The case for a national AI flywheel
- Upskill teams to build and govern AI responsibly
- We’re building alien minds. Now what?
- Google: waging a three-front war
- Embed cross-functional collaboration
October
- Build enterprise-ready foundations
- Microsoft: beware the big Brussels unbundling
- How a tick’s world explains AI alignment might be impossible
- Context is king: mapping the battlefield
- From strategy to scale: Ireland’s AI advantage
- Always design for human-in-the-loop governance
- OpenAI says: context reigns supreme
- Context is king. But why?
- The AI flywheel in action: lessons from Citi
- Start with high impact, low risk use cases
- The new era of digital intelligence
- Building an AI flywheel without betting the farm
- The next leap in AI: will built-in governance temper the risks?
- World models: what they are and where they might matter
- Success starts with the flywheel