Dan Bennett

For everyone who is playing around and trying to get agentic frameworks and systems in production, here is a great cheatsheet on Context engineering. https://lnkd.in/eqjTNbNU And the 12 Factor Agents. Patterns let me remember cohesive concepts, so I really appreciate them https://lnkd.in/e7GHqYUq

11

Last night at Founder Institute night out in Chicago, I met Wendelin Dunlap MSD, M.Ed., CDE, CPC, a user on ChaChing Social and an expert on building accessible products. I was showing her our blog reader on ChaChing Social and asked how we can make it more accessible for people with dyslexia, she mentioned there's an actual font called the Dyslexia Font, and that it's proven to make reading easier for people with dyslexia. So I went home, dived into a rabbit hole, had a deep conversation with Gemini and found these insights: - Line Spacing (Leading): Aim for 1.5 (150%). This provides enough "breathing room" between lines so the reader doesn't accidentally skip a line or reread the same one. - Letter Spacing (Tracking): At least 0.12 times the font size. For example, if your font is 12pt, the letter spacing should be at least 1.44pt. - Word Spacing: At least 0.16 times the font size. Words should be clearly distinct from one another to prevent them from merging into a single block of text. - Paragraph Spacing: At least 2 times the font size. This helps visually "chunk" information into manageable pieces When you read our blogs, you can switch on "Reader Mode" on the top right-hand side, and you can read our blogs in a more accessible way. Does this make reading much easier? Are there other ways we can make it easier?

17

2026 is shaping up to be another big year! With my thesis now submitted, I get to do the part I've been looking forward to most - getting out there and sharing what I've learned. I'm thrilled to have been invited to speak at and attend these events over the coming months: - IT Revolution Enterprise Forum & Enterprise AI Summit - San Jose, April 7-10. An invite-only forum followed by a plenary talk at Gene Kim's conference on the invisible gap my research found between productivity and experience - and why leaders need new vocabulary for the work their engineers are actually doing now. - Copenhagen Symposium on Human-Centered AI in Software Engineering - Copenhagen, May 5-8. A working symposium organised by Prof. Dr. Daniel Russo, bringing together researchers and practitioners to shape how AI is adopted in software engineering. Looking forward to contributing to the workstreams and reconnecting with familiar faces from the academic world (a space I love keeping a foot in). - Web Directions AI Engineer Summit - Melbourne, June 3-4. John Allsopp is bringing the renowned AI Engineer conference series to Australia, and I'll be delivering a keynote on what the shift to AI-assisted engineering actually *feels* like, and what you can do about it. Three events, three continents, three very different audiences. I genuinely can't wait to see how the same ideas land in each room. Now if you'll excuse me, I need to go have a word with my suitcase. It's looking at me nervously 🙈

109

9 Comments

As AI systems influence ever more consequential decisions, governance is moving from abstract ethics to concrete infrastructure. Read more 👉 https://lttr.ai/AnCle #2026 #Data #Power #Technology #Experian

12

I wrote this before heading to the FDX Global Summit this week, and honestly, it could not have been more timely. AI was the topic in every session, every hallway, every conversation. Not in a buzzword way. In a "we need to figure this out now" way. What does agentic engineering actually mean for banks and credit unions? Not the hype, but the real shift in how people work, how decisions get made, and how you govern all of it in a regulated industry. My take is on the Tyk blog. Would love to hear what you're seeing on the ground. 👇 https://lnkd.in/gKcg7uBs #FDX #AgenticAI #FinancialServices #OpenBanking #AIGovernance Financial Data Exchange, Tyk,

5

AI has made execution faster. It has not made commitment cheaper. This is the distinction most technology leaders are not making clearly enough right now. And the cost of missing it compounds quietly, until someone inherits the answer. In my first post of Season 2 of Systems and Sutras, I write about something I encountered early in my career. A decision made under delivery pressure that was already live by the time I arrived. Synchronous where it should have been asynchronous. Fast where it should have been deliberate. By the time anyone questioned it, the system had reorganised itself around that decision. Partners had integrated against it. Contracts referenced it. Changing it would have required a migration window nobody had appetite to fund. A temporary decision had become permanent architecture. AI accelerates this pattern in two distinct ways that are worth separating carefully. The first. speed surfaces existing weaknesses faster. Ownership gaps, architectural fragility, unclear boundaries. These were always present. AI simply removes the buffer that used to obscure them. The second is specific to AI-assisted development itself. Code that is locally coherent and systemically fragile, written faster than the organisational immune system can respond. Both risks matter. But they require different responses. And conflating them leads to the wrong one. The full post covers both. It also covers why reversibility is a governance discipline, not a technical preference. And why the leaders who will define this era are not the ones who move fastest. They are the ones who know which decisions to slow down, and why. Season 2 is about responsibility when execution is cheap. This is Post 1 : https://lnkd.in/eMVXKNac Execution ends. Commitment begins. Few notice the crossing.

21

3 Comments

🧭 The Most Underrated GenAI Superpower? Antipodal Generation. Ask an LLM: “Give me 30 examples that are nothing like these 300—but still on mission.” It doesn’t flinch. It leaves the semantic neighborhood with purpose. Not edge cases. Not counterfactuals. Not hallucinations. 🧠 It performs Latent Antipodal Generation—sampling from semantically distant but valid regions of meaning space. Why does this matter? Because in banking, insurance, and capital markets, it’s not the known cases that kill you. It’s: • The fraud attempt that doesn’t look like fraud • The claim that breaks your intent detection • The investment query phrased like no prompt you’ve ever seen With antipodal generation, you can: • Simulate plausible-but-unseen customer behaviors • Stress-test systems with structured semantic outliers • Pre-train agents on failure modes before they fail in production And here’s the kicker: You can dial the divergence. Turn it up, and you get wildcards that redefine risk boundaries. Turn it down, and you get high-value synthetics that plug subtle gaps in your data. This isn’t prompting. It’s vector-aware imagination—at enterprise scale. ⸻ The future of GenAI isn’t in just finishing your sentence. It’s in inventing the sentence you never thought to write. #tPower #mlai #GenAI4FS #inc81starch

40

2 Comments

Very interesting piece of news (to me) that ChatGPT found and framed at its own initiative in today’s Pulse magazine, thinking I would be interested. “That inversion invites rethinking scheduling logic more broadly. In multi‑actor systems (governments, funders, contractors, regulators), readiness becomes a shared state — one actor’s delay undermines the collective viability. The classic temporal ordering (start time, finish time) gives way to a network of readiness dependencies.” ChatGPT

8

1 Comment

I spend a lot of time anxious about if we are behind in what is clearly a race to an uncertain future. In the last few weeks, I've started to see the curtain pulled back on other approaches to ML and automation, and I'm less worried about others and more excited about the opportunity. The human factors though, still prevail. no AI is going to take away the misunderstanding and challenges that arise between people - but with the small things being automated away the risks can be found a lot sooner, and eliminated with rigour.

27

1 Comment

Consider this your next read. Clear, useful info that supports many stakeholders. Good job Mark Parsons and Steve Diggs!

5

I’m looking forward to being part of the Iowa Technology Summit this year on a panel called “The Era of Reinvention: Leadership in an AI‑Driven World.” AI is changing the way we work and lead faster than ever. Our conversation will focus on what that actually looks like in practice—how leaders adapt, how culture shifts, and how organizations navigate constant change without losing sight of their people or their purpose. For me, leading in this space starts with a simple idea: AI should amplify the human experience, not replace it—and the real work is helping teams navigate change with clarity and purpose. It’s going to be a great conversation with Jacey Heuer, Theresa Swaim, and Crystal Wells. #IowaTechnologySummit #DrivingInnovation Technology Association of Iowa https://lnkd.in/dCm6uDub

97