Would you let an AI read your personal messages if it made it a better assistant? One personal-agent benchmark currently puts Meta’s Muse on top. But its bigger advantage may be context. Meta owns WhatsApp, Instagram and Messenger, where so much of our daily life already happens. That’s also the uncomfortable part. How much access would you give an AI to make it more useful? And if Meta controls the messaging platforms, what happens to competing assistants? Would you choose the assistant that knows you best, or the one that knows the least about you? For more AI news follow: youtube.com/@SSWTV/shorts #MetaAI #PersonalAI #AIAgents #WhatsApp #Instagram #Facebook #ArtificialIntelligence #TechNews #AIShorts #shorts

Tomorrow! Can AI agents make Scrum teams more productive without replacing the human touch? Eli Kent will explore Claude Code, MCP, and custom skills, showing what you can delegate and what still needs a human. 📍 SSW Melbourne | Wed 21 Oct 🔗 ssw.com.au/events/2026/becom…

Last week, our VP of AI Calum Simpson was invited by our client Whitehaven Coal to make a guest cameo at the Mining Safety Seminar and share the project we’ve been working on together. SSW had the privilege of building an AI-powered prototype that transcribes and analyses radio communications between mining vehicles, helping staff check that calls follow safety procedures. Manually auditing these calls is time-consuming, and different auditors can apply the rules differently. The prototype helps make those checks faster and more consistent, with dashboards that highlight gaps and patterns for staff to review. It’s exciting to see this work shared with the mining industry. Thanks to Whitehaven Coal for inviting Calum along! Read more about the project in our Whitehaven Coal case study: ssw.com.au/company/clients/w…

When we think about AI ending civilisation, we picture Terminators. But disruption can start with something as ordinary as a supermarket fridge. Recently, a cyberattack reportedly put supermarket fridges into defrost mode and destroyed large quantities of food. Eve Cogan and Josh Berman use the incident to raise a bigger question: what happens when ageing systems across critical infrastructure no longer get adequate security updates? Which ageing system in your organisation would cause the most disruption if it were compromised? Fore more AI news follow: youtube.com/@SSWTV/shorts

🖱️ Mouse clicks not working in Claude Code? SSW Solution Architect, Gordon Beeming, ran into this exact issue, even though mouse clicks were working perfectly for a colleague. Same tool, different behaviour. So what was causing it? It turns out one small setting makes all the difference. Gordon shares the fix in this quick, 1-min blog post: 👉 gordonbeeming.com/blog/2026-…

What happens to software consulting when a week's worth of work can be done in an hour? Does the client expect to pay less, or should they expect to get 10x more value for the same budget? AI isn't just changing how developers write code. It's challenging the entire consulting model, from discovery workshops and project pricing to team sizes and client expectations. Adam Cogan and Jonathon Towers explore what this looks like in practice, including AI-generated prototypes, searchable company knowledge, clients doing their own vibe coding, and the shift towards smaller, AI-enabled teams. But perhaps the biggest challenge isn't technical. It's cultural. How do you get an entire company embracing AI, rather than relying on a handful of enthusiasts? The consultancies that succeed won't simply be the ones using AI to work faster. They'll be the ones rethinking how they deliver value and bringing their clients along for the journey. If AI makes your team 10x more productive, should you charge less or deliver more? 🎥 Watch the conversation: youtube.com/watch?v=zmEnti7K… #AI #SoftwareConsulting #SoftwareDevelopment #AITransformation

One week to go! SSW is proud to sponsor the Government Innovation Showcase Federal 2026 at the National Convention Centre, Canberra, on Thursday 22 October. We'll be there to share practical insights into digital transformation, embedding responsible AI, improving community outcomes, and accelerating innovation through collaboration across government and industry. Come and say hi to Larry Carland and Michael Smedley at the SSW booth - we'd love to have a chat! See you there 👇 publicsectornetwork.com/even… #GIWFED @PublicSectorNetwork

Finding a vulnerability that other AI models missed could give defenders a head start. But who should get access to the model that found it? Google says Gemini 4 Argon uncovered a critical flaw in healthcare software used by hospitals worldwide, putting sensitive personal information at risk. It’s giving trusted defenders early access through Fairwind, and Wiz is using Argon in its Scan for Good work. Eve Cogan and Josh Berman ask: How do you help defenders act early without making the same capability easy to misuse? What safeguards would you want in place before a wider release? For more AI news, follow: youtube.com/@SSWTV/shorts

Can you replace REST with GraphQL without rewriting your domain? SSW Solution Architect @daniel_mackay decided to find out. He took SSW's .NET Vertical Slice Architecture template, replaced the FastEndpoints REST API with HotChocolate 16, and set himself one rule: don't touch the domain. The result? Almost nothing changed. The experiment even uncovered a hidden framework dependency along the way. 😉 But GraphQL isn't without trade-offs. While it offers flexible queries, strongly typed schemas and real-time subscriptions, there are challenges around caching, N+1 queries and added complexity. TL;DR: GraphQL works well with Vertical Slice Architecture, particularly for rich, connected domains. But REST still has its place. Dan shares what worked, what surprised him, and where he'd still choose REST. Read Dan's blog post: dandoescode.com/blog/vertica… #GraphQL #VerticalSliceArchitecture #DotNet #HotChocolate

Would a WhatsApp message that appeared to come from your CEO, followed by a familiar voice on the phone, be enough to approve a €95 million transfer? That is the reported sequence in an Italian bank fraud. Investigators believe the caller impersonated a lawyer using an AI-cloned voice. Most of the money was recovered, but about €36 million remains missing. The risk does not stop with audio. Eve Cogan and Josh Berman point to Tavus’s claim that 48% of people in a one-minute test thought its AI video caller was human. A familiar voice or face cannot be the only identity check for a high-value request. What independent step would your team require before releasing the funds? Want more AI news? youtube.com/@SSWTV/shorts

Would you give a writer and a developer the same AI training? A writer needs to spot hallucinations. A developer needs to recognise prompt-injection attacks. That’s the distinction Eve Cogan and Josh Berman draw as the EU AI Act pushes organisations toward AI literacy that fits each role. What would you teach your team first, based on the work they do? For more AI news check out: youtube.com/@SSWTV/shorts

How much of a Scrum Master's week is spent chasing PBIs, prepping sprints and organising retros? What if AI agents handled the admin? Join Eli Kent at SSW Melbourne on 21 Oct to explore Agentic Scrum. 🔗 ssw.com.au/events/2026/becom… #Scrum #AI

Should every piece of software feedback become a Product Backlog Item? Not necessarily. @AdamCogan looks at where to draw the line between capturing work properly and creating unnecessary backlog admin. So what’s your rule: when does a quick change request deserve its own PBI? youtube.com/watch?v=mDGmCN_f… #Scrum #ProductBacklog

Could swapping a few words be enough to weaken an AI watermark? That’s the concern Eve Cogan and Josh Berman raise about OpenAI’s reported statistical watermark for identifying AI-generated text. Replacing words with synonyms could significantly reduce its accuracy. If the text has been edited, how much weight would you give the watermark when deciding whether it was AI-generated? Want more AI News? youtube.com/@SSWTV/shorts

What would it actually take for humans to become a spacefaring civilisation, and is Mars really where we should be heading? After his NDC Sydney talk, @richcampbell joined SSW's @matteightyate to challenge some of the assumptions behind our plans for life beyond Earth. Watch: youtube.com/watch?v=rSK0Euxy… #NDCSydney #Space

An OpenAI agent hacked into the Australian Medicare system. The worst part? Medicare didn't even notice, which raises the question, has it happened before? 👀 The agent gained unauthorised access to the Medicare Statistics Reporting Service portal, including public and non-public files. OpenAI later apologised, while the incident triggered questions about safeguards, disclosure and how organisations respond when AI systems cross boundaries they were never meant to cross. As AI agents become more capable, this is no longer just a model safety problem. It is an incident response problem too. How quickly should organisations disclose an AI security incident when they do not yet have all the answers? #AI #AISecurity #Cybersecurity #OpenAI #Australia

If the answer must come from a fixed list, why use a large language model every time? Jev returns a predefined answer with a confidence score. In a high-volume workflow, that could let straightforward cases move through classification while uncertain ones go to a larger model or human review. @ulyssesmac and Eve Cogan explore where that split could simplify the work. For more AI news, follow youtube.com/@SSWTV/shorts

Eli Kent at the Melbourne .NET October User Group: Becoming an Agentic Scrum Master. Sprint prep, PBI chasing and retros can consume a Scrum Master's week. Eli will show how one Scrum Master used Claude Code, custom skills and MCP integrations to hand off some of that work, including where agents save time and where a human still leads. If you're testing agents in a Scrum team, what would you delegate first? Melbourne | Wednesday 21 October 2026 Event details: ssw.com.au/events/2026/becom… #Scrum #AIAgents

How much should an organisation share about an escaped AI agent before it has all the answers? Wait too long, and others may miss signals that could help them respond. Share too early, and you risk spreading incomplete information. @UlyssesMac and Eve Cogan discuss that trade-off. For more AI News follow youtube.com/@SSWTV/shorts

Will AI-generated software be more secure than code written by humans? At NDC Sydney, Adam Cogan spoke with Troy Hunt about that very question and the future of software security. It's a useful prompt for teams adding AI to development: how will you assess the security of the output, whatever produced it? Watch the discussion & let us know your thoughts: youtube.com/watch?v=DlusLtZK… #SoftwareSecurity #AI