Pooya Saeedi

Canada’s AI game? Our labs are world-class—but our businesses are still trailing. That’s the reality, according to Canada’s AI Minister, Evan Solomon: “We have some of the best researchers… but Canadians and Canadian companies are lagging on adoption.” https://lnkd.in/gBBFaRvj The challenge? Adoption isn’t just a technology issue—it’s a trust, clarity, and ROI issue. - Too many organizational leaders don’t see tangible business impact - Regulations and governance feel murky, slowing progress - Businesses—already stretched thin—need confidence, not hype ⸻ Why this matters to you: AI can be a game-changer for productivity—but only if it’s deployed strategically, ethically, and with real-world impact. ⸻ What we bring to the table: - ROI-Centric Adoption: Pilot frameworks that show quick value and build momentum - Governance & Regulation: Clear, compliant AI policies that earn stakeholder trust - Human-Centric Change Management: Workshops and coaching that raise internal buy-in and readiness Want to stop lagging and start leading? Drop a comment or DM “ADOPT” to get access to our AI Adoption Blueprint—designed for those who are ready to level up. #CanadaAI #AIadoption #ThePhoenixFirm #DigitalTransformation #TrustInAI #AILeadership

2

When Tobi Lütke of Shopify announced their AI-first policy last week, it got a lot of people talking. At LTV, we've also been moving towards a similar AI-driven vision — and one big question kept coming up: "How do you make AI feel fun and empowering, instead of like it’s replacing parts of people's jobs?" Recently, I tried an experiment with 1 of my teams, inspired by Mark Rober’s "Super Mario Effect" TED Talk — the idea that people succeed more when challenges are gamified. 🎮 The Idea: Gamify AI learning through custom GPT-powered games during our remote team happy hour! Here’s what we built: - pAIctionary: Draw doodles, upload them, and have a custom GPT guess them — points for every correct guess! - codenAImes: A word-association game where teams craft clues to make GPT guess the right words without accidentally hitting the wrong one. - pAIsswords: Like "Password" but with AI — players get the AI to guess a secret word without using restricted hints. - AIseeAIcompare: Players prompt GPT to recreate given images, and a custom tool scores how closely they match. The Results? - Each of our 6 vertical teams competed in fast-paced 10-minute games. - Each team was required to have 2 champions participating actively in each game. The champions strategized, but everyone contributed — cheering, suggesting prompts, and collaborating. - The competition brought energy and laughter, but more importantly, it built confidence and creativity around using AI tools. 👉 Key takeaway: When you gamify AI adoption, you shift the mindset from fear to fun; from threat to tool. 💬 How are you encouraging AI adoption on your teams? I would love to hear any creative ideas or strategies you’ve seen work! 🚀 #AI #Leadership #CultureBuilding #Teamwork #Innovation #FutureOfWork #LearningThroughPlay

49

2 Comments

Product leaders from across Canada met today for PLX Lean Coffee for a discussion of hot topics from the trenches! We took on two burning issues: HOW TO PRIORITIZE - CUSTOMER REQUESTS, LONG TERM VISION, ETC. - We discussed the high stress environment of competing priorities and how to allocate scarce resources among so many priorities. - Key to our discussion was that this was a very early stage start up (3 co-founders) in a very competitive and fast-moving market. - "What do you need to do to raise your next round of funding" became the key question. From that, clarity came quickly. - The company needs to turn four beta customers into long term contracts. We discussed what that would take, and how the success of the trials was being measured and communicated to the customers. - The co-founders must be completely aligned. There is no room for pulling in different directions. - The product leader decided her most important next action was to come up with better metrics for the trials, communicate progress against those metrics consistently with those customers, and let those metrics drive the decisions on product priorities. WHAT IS THE PLACE OF PRDs IN A WORLD OF VIBE-CODING AND NEED FOR SPEED? - A product leader shared he was questioning the value of the PRDs he is currently expected to deliver. - There were not many fans of PRDs in the group. PRDs require a lot of time to research and write and require lengthy sign-offs. They assume all requirements can be known up front. - The preference was for multiple flexible documents, most which live in Jira and Figma, that are more useful to developers and designers. Vibe-coded prototypes can be used very effectively to communicate requirements. These are the "what" and "how" parts. - Some parts of the PRD are still needed - the "why" parts - product strategy, link to business goals, success metrics. This is done for the product leader's own thinking and to communicate up. It does not have to be a PRD. With participation from Vancouver, Calgary, and Toronto, Lean Coffee keeps growing and getting better. If you are a product leader (CPO, VP, Director) and want a great place to bounce questions and ideas off your peers from across Canada, join us for our next one.

13

Canadian Tech Sovereignty: The Long Weekend Problem As we roll into a long weekend here in Canada, I’ve been thinking about something a little heavier than BBQ menus and motorcycle routes: building truly Canadian technology solutions. Sounds noble, right? Keep our data here, keep our IP here, keep our decisions here. The reality? It’s like deciding you’ll only shop local and then realizing your town doesn’t make running shoes, dish soap, or -- ironically -- half the parts of your barbecue. When I set out to create sovereign-friendly solutions, I thought: How hard can this be? Spoiler: harder than riding my Goldwing through Toronto traffic at rush hour. The challenges? • Vendors: Not many fully Canadian ones, and fewer still that can scale at the pace clients expect. • Rules: We love our regulations, but they don’t always keep up with the tech. • Clients: They want compliance, privacy, and sovereignty -- without sacrificing speed or cost. (Tall order, eh?) The opportunities? • Differentiation: Not every customer wants their data wandering across borders like a hitchhiker. Some care -- deeply -- and will pay for assurance. • Innovation: Constraints force creativity. I’ve had to rethink how services are built, delivered, and secured. It’s exhausting… but in a good way. • Community: There’s a growing number of people who care about this, and they’re starting to find each other. (Think “small town farmers’ market,” but with Kubernetes clusters instead of kale.) The funny part? The more I lean into solving sovereignty issues, the more doors open. Clients, partners, even regulators are suddenly interested in what we’re cooking north of the border. So as you’re enjoying your long weekend (and maybe your own barbecue), here’s something to chew on: Do we want Canadian tech sovereignty, or do we just want to say we want it until the bill comes due? I don’t have all the answers -- but I do have some hard-won experience. And if you’re wrestling with these same questions, let’s talk. I promise I’ll bring the sass (saas?) and maybe even some practical solutions 😜 Happy long weekend, Canada. 🍁

3

1 Comment

I’ve been thinking about the recent news that Waymo is exploring what it would take to deploy autonomous vehicles in Toronto. It’s worth remembering that Toronto has been here before — Uber previously tested self-driving vehicles in the city and found that full autonomy was only achievable within a very narrow operational design domain. Since those early tests, the technology has clearly matured — but whether it’s good enough for Toronto’s streets is still an open question. Toronto is not an “easy” city for autonomy. Our streets are dense, irregular, and constantly in motion — streetcars, cyclists, pedestrians, delivery vehicles, construction zones, winter weather, and curbside activity all layered into the same space. That raises a fundamental question: are today’s autonomous systems truly ready for cities like Toronto, or are we scaling up experiments before proving resilience in complex urban conditions? The biggest uncertainty is how AVs would coexist with vulnerable road users. Toronto’s streets rely heavily on human negotiation — eye contact, informal yielding, and shared assumptions. Autonomous systems don’t participate in that social contract. If robotaxis arrive, will they genuinely improve safety outcomes, or will pedestrians, cyclists, and transit users be expected to adapt to machine behavior? This isn’t just about whether the technology works. It’s about whether it belongs — and under what conditions — on some of the most complex public roads in North America. #AutonomousVehicles #UrbanMobility #SmartCities #RoadSafety #VisionZero #Toronto https://lnkd.in/gWHZsjWz

53

8 Comments

hard not to imagine a future where better run countries like Canada step in and announce that they are willing to help AI companies that don't want the tech supporting a lawless and corrupt regime migrate

7

Getting some valuable insights and excellent questions. My methodology: 1. We really need to evaluate ROI, start small as proof of concept AND value. 2. Begin with some time-consuming task that is not critical, adds value, and is repetitive. 3. Measure value, then find other tasks. 4. Build out the system - this may mean a small server running the local LLM, a proper RAG to keep your templates, documents and relevant legal/privacy information. 5. Tackle those bigger issues. Always, always, always retain control of your information, and keep your data private.

2

Software engineering isn’t just about shipping code—it’s about shaping teams that scale, systems that adapt, and cultures that endure. One of the biggest lessons I’ve learned in engineering leadership is that technical debt is easy to see, but organizational debt is silent — and far more expensive. We spend hours debating frameworks, performance tradeoffs, and system design. But how often do we apply that same rigor to the design of our team processes, feedback loops, or decision-making velocity? Here’s what I’ve seen work: • High-context environments beat high-control ones. Engineers thrive when they understand why, not just what. • Velocity emerges from clarity. The best teams don’t move fast because they cut corners—they move fast because they don’t waste cycles on ambiguity. • Great engineering managers are multipliers. Not the best engineer in the room, but the person who ensures every engineer is at their best. If you’re scaling a team or facing complexity creep, ask yourself: Are we debugging our systems as actively as we debug our culture? #SoftwareEngineering #Leadership #EngineeringManagement #TechCulture #TeamScaling #OrganizationalDesign ⸻ We are proud of the culture and organization we’ve built in Amazon Vancouver. Come meet us! Want to discuss about these in person? See you next Thursday, May 15th at the Amazon Payment Products event!

53

1 Comment

Gatik CEO Gautam Narang joined SiliconANGLE & theCUBE’s John Furrier on NYSE Wired’s Robotics & AI Leader Series to discuss what it takes to achieve commercial scale and safety in autonomous trucking. In this segment, Gautam walks through our journey from commencing our partnership with Loblaw Companies Limited in 2020, to launching Canada’s first driver-out commercial routes in 2022, to a new 5-year expansion that will scale to 50 trucks serving 300+ stores across the GTA. Stay tuned for more insights from the conversation. For the full interview, check it out on theCUBE + NYSE Wired: https://lnkd.in/gqPwJzQa #AutonomousTrucks #SupplyChain #Robotics #AI #Logistics #NYSEWired Brian J. Baumann David Vellante

86

1 Comment

The U.S. and Canada are tightening expectations around crypto platforms. Dank Daily #0053.2 — Forbes’ 2025 list of Canada’s top crypto exchanges points to a clear trend. The platforms rising to the top take regulation seriously. With tighter rules on custody, stablecoins and trading limits, the industry is being pushed to mature, and the exchanges that embrace registration, fair pricing and solid security are standing out. #DankDaily #DigitalFinance

1

1 Comment

A recent study highlighted by CTV News found that overlooked workplace gestures can carry real costs — and engineering teams feel this fast. As lead engineers, we set the tone more than we realize: • How we respond in code reviews • Whether we interrupt or truly listen in stand-ups • How we react when something breaks • If we acknowledge effort, not just output A quick dismissive comment, a sarcastic reaction, or silence after someone flags a risk can shut down future communication. And when engineers stop speaking up, bugs linger, incidents escalate, and innovation stalls. On the flip side, small positive gestures compound: ✔️ Asking clarifying questions instead of assigning blame ✔️ Thanking someone for raising an issue early ✔️ Treating “basic” questions as valid ✔️ Staying present during discussions Strong engineering leadership isn’t just about technical decisions — it’s about creating psychological safety so problems surface early. 👉 Reminder for leads: Culture isn’t built in architecture docs. It’s built in code reviews, incident calls, and everyday interactions. What’s one small habit you’ve adopted that made your team stronger? #LeadEngineer #EngineeringLeadership #TechCulture #PsychologicalSafety #SoftwareEngineering #DevCulture

4

1 Comment

Canada and the Architecture of Democratic AI Power While the U.S. races to scale compute and China accelerates algorithmic dominance, Canada is quietly attempting something more ambitious: redesign the operating system of democratic AI governance itself. Last month, the Center for AI and Digital Policy submitted a possible roadmap for the Ministry of Innovation, Science and Economic Development Canada. On paper, it’s about ethics; in substance, it’s about sovereignty. CAIDP’s key recommendations form a new social contract for the algorithmic age: Create sovereign, low-carbon compute clusters open to universities and SMEs. Ensuring domestic access to high-performance computing without total reliance on U.S. hyperscalers. Tie federal funding and procurement to algorithmic transparency, explainability and safety testing. Enforce algorithmic registries and impact audits, so citizens know which systems shape their credit, healthcare or security outcomes. Embed Indigenous and civil-society representation directly in federal AI councils. Align regulation with the Council of Europe’s AI Convention, connecting domestic legitimacy to international credibility. The report also raises a paradox at the heart of democratic AI: how to lead morally without losing materially. Canada has the values, but not yet the infrastructure to express them. That’s where new recommendations become crucial: Build a Canadian Compute Cloud. A federated infrastructure linking universities and public agencies through sovereign nodes, potentially in partnership with Quebec’s hydro-powered data centers. Establish a National AI Testbed to certify trustworthy models, similar to the FDA’s approach to medical trials. Introduce an AI Stability Mechanism, a fiscal buffer protecting domestic AI research from foreign subsidy shocks or capital concentration in a few hyperscalers. Still, structural limits persist. Canada’s market scale and industrial base cannot match the U.S., China or the EU. A regulatory-first approach could slow deployment if it lacks reciprocal industrial strategy. In other words, you can’t regulate a future you haven’t built yet. If Canada succeeds, it will show that democratic accountability can scale. Not as moral constraint, but as industrial strategy.

7

1 Comment

⛑️ OpenAI released Sora 2 and an invite-based iOS social video app (U.S./Canada first). The model adds synced audio, stronger physics/realism, and more steerability. A consent-gated “cameo” feature lets verified users allow trusted friends to use their likeness. A System Card outlines safety limits (e.g., upload restrictions, teen safeguards, ongoing moderation). Ultra-realistic, social-by-default video raises new operating questions for enterprises: managing likeness rights, verifying provenance/watermarks, setting pilot guardrails for brand and copyright risk, and mapping controls to your governance framework (e.g., NIST RMF / ISO 42001). Early momentum is real—and so are concerns over deepfakes and harmful content—so policies should ship before pilots scale. 👉 If your org could access Sora 2 today, what’s the first written rule you would require before launching a pilot? #AILeadership #AIGovernance #Sora2 #GenerativeAI #AIVideo #Provenance #Watermarking #TrustAndSafety #Deepfakes #RiskManagement #NIST #ISO42001 #BoardOversight #GenAIGlobal Tap follow for no-fluff AI news & insights from MIT’s Gen AI Global founder Oscar Sanchez 🧠

21

1 Comment

Can a Single AI Agent Replace an Entire Team? New research from the Trusted and Efficient AI (TEA) Lab at UBC explores a pivotal shift in AI design: moving from high-overhead Multi-Agent Systems (MAS) to streamlined Single-Agents with Skills. The Problem: Multi-agent teams are great for complex reasoning but are slow and expensive. They constantly repeat context and pass messages back and forth, leading to high latency and massive token usage. The Solution: "Compilation" The researchers found we can "compile" these teams into one single agent by turning agent behaviors into a library of "skills." 54% Reduction in Tokens: By sharing one context window, the AI stops repeating itself. 50% Faster Latency: No more waiting for multiple API calls to finish "talking" to each other. Preserved Accuracy: Single agents matched or even beat team performance on benchmarks like GSM8K and HotpotQA. The Catch: The "Cognitive Load" Breaking Point 🧠 It’s not all sunshine and rainbows. AI agents, like humans, have a "magical number" capacity limit. The Phase Transition: Accuracy remains stable until the library reaches about 50–100 skills, then it drops sharply. Similarity is the Enemy: The AI fails not just because of the number of skills, but because similar skills (e.g., "Calculate Sum" vs. "Compute Total") cause "semantic confusability." The Fix: Hierarchical Routing 📂 To scale beyond 100+ skills, we must organize them into categories. By using a two-stage "category first, skill second" selection process, researchers restored accuracy from ~45% back up to 85%. The Takeaway for Developers: Audit your skill library: Merge near-duplicate skills to reduce confusion. Organize at scale: If you have 100+ tools or skills, implement a hierarchical router. Optimize Descriptors: Selection depends on the meaning of the skill, not just the code. Read the full paper: “When Single-Agent with Skills Replace Multi-Agent Systems and When They Fail” by Xiaoxiao Li. #AI #LLM #MachineLearning #GenerativeAI #AIProgramming #TechResearch

5

The logistics tech scene is having a moment — from fellow Canadians to boot 🇨🇦 Toronto-based Waabi just raised $1B to bring autonomous trucking (and now robotaxis) to market. It's the biggest fundraise in Canadian history. The future of supply chain is complete autonomy — AI coordinating freight, and autonomous trucks moving it. Wilson will have his own fleet of trucks one day, they might just be Waabis. https://lnkd.in/gDG4JnBg

46

1 Comment

Canadian AI firm Cohere is expanding into Europe with a new Paris office, joining President Macron’s push to build a European AI hub. Valued at $6.8 billion, Cohere is taking on rivals like Mistral, OpenAI, and Microsoft — with a focus on enterprise AI, data security, and digital sovereignty. Can European AI finally compete with U.S. dominance? Let’s break it down. #OpenAI #MistralAI #Microsoft #FutureTech #VentureCapital https://lnkd.in/eXrVpUJH

2

6 Comments

Today is a big day. For years, I've watched brilliant Canadian entrepreneurs struggle to access AI tools that transform Fortune 500 companies. That's not just unfair - it's economically backwards for our country. Today, agentiiv is solving that equation with the launch of You.Scaled.: $4 million in free AI services for 500 Canadian small businesses and nonprofits across Canada. The goal? To democratize access to enterprise-grade AI across industries and regions and give every organization a chance to scale with the help of smart technology. Read the release and get more info at the links in my comments. And a big thank you to our partners on this: Canadian Marketing Association Ownr Staples Canada CCNDR #AI #CanadianAI #SME #SMB

114

27 Comments