About
On a mission to build AI-driven platforms that power smarter, faster decision-making for…
Articles by Cássia
-
AI Won’t Kill Engineering. Weak Leadership will.
AI Won’t Kill Engineering. Weak Leadership will.
This article was first published on Medium here. You can check out all my published articles here.
2 Comments -
If Every Decision Goes Through You, That’s the Problem
If Every Decision Goes Through You, That’s the Problem
This article was first published on Medium here. You can check out all my published articles here.
-
Decide Anyway
Decide Anyway
This article was first published on Medium here. You can check out all my published articles here.
2 Comments -
50 ChatGPT Prompts I Use as an Engineering Manager
50 ChatGPT Prompts I Use as an Engineering Manager
AI is changing the way we handle work. It saves you time by automating routine tasks so we can spend more time on the…
-
Engineering Excellence Isn’t a Framework. It’s a Habit
Engineering Excellence Isn’t a Framework. It’s a Habit
This article was first published on Medium here. You can check out all my published articles here.
-
Does organisational culture affect software delivery performance?
Does organisational culture affect software delivery performance?
This article was first published on Medium here. You can check out all my published articles here.
1 Comment -
Why Product-Centric Engineering Teams Outperform the Rest
Why Product-Centric Engineering Teams Outperform the Rest
This article was first published on Medium here. You can check out all my published articles here.
-
Data-Driven Decisions for Leaders
Data-Driven Decisions for Leaders
This article was first published on Medium here. You can check out all my published articles here.
Activity
1K followers
-
Cássia Almeida shared thisMy 3 favourite leadership books were written by a SEAL, a runner (and ex-SEAL) and a bodybuilder. I read a lot of leadership books. And in reality, most of them cover the exact same thing using different words: "communication, metrics and frameworks". These 3 don't teach management. They teach how you operate first. → Extreme Ownership (Jocko Willink): calm under pressure. Delegate and empower. → Can't Hurt Me (David Goggins): how much you have left when you think you're done → Be Useful (Arnold): discipline over motivation Full write-up here 👇 https://lnkd.in/eeyjDiMiMy favourite leadership books were written by a SEAL, a runner and a bodybuilderMy favourite leadership books were written by a SEAL, a runner and a bodybuilder
-
Cássia Almeida posted thisSmall thing that will upgrade your prompts: knowing the proper name for what you want. If you ask for a "menu", you get something generic. But if you ask for a "combobox" or a "segmented control" the output comes with more precision because your input was specific. What an agent gives you is capped by the precision of what you give it. Vague words == vague results. Building a shared vocabulary isn't pedantry. It's literally the cheapest and simplest thing you can do to improve your output quality.
-
Cássia Almeida posted thisAn agent can write ten versions of a feature before lunch, but it can't tell you which one your users actually needed. That's the gap now. It's easy to miss, because ten versions really feel like progress. The output is there. You can see it, you can demo it. But what people often forget is that the first nine of them were never worth building, and nothing in the tooling is going to flag that for you. The agent just gave you back the hours you'd have spent writing it. Put those into deciding what's worth building, not into building more. Sit with the ticket before you generate anything. Ask what happens if you don't build it at all. Generation is cheap now. Prioritisation, taste, knowing when good enough is genuinely enough, that's the scarce part.
-
Cássia Almeida posted thisSomething I keep noticing when I watch other people work with agents: the slow part isn't generation, it's the re-explaining. For every new chat they rebuild the whole picture from scratch. What the system does, why that constraint exists, what they already tried. The agent forgets, so they pay for it again. The tricky part is that the cost of this is nearly invisible as it's spread across 20 different sessions. You barely feel it. Start writing things down once, somewhere the agent can read it. You were probably going to have to explain it to a human eventually anyway. The people getting real leverage here aren't prompting better, they're externalising context so it stops living in their heads. Use your AGENTS.md files. Create decision records and tests that encode the why and not just the what. Same notes serve your future self and your next agent. Codify once. Stop re-explaining.
-
Cássia Almeida shared thisMarkPad. A lightweight Markdown editor for Mac and Windows. I built it because every option I found was paid, not cross platform, too heavy, or it just showed me the raw syntax rather than the document. 3.3 MB. No account, no sync. Free and open source. ➡️ https://lnkd.in/ecsUD-VE
-
Cássia Almeida shared thisStarting has never been cheaper, but finishing costs exactly what it always did. This widening gap is where most of my projects die. I have six side projects in progress. None of them finished. I used to think that was a discipline problem. Then I noticed they all die at about the same point. Not at the start, and not near the end... but right after the hard part is solved, when all that needs is some final touch so that it is good enough for someone else to use. That's also the exact moment a new idea turns up looking much better than the thing in front of me. Starting got very cheap over the last few years. Finishing costs what it always did. https://lnkd.in/er2YcMPb
-
Cássia Almeida shared thisFred Brooks ran the OS/360 project at IBM in the sixties. It went late. So they added people. It went later. He wrote a book about it in 1975, and the line everyone quotes is that adding manpower to a late software project makes it later. And although this was fifty years ago, we still do it. I've been thinking about why. When a project is slipping, adding people is the only thing that feels like helping. Cutting scope means someone isn't getting what they asked for. Moving the date means saying the plan was wrong. https://lnkd.in/e7uNM_jC
-
Cássia Almeida shared this😂😂Cássia Almeida shared thisDon't miss the ending 🤣 BTW, if you're trying to keep up with AI, this newsletter makes it easy. *Free*, 5 minutes a day, and packed with the latest AI news: https://lnkd.in/eqhKTg5W When was the last time you had a Friday like this? 😬
-
Cássia Almeida shared thisThose who know me know that I love building side projects. That is until I have 6 in progress, none of them finished and I can't remember where I left off on any of them (yes, I know). So I built myself a little Side Project Tracker in Notion to keep my sanity in check. It's where I dump all my WIP and shipped projects, as well as the random ideas I'll probably never start. Sharing it free for anyone to grab 👉 https://lnkd.in/ew8knChP
Cássia Almeida commented on a post
1w
can't wait to read the rest
-
Cássia Almeida liked thisCássia Almeida liked thisI don't say a word about AI risk and regulation without a lawyer present 👀 Which is why, at 3:40pm on Wednesday at Ashby One in London, Tourlane's Maximilian Rauschen, Ashby's own Kelly Fish, and I are walking onto a stage together. Two Heads of Talent and one brilliant Senior Legal Counsel, talking about how you actually say yes to AI with confidence. My team builds with AI every day. We're very deliberately all in on it, but that also means getting much better at understanding what we're building, what data it's touching, who it affects, and what happens if it gets something wrong. If your team is building too, or is about to, come spend 40 minutes figuring out how to do it with confidence. And can a compliance panel have the best stage warm-up and cool-off music of the day (Elizabel 😄)? Come and find out!
-
Cássia Almeida liked thisCássia Almeida liked thisOne Special Walk! 🚶➡️💛 Please support us 🙏 A team of 9 of us from our Guildford Wargaming office are taking on One Special Walk (an 18 mile walk) in London for Special Effect. Special Effect is an incredible charity that uses technology to help people with physical disabilities experience greater independence, inclusion, and fun. Their mission is especially close to our hearts as their life changing work enables people to play video games, connect with others, and enjoy experiences that many of us take for granted. Please help us support this fantastic cause by sponsoring our team. Every donation, big or small, will help Special Effect continue making a real difference to people's lives. We'll be incredibly grateful for your support while we're battling sore feet, tired legs, and 18 miles of London streets! 👉 Link to donate here: https://specialeffect.enthuse.com/pf/wargaming-guildford
-
Cássia Almeida reacted on thisCássia Almeida reacted on this14x growth in 15 years. Tim Cook proved to be the right choice after Steve Jobs. Cook became Apple’s new boss after Steve Jobs in 2011. He had a huge impact on Apple’s value, sitting now at 4.7 trillion dollars. Larger than entire economies. Some of the things Apple achieved under his supervision changed our lives: → Launched Apple Watch, Air Pods, and acquired Beats by DRE. → Apple Pay, a huge milestone on its own. → 2.5 billion Apple devices worldwide. → Hired 100,000 more people. Are we missing any of his achievements? Now the Iron Throne of the biggest company in the world belongs to John Ternus. Get used to hearing his name. Are you excited or doubtful? The Open Agent Hackathon 2026 gives you 72 hours to learn, build and win up to $20,000 in prizes across 8 tracks. Register for free before spots close: https://lnkd.in/eea7YfGd
-
Cássia Almeida liked thisCássia Almeida liked thisThe ARR conversation in AI startups has gotten completely unhinged. When startup founders want to hype up their business to the market or investors, ARR is an easy metric to inflate. The reason isn’t a mystery: ARR has no standard definition. So it’s easy to invent your own flattering number. Some of the interesting methodologies I’ve noticed lately: → A one-time project somehow gets counted as recurring revenue for next month too → One good month of usage becomes the new assumed baseline → Customers who haven’t even onboarded yet still get counted → Some startups are even reporting weekly instead of monthly, just to make their growth look steeper None of these would survive any financial rigor, but nobody’s really checking. ARR is still the most important metric when it comes to startups. So long as it's realistic and backed by numbers that make sense. As accounting and finance leaders, it's our job to educate the market with the 2 metrics that tell the full ARR story: 1) GAAP revenue. AKA what’s actually sitting on your profit and loss statement 2) Accounts receivable balance. Who hasn’t paid you back? A lot of companies still count revenue from accounts that have functionally churned but haven’t officially said so yet. If you can't explain why your GAAP revenue looks substantially different from your ARR, that's a problem. If your AR balance is a growing share of your ARR, you're probably overstating your ARR without even realizing. Finance and accounting leaders can spot vanity metrics a mile away. Maybe ARR isn’t one quite yet, but it’s on its way.
Experience
Licenses & Certifications
Honors & Awards
-
On the Spot Award
-
Award presented by Dell and Dell Financial Services for my contributions in leading the team and providing a solution in order to resolve a critical production incident.
Recommendations received
9 people have recommended Cássia
Join now to viewOther similar profiles
Explore collaborative articles
We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.
Explore More