About
I work for Coder, we are moving software development off local machines into any cloud. ☁️
Activity
985 followers
-
Ben Potter reposted thisTwo weeks ago, our newborn started laughing. A second kid is like driving back to somewhere far - it doesn't feel nearly as long as it did the first time. The hard moments are temporary. They all are. It's all fleeting, and I get to be here with this moment. I didn't expect to have indefinite time off, but here I am. I'm making the most of it: being with this new human and with our older kid, being as present as possible while still getting to jiujitsu, doing house projects, and always folding laundry. Now that childcare is on the horizon, I'm ready to ramp up the job search. If you know anyone looking for a documentation lead or senior technical writer, I'd love to hear about it.
-
Ben Potter reposted thisBen Potter reposted thisHere we go again. Tech terminology torrent time. The only thing more confusing than the emerging term "Agentic Development Environments" is its predecessor term, "Integrated Development Environments". ADE is derived from IDE, but IDE was never clear. For most people, IDE = Code Editor. Think VSCode. But VSCode is not a truly functional IDE until you load extensions for compiling/interpreting, debugging, etc. Now we're doing it all over again. ADE could just be a code editor with AI code completion. Or it could be an agent chat interface in a terminal. Or it's a fully integrated stack. To make matters worse, vendors use "Agentic IDE", "Agent Workflows", and "Agentic Workspaces" in product descriptions. Here's my PoV. It's all the same f@$%ing thing. Yes, there are technical distinctions. But in the eyes of the customer, ADE is how we'll generate and review code, just like the IDE is how we traditionally generated and reviewed code. I see customers deploying the same three-part pattern: App->Agent->Workspace. The app is increasingly just a chat interface with a code viewer. Cursor 2.0 is best (but don't sleep on JetBrains Air or Mux). The Agent is your favorite coding LLM in a loop with MCP tools. Claude Code running Opus 4.5 hosted in Bedrock is best (but don't sleep on ChatGPT 5.2 Max for lengthy code reviews). The Workspace layer is an ephemeral VM or container sandbox for the agent to do it's work. I believe Coder is the best. My plea. Let's agree on the intent of the term ADE, or we'll forever be in a pedantic syntax argument. IDE = Human -> Edit -> Error -> Human -> Fix. ADE = Human (intent) -> Agent (Edit -> Error -> Fix loop) -> Human (review). The human role becomes declaring intent and reviewing outputs, rather than one of generation. Anytime there is a transformational change it's helpful to use a new term to describe the new way. That's why ADE matters. Maybe I'm too much of a taxonomical nerd, but words matter and terminology torrents are terminally tough. Let'd do better?
-
Ben Potter reposted thisBen Potter reposted thisThis is a must watch for every software engineering leader. This week we hosted Elliot from Skydio. It's a truly inspiring podcast. Elliot is a big Coder fan and I'm proud we're part of his journey. But this is not about Coder. It's about culture. I'm inspired by how Elliot brought AI into the the culture of engineering. And he did in an enterprise environment working on real production code. For any software engineering leader reading this post. Or wondering how to get ROI from AI. Watch this 5-minute snippet, but I encourage you to check out the full session. AI coding is real and, when done right, has a positive impact on the business and your engineering culture. https://lnkd.in/guxTDRjW My three takeaways: 1. You have to jump in. Pick a tool and a problem that no one else wants to solve. Elliot chose linting. 2. Use your existing, secure tooling. Elliot chose Bedrock and Coder to integrate Claude Code. Now the agent can work on linting. 3. Give it context in markdown. Give the agent human-readable guidelines in markdown. In this case, linting best practices. AI improved from getting it right 40% of the time to 80%, and in a way that was embraced and valued by engineers. There are *so many* good lessons from this podcast. I will post more snippets soon. Apologies if I spam my feed with insights. #sorrynotsorry
-
Ben Potter shared thisJosh is a fantastic leader. Great listen :)Ben Potter shared thisPressure felt like leadership (until it wasn’t) Josh Epstein, President & Chief Business Officer at Coder, learned this the hard way. I’ve seen urgency do two very different things to teams. Sometimes it sharpens focus and everything moves faster. Other times, it creates stress, confusion, and quiet burnout—while leaders think they’re “pushing for results.” Josh shared how early in his career, pressure felt like the right move… until it started costing trust and momentum. Increasing pressure wasn’t creating more effort...or the results he expected. It was one uncomfortable shift (that felt slower at first)... ..that ended up making everything faster. We unpack the full story in this week’s episode. So: 👉 When things slow down, how effective has pressure been for you?
-
Ben Potter reposted thisThis is an amazing opportunity to learn from Crista Wyrocki and build the foundation of a career in enterprise sales. Shoot me a DM me if you know anyone who could be a fit!Ben Potter reposted thisWe're hiring! Apply today or share this post with a great candidate in your network.
-
Ben Potter shared thisBest video I've seen around using AI in complex brownfield codebases and reducing slop/rework: https://lnkd.in/dfVGGNzwNo Vibes Allowed: Solving Hard Problems in Complex Codebases – Dex Horthy, HumanLayerNo Vibes Allowed: Solving Hard Problems in Complex Codebases – Dex Horthy, HumanLayer
-
Ben Potter reposted thisBen Potter reposted thisFor developer experience to work, the C-suite has to get it. The best organizations have CTOs who log into the developer platform, use the tools, and engage with the wider developer community. They understand why these investments matter. When there's no C-suite buy-in, proving value through metrics becomes critical. But AI creates a new opportunity. For the first time, you can show executives what you're working on in ways they can actually see and understand. Open your environment. Show them the tools. Demonstrate how processes improve. AI makes it easier to bring leaders closer to the work. Use that to build understanding and secure buy-in. Don't tell them. Show them. ◼️ Ben Potter
-
Ben Potter reposted thisHarish Gaggar and I have collaborated for years, and he consistently drives innovation for data scientists and ML engineers at Credit Karma. Join us to hear how Credit Karma is using Coder to deliver better products and experiences to customers, faster and more efficiently.Ben Potter reposted thisThe future of data work is both faster and highly configurable. Credit Karma processes massive data volumes across diverse teams, each with unique needs. Python data scientists. Scala engineers. ML researchers that require GPU access. To make sure all of these teams can move with the speed and security they need to confidently experiment with AI agents, Credit Karma used Coder to make deploying cloud development environments scalable and fast. Join us on November 12 as Coder's Ben Potter speaks with Credit Karma's Amit Mishra and Harish Gaggar to reveal: ✓ How they built a paved path for AI agent prototypes ✓ Strategies for optimizing expensive GPU resources ✓ Their approach to secure, compliant data operations at scale Register now to learn how Coder is bringing platform engineering into the AI era. http://cdr.co/4NzczMw
Ben Potter commented on a post
3mo
Congrats, Carolina! Huge
Ben Potter replied to a comment
6mo
Sophia F. IMO it has to be enterprise VPs understanding what success means! And realistic targets for agents.
-
Ben Potter liked thisBen Potter liked thisIf you are on KubeCon, come and swing by Booth 794. Happy to chat about dev productivity and AI Governance.
-
Ben Potter liked thisBen Potter liked thisThe “It’s Broken, Fix It” Problem, and Why AI Is No Different… Early in my career doing desktop support, the most common ticket I received was some version of: “It’s broken. Fix it.” No context. No steps to reproduce. Just frustration, and the assumption that the technology, or the person supporting it, had failed them. Years later as a server admin, I saw the same pattern. A small percentage of engineers would give me a clear, detailed incident report. The rest? “It’s broken. Fix it.” Now I’m watching the same dynamic play out with AI. Someone tries a chatbot once or twice, maybe a frustrating customer service bot that couldn’t understand them, and concludes that AI is useless. Or they type a vague description into an image generator, get something weird, and decide the whole thing is overhyped. Totally understandable reactions. But here’s the thing: the technology didn’t fail them. The skill of articulation did. This isn’t a criticism, it’s a pattern I’ve watched repeat at every stage of tech adoption, across every level of user. Communicating what you actually need has always been the difference between people who feel empowered by technology and people who feel defeated by it. AI just makes that gap more visible than ever. And before you ask, yes, I used AI to draft, iterate, and finalize this post. 😄
-
Ben Potter liked thisBen Potter liked thisI've been spending a lot of time lately helping people in my network with their job searches — everything from optimizing their LinkedIn profiles, to interview prep and follow-up strategy, to the psychological shit show that is dealing with rejection in this market. What has disappointed me the most? Recruiters. I started my career in high-volume recruiting. I know firsthand that it's impossible to reply to every applicant with a personalized explanation. But what I have always strived to do — and still do — is keep candidates already in the process informed. I am shocked by how many people tell me they've spent 5–10 hours interviewing at a company only to wait weeks before getting a rejection. That's not how people should be treated. Have I probably left someone hanging in my career? Honestly, yes. But the tools we have now make that inexcusable. Modern ATS platforms (shout out to Ashby) let you set reminders, automate follow-ups, and use AI to ensure no candidate falls through the cracks. The "I'm too busy" excuse doesn't hold up anymore. If you're a recruiter or a CEO reading this, here are some things that cost you almost nothing but mean everything: 1. Send "no update" updates. It's Friday. You don't have next steps yet. Send this: "Hi [Name], I was hoping to have an update for you today, but the earliest I can connect with the hiring manager is Monday. I'll follow up then." That takes two minutes to write and can completely change how someone spends their weekend. These candidates are nervous about paying their next bill, supporting their families, and just feeling wanted — and we're leaving them in the dark. They worry that reaching out too much makes them look desperate, when really they just want answers. 2. Reject in real time. If you know someone isn't moving forward, tell them now. Not next week. People go to sleep dreaming about working at your company. Leading them on is cruel. 3. Reject with feedback. If someone spent hours interviewing and you hit them with "we went with someone better suited for the role," you're wrong for that. Use AI to help synthesize your team's feedback into a meaningful rejection, so candidates know this was a serious decision, not a template. 4. Pay for take-homes. If you're asking someone to spend 3+ hours of their personal time on an assignment, pay them. Full stop. Okay, end rant, have a nice weekend :)
-
Ben Potter liked thisBen Potter liked this"Coder will be our golden path to deploying Claude Code." That's what a CTO told Eric Paulsen and Jiachen Jiang after their NDC London presentation on agent governance. Not "we're evaluating." A decision. Enterprises have moved past AI agent pilots. Platform teams need infrastructure that governs autonomous AI the same way it governs human developers—with deterministic safety, complete observability, and zero compliance gaps. Here's what makes the difference: → Agent Boundaries enforces network isolation at the process level → AI Bridge provides centralized authentication and audit trails → Workspaces give humans and agents identical context without fragmentation When a CTO says you're their "golden path," they're saying your infrastructure solves the blocking issue between pilot and production. For teams evaluating Claude Code: the bottleneck isn't the agent. It's whether you can deploy it safely at scale.
Experience & Education
-
Coder
View Ben’s full experience
See their title, tenure and more.
or
Recommendations received
2 people have recommended Ben
Join now to viewOther similar profiles
Explore more posts
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top contentOthers named Ben Potter in United States
164 others named Ben Potter in United States are on LinkedIn
See others named Ben Potter