Konstantin Piliugin

To date, writing extensions for Envoy has been a laborious process that involves writing C++ and compiling a fully custom build of the entire proxy.  Can we make that easy and make Envoy more approachable by developers?. Can we create an ecosystem of reusable extensions around it - akin to how npm ecosystem fundamentally altered Node.js's trajectory. With that vision, we at Tetrate have been working with Envoy community to first make Envoy easy to extend by any developer. We added dynamic modules a year ago and then added a proper Go and Rust SDK around it. Today, we are making the developer experience much easier and announcing a home for Envoy extensions with an open source extension marketplace around it. Thanks Matt Klein Anna Berenberg for your support and thanks Rohit Agrawal Ben Plotnick Anubhav Aeron for early feedback and trials. You can read more about Built on Envoy here: https://lnkd.in/gbNz9-3v and see a cool demo video from our engineering lead Ignasi Barrera on how to write a simple Go extension and test it locally for Envoy to not allow AI calls to specific models. https://lnkd.in/gcpYJyjD

74

I was showing Ichi to about 120 building department staff when something interesting happened... During my webinar with CALBO last week I was demonstrating how we pull code sections and jurisdiction-specific requirements. Standard demo stuff. But these folks didn't only care about the speed (though yeah, turning 30-minute code searches into seconds is nice). They want to know, how can they trust it. This is where being able to show them the exact source of information matters. Makes total sense when you think about it. These are the people who carry the liability when something goes wrong. They're not going to trust some AI that spits out answers without showing its work. That's why we built Ichi to be completely transparent. The plan reviewers and inspectors I talked to after were genuinely relieved. Someone finally built tech that respects how they actually need to work. If you want to see what transparent AI for building departments actually looks like, I'll share the full demo link in the comments.

34

4 Comments

SMS-iT Update: 400+ upgrades shipped. We’ve been building quietly, not pitching. What we’ve done: * Made the entire platform faster, more stable, more reliable. * Cleaned up the UX/UI so it’s easier to use every day. * Added new features instead of just polishing old ones. * Baked Agentic AI into real workflows, not as a gimmick. Major wins: * Team-ready: assign leads, contacts, pipelines, calls, appointments to specific teammates. Smart Inbox is now team-aware. * Core modules overhauled: Appointments, Pipelines, Smart Inbox, Voice, Reports, Contacts. * Messaging (text, email, calling) is sharper and more dependable. * Workflow Automation upgraded: more powerful nodes, better performance, and you can trigger real actions across SMS-iT directly from AI-driven workflows. * Global search, timeline, header actions (call / notepad from anywhere) are live. * You can route specific inbound numbers to specific teammates. * Build-iT got huge upgrades: import templates, see analytics, and next is “build with AI” (full landing pages, funnels, etc.). * Web Agent-iT, Voice AI, Cloud Numbers (buy numbers inside SMS-iT, no config), and compliance/billing are in final stages. * Blockchain layer (STRING token) is in progress. What’s next: We aim to finish remaining work by Dec 31, 2025. Jan 1, 2026 we move into marketing and officially treat SMS-iT as out of beta. You’re getting: * A platform built for teams. * Faster, cleaner, smarter operations. * Agentic AI built into how you actually work. * One login for messaging, calling, AI workflows, site building, voice AI, search, and more — with almost no setup. We’ve been quiet because we’ve been building. You’re about to see why. #SMSiT #AgenticAI #AI #nostack

8

Twilio just acquired Stytch. This is a great move, here's why 💡 David Jones-Gilardi and I spoke with Max Gerber (Stytch Product Engineer) on our podcast The Flow recently about all things MCP and OAuth. One of the big gating factors that is slowing the adoption of agents is the desire for them to act on behalf of end users. Think about buying a movie ticket or making a reservation at a restaurant. Stytch has built an enterprise-ready platform for authentication and authorization that millions of people experience when they log-in to a product their using. This is built on top of industry standards like OAuth. But people might not know that those standards are racing to embrace the needs to AI and agents, and Stytch is right in the middle of that work. Listen to us dig into this: https://lnkd.in/gDu4BPgf

53

3 Comments

Replit agents, Supabase, v0, Cursor… everything connects now. Each new link feels like progress. But too much orchestration is just fragile glue. A tower of dependencies waiting to break in ways you won’t see coming. The speed is addictive. You feel like you’re building fast, but really you’re piling on risk. The smarter play isn’t avoiding integrations. It’s knowing which ones you can’t afford to outsource. Own the critical path. Because when the API disappears, so does your product.

2

Everyone in the mobile app space should be paying attention. The road to launch on new ventures is getting shorter and shorter…

5

1 Comment

🚨 Agentforce Builder is officially GA. But if you're only looking at the builder, you're missing the real game-changer: Agent Script ✨ Many developers are still relying entirely on an LLM’s interpretation, hoping for the best. Agent Script changes the game by letting you build predictable, context-aware agent workflows that put you - not just the AI in the driver's seat. So, how do you champion Agent Script and stay ahead of the curve? It comes down to architectural thinking, business logic, and deep platform expertise. Here are the 5 core skills you need to master: 🧠 1. Finite State Machine (FSM) Design It’s time to adopt an "Architecture Over Prompts" mindset. Move away from blindly trusting the LLM and start structuring your conversations into predictable, auditable paths. ⚙️ 2. Basic Deterministic Logic You don't need to be a hardcore Python developer, but you do need a solid grasp of basic programmatic expressions to effectively guide your agent's reasoning. 💾 3. State & Variable Management An AI agent is only as smart as its memory. You need to understand how to manage state throughout a conversation—specifically, mastering the difference between Mutable and Linked variables. ☁️ 4. Salesforce Ecosystem Expertise Agent Script orchestrates the conversation, but the actual heavy lifting happens in Salesforce. You need to know exactly when to invoke Flows for data operations, Apex for complex calculations, and Data Cloud for knowledge retrieval. 🛠️ 5. Developer Tooling vs. Conversational Dev Yes, you can build agents conversationally using the Agentforce AI assistant. But to champion this architecture at an enterprise scale? You need to get comfortable under the hood with developer tools like VS Code and Salesforce DX. The era of purely prompt-driven development is evolving into structured AI architecture. 👇 Follow me (if not already 😊 ) for more deep dives into Agentforce, AI architecture, and mastering the Salesforce platform. Let's build the future together. Get Started with Agent Script - https://lnkd.in/efdKwW5H #Agentforce #Deterministic #AI #Agents #Architecture

11

1 Comment