arun-gupta

The problem is not that we don’t write specs. It’s that specs are not the system of record. Code is. In practice, the spec lives in a ticket, gets closed, and drifts. With AI generating code faster than teams can review it, that gap becomes expensive fast.

Specification-Driven Development flips that model. The spec comes first and stays the source of truth. It is a precise, versioned definition of intent that drives architecture, implementation, tests, and docs. The idea has existed in fragments: TDD, BDD, API-first design all push toward defining intent earlier. What’s different here is making the spec the primary artifact throughout, not just at the start.

Tools like SpecKit, AgentOS, Tessl, SuperClaude, BMAD, and Kiro are all exploring this space, each from a different angle.

I’ve been experimen

arun-gupta

Running NemoClaw on macOS: Sandboxed OpenClaw with NVIDIA Guardrails

Following my earlier post on running OpenClaw in Docker on macOS, I wanted to explore the next level: NemoClaw, NVIDIA's open source reference stack that wraps OpenClaw in a proper security sandbox powered by NVIDIA OpenShell.

NemoClaw was demoed at the GTC keynote and is fresh out of the oven (alpha, released March 16, 2026), so this is an early look. Let's get into it.

---

arun-gupta

Connecting OpenClaw to Claude and Telegram

In the previous blog we got OpenClaw running in Docker. Now, we give it a brain and a voice. The brain is Claude via the Anthropic API. The voice is Telegram. By the end, you will be chatting with Clawde from your phone.

Here is the full stack we are building:

Telegram (your phone)
      ↓
@askclawde_bot (Telegram Bot)

arun-gupta

OpenClaw on Docker (macOS)

What is OpenClaw?

OpenClaw is an autonomous AI agent gateway that connects large language models to the real world. It can browse the web, execute code, manage files, send messages across platforms (Slack, Discord, Telegram, email, and more), and interact with external services — all driven by natural language instructions. You can think of it as a self-hosted, extensible AI agent that runs continuously and acts on your behalf.

In this workflow, we'll call our agent Clawde — a nod to both Claude (the underlying model) and OpenClaw.

---

arun-gupta

OPEA ChatQnA on IBM Cloud

Setup Virtual Server

• Location:
• Image and profile:
• Storage:

arun-gupta

Knowledge graph and OPEA

Test

• Configure

  sudo apt update -y
  sudo apt install python3-pip -y
  pip3 install opea-comps --break-system-packages
  git clone https://github.com/opea-project/GenAIComps
  

arun-gupta

More OPEA Examples using Docker Compose

• Spin up Ubuntu 24.04 VM and install Docker following the instructions at https://gist.github.com/arun-gupta/7e9f080feff664fbab878b26d13d83d7

CodeGen

• Pull the Docker image:

sudo docker pull opea/codegen:latest

arun-gupta

OPEA ChatQnA on GKE

Create GKE cluster

• On https://console.cloud.google.com/kubernetes, click on Create on top left, choose Standard, Use Setup Guide, My first cluster, CUSTOMIZE
• Name: opea-demo
• c3-standard-8 (8 vCPU, 30 GB memory)
• Boot disk: 500 GB
• Change the number of nodes to 1
• Click on CREATE

arun-gupta

OPEA on Microsoft Azure using Docker Compose

Create your instance

Ubuntu 22.04

• https://portal.azure.com/
• Name: opea-demo
• Region: (US) West US 2
• Availability zone: Zone 2

arun-gupta

OPEA on Google Cloud using Docker Compose

Create your instance

Ubuntu 22.04

• https://console.cloud.google.com/
• c3-standard-8
• Change boot disk to 500 GB
• Ubuntu 24.04 LTS