Sponsored LinkedIn post
A dedicated post to introduce your brand to my audience.
Co-Founder @ Studio1
Companies pay Arindam to write a post about their product on LinkedIn or X.
This page is where they book it. Below: Arindam's audience, recent posts and price. Naano, the platform behind this page, handles the brief, your approval of the post and the payment.
€100per X post, fixed price
You approve the post before it goes live
Looking for Arindam? View LinkedIn profileView X profileA creator yourself? Create your own page
I build practical AI tools, experiments, and content around LLMs, AI agents, and developer tools. I’m an AI Engineer and Developer Advocate at Studio1, where I work with AI, SaaS, and DevTools companies on technical content, agent demos, developer education, and advocacy. What I share: → Practical AI tutorials and experiments → LLM, RAG, and AI agent workflows → Open-source projects you can actually build with → Technical breakdowns of new AI tools and models → Lessons from building and shipping in public I’ve built 100+ agentic examples and open-source projects including: → awesome-ai-apps, a collection of practical AI applications → CC-Lens, a real-time Claude Code monitoring tool → Formaly, a SaaS for conversational forms → termui, an open-source terminal UI project I also create technical content across YouTube, blogs, and social platforms, with 1M+ blog views and 150K+ YouTube views. My work sits at the intersection of AI engineering, open source, developer education, and developer advocacy. I like taking something that sounds complicated, building it myself, and turning it into something people can understand and use. Through Studio1, I also work with AI, SaaS, and DevTools teams to make their technology easier for developers to discover, understand, and adopt. If you're building something interesting in AI or developer tools, I'd be happy to connect. Build less hype. Ship more useful things.
LinkedIn export · Aug 18, 2026 – Nov 15, 2026 · Provided by the creator, not independently verified by Naano
X
Public X data · Captured Nov 15, 2026
I just received $1,000 in credits from Neon for building vid0.in. I've been a Neon user for a long time. Whenever I start a new project, my stack looks the same: Drizzle and Neon, paired together. Why it works for me: 1. Drizzle and Neon fit together well. Schemas, migrations and queries just work. 2. I spend my time on the product instead of fighting the database. 3. It handles the small details of storing and tracking analytics data without extra setup. A big thank you to the Neon Postgres team for supporting me as a builder. At the early stage, every bit of support matters. It means a lot to see tools back founders with credits, and I'm planning to keep building on Neon for a long time. vid0.in is public and still in beta, so feel free to try it out and tell me what you think. What's your go-to database stack for new projects?
Prompt engineering teaches you how to tell an AI agent what to do. Harness engineering teaches you how to make sure it doesn't need to be told the same thing 15 times. And I think we're seriously underestimating the second skill. Found a free, open-source GitHub course that teaches Harness Engineering from scratch. 14 lectures. 8 hands-on projects. Here's what you'll learn: 1. The 5 building blocks of an agent harness Instructions: What the agent should do State: What it needs to remember Verification: How its work gets checked Scope: What it's allowed to change Lifecycle: How tasks start, finish, and recover 2. How to make agents catch their own mistakes One agent writes. Another reviews. Build feedback loops where agents verify their own work instead of waiting for you to catch every mistake. 3. How to build multi-agent workflows Move beyond simple agent loops. Learn agent graphs, parallel execution, multiple agents handling different tasks, and failure recovery. 4. How Claude Code, Codex, and DeepSeek approach it Claude Code's layered memory system Codex treating the repository as its source of truth DeepSeek's approach to plugins A look under the hood at how these tools are actually built. 5. A skill that builds a harness for you It generates instruction files, feature lists, and setup scripts directly inside your repository. 6. A script that audits your existing agent setup Point it at any repo. It evaluates the five parts of your harness, scores your setup, and tells you what's missing. And the best part? It's completely free, open-source, MIT-licensed, and translated into 15 languages. I particularly like the audit script. Because when an agent keeps making the same mistake, our first instinct is often to blame the model. But the real problem might be the instructions, missing checks, or boundaries we've given it. We're moving from writing better prompts to building better environments for agents to work in. That's a skill worth learning. Course: https://lnkd.in/dmmWg9b7 What part of your agent harness needs the most work right now?
prompt engineering is not what makes your agent reliable. the harness is! (google research's harness² lifted glm-5 from 64.2% to 75.7% on a benchmark without touching the model) the formula is simple: agent = model + harness the model reasons. the harness is everything around it: the loop, the tools, memory, permissions, and checks. same model, different harness, very different agent. here's how i'd build one, step by step: → guides: agents.md, rule files, constraints. every line should be a past failure turned into a permanent fix → sensors: linters, tests, and validation scripts the agent runs on its own output before you ever see it → the loop: plan, execute, verify, fix. cap the retries and the budget, and escalate when it's stuck → memory: the model forgets every session, so the harness keeps state, decisions, and artifacts → permissions: which tools, how many writes, and what needs approval. the model won't enforce safety, the harness has to → observability: log every tool call, cost, and retry so you notice when behavior drifts google's developer team makes a point i agree with: stop judging agents only by one big benchmark score. write small behavioral evals instead. did it run the validator? did it ask before guessing? those catch regressions a composite score hides. the honest tradeoff: a harness is real software. it needs tests, upkeep, and someone who owns it, and a heavily tuned harness can overfit to one kind of task. this is what comes after prompt and context engineering.
Choose a format for your next LinkedIn collaboration.
A dedicated post to introduce your brand to my audience.
Choose a format for your next X collaboration.
A dedicated post to introduce your brand to my audience.
Quotes your X post with their own comment.
Your collaboration, managed on Naano
How does this price compare? B2B creator rates 2026
People who saw my posts, from my LinkedIn export.
People who saw my posts, from my LinkedIn export.
People who saw my posts, from my LinkedIn export.
People who saw my posts, from my LinkedIn export.
People who saw my posts, from my LinkedIn export.
X
Kimi K3.1 Pro is 5.4x cheaper than Opus 5.5! (and the price list says the gap is only 2x) Price per token is the wrong number to compare models on. On paper, Opus 5.5 costs exactly 2x Kimi K3.1 Pro. $4 vs $2 per 1M input, $20 vs $10 output. So I measured what one task actually costs. Same 12 support tickets, same 12 NER texts, same prompts. Kimi K3.1 Pro ran on @nebiustf Token Factory, Opus 5.5 through Claude Code. Both got 12/12 correct on both tasks. per 1,000 tasks: → Classification: $0.93 vs $5.36 (5.8x) → NER: $2.12 vs $7.19 (3.4x) four things explain it, and they apply to any model you pick: 1. you pay for tokens you never see reasoning was Kimi's biggest cost line: 35% of the bill on classification, 45% on NER. Even for one-word answers. 2. your harness is part of your bill 93% of Opus's classification cost was Claude Code writing ~620 tokens to a 1-hour prompt cache on every call. Stock coding agents also send 22k to 32k prompt tokens before your input arrives. For a one-word classification, that's 30 to 49x the cost of a stripped-down call. 3. instruction following is a cost lever, if the model is verbose adding "exactly one line of JSON, no prose" cut Kimi's NER cost by 35%, mostly by halving its reasoning. On Qwen3-30B it added 11 to 25%, because Qwen was already answering short. 4. the cheapest model has a price too Qwen3-30B on the same Token Factory account came in 14 to 27x under Kimi, but missed one entity in 12 on NER (¥12 billion). On some workloads that trade is fine. On others it isn't. On these tasks, Kimi K3.1 Pro matched Opus 5.5 at a fraction of the cost. For high-volume work like this, that middle of the curve is where I'd start. the honest tradeoff: I used no API keys. Every token count comes from the CLIs' own JSON output, multiplied by list price. It's 12 items per task, so read it as directional. Kimi's cost also moved between runs because of Token Factory's cache hit rate. the takeaway: before you pick a model, measure your own workload. Include the reasoning tokens, the harness overhead and the prompt. Kimi K3.1 Pro is live on @nebiustf Token Factory. The full breakdown is in the video 👇 (bookmark this)
Running 1,000 coding agents on a single job sounds chaotic. But Claude Fable 5.2 has a harness designed to make it work. The core idea: 1,000 agents need clear briefs, not just prompts. Here’s how the system works: → Each brief defines one piece of work, with files that no other task can touch. → The builder agent completes the task and returns a receipt. → A separate verifier, which never saw the builder's reasoning, checks the result. → Jev Engineering routes prompts to the right seat, checks GOAL.md in around 100 ms, and escalates anything it's unsure about. → Destructive actions like deletes, pushes, and deploys always require your approval. Your role becomes much simpler: write the briefs and approve the deploys. The whole approach comes down to one principle: Give every agent one job and a stranger to check it. https://t.co/HuLAgOF1jx
Storefront by Naano · Create yours
Arindam Majumder is a B2B LinkedIn creator based in the United States with 15,751 followers, posting about AI, SaaS, and Software. On Naano, one sponsored LinkedIn post by Arindam Majumder costs €250, a fixed price per post. A sponsored X post costs €100.
The Naano team casts the creators, writes the briefs, reviews every draft with you and reports results per creator. You approve the shortlist and every draft.