Manveer Chawla

Building @tryzenithai | Runner | Alum @ Confluent, Dropbox, Facebook, IIT Bombay

[The New Stack @thenewstack]

Integrating AI Into Apache Kafka Architectures: Patterns and Best Practices
from @confluentinc, by @manveerc
https://t.co/Lk1IVNsUz6

[Manveer Chawla @manveerc]

People still say @ClickHouse is append-only. That is an eight-year-old myth. Standard #SQL UPDATE shipped in 2025. Lightweight updates use patch parts, skip table rewrites, and run 1000x faster. I analyzed 100+ merged PRs to prove it.

https://t.co/c5l3HPzEgA

#database #clickhouse #dataengineering #analytics

[Mateo Torres @torresmateo]

"The model is the brain. The harness is the body."

I sat down with @robinnewhouse from @cline

We got into harness design, model-agnostic agents, and why the bottleneck in coding agents has quietly shifted.

🧵👇

[Manveer Chawla @manveerc]

Most AI SRE tools fail because they try to be "autonomous." I am skeptical of that. A model can't remediate production unsupervised, and it shouldn't.

The real win is the "Companion" model.

The AI does the legwork (triage, timelines, correlation). The human makes the judgment.

I broke down 5 #AI #SRE workflows and the #MCP runtime infrastructure needed to make them safe in production on the Arcade blog.

The goal: let the on-call start on page 5 instead of page 1.

Full post here: https://t.co/aLCDY52kPA

#AI #SRE #RELIABILITY #ONCALL #DEVOPS #MCP

[Manveer Chawla @manveerc]

2,500x faster than #MongoDB. 10x faster than #Elasticsearch. 9,000x faster than #DuckDB. On 1 billion JSON documents. Single node.

#ClickHouse's native JSON type stores each path as a columnar subcolumn with type preservation and primary key support.

I traced 80+ PRs across 4 years. The full evolution from String blobs to production-ready columnar JSON.

https://t.co/tVtmLh2gxf

@ClickHouseDB #JSON #Performance

[Manveer Chawla @manveerc]

#ClickHouse JOINs in 2026:

6 algorithms. Cost-based reordering. Parallel hash by default. Runtime bloom filters. Correlated subqueries decorrelated.

I audited 50+ PRs from 2022 to 2026. The "avoid JOINs in ClickHouse" advice is 3 years out of date.

https://t.co/fAZgqUqz8C

[Manveer Chawla @manveerc]

#OpenClaw's promise is great but the security model isn't there. I was already using #ClaudeCode and #WhatsApp constantly, so I built on that instead. Now I can text my assistant "prep me for my 2pm" or "latest for a customer project", seconds later I get a response right in my WhatsApp chat.

Full working code: https://t.co/uSZL7xWTY0

Tutorial: https://t.co/B3KI0vvfoV

[Manveer Chawla @manveerc]

WebMCP asks every developer on the internet to build a second interface for AI agents. Meanwhile, server-side MCP already exists and the browser already has the Accessibility Tree. Why are we adding a middleman?

https://t.co/pL8YufktRZ

#AI #MCP #WebMCP #LLMs

[Manveer Chawla @manveerc]

Building personal agents like @openclaw forces us to rethink how we run AI agents

The reality: if you're running untrusted agent logic, a shared kernel is a liability. The blast radius is just too big. Wrote down some thoughts here on different options:

https://t.co/XWPPI7M4ON

[Manveer Chawla @manveerc]

AI SREs will fail on missing data before they fail on missing IQ.

Most observability stacks aren't built for agentic workflows. They drop the exact high-cardinality dimensions like User IDs or Request IDs that an AI needs to find a root cause.

My thoughts on this are up on the @ClickHouseDB blog.

https://t.co/qZoccrQJfk

#AI #SRE #DevOps #observability

[ClickHouse @ClickHouseDB]

As we head into 2026, we expect agentic AI SREs to move more from hype to production.

Though that fail, won’t fail on models. They’ll fail on data.

Manveer explains why ClickHouse solves this problem and fits this workload:

- Long retention, full-fidelity telemetry
- High-cardinality data without pain
- Fast SQL for agentic investigation loops
- Built for correlation, not dashboards

AI SREs need better observability, not bigger models.

https://t.co/2IiNeuSF3n

[Manveer Chawla @manveerc]

The old observability model is broken. Siloed tools can't handle high cardinality data well. We analyzed a unified stack for ClickHouse.

https://t.co/u6vRTdOJiZ

[Manveer Chawla @manveerc]

@motherduck https://t.co/tq1hkRLGFS

[Manveer Chawla @manveerc]

SEO is changing. Fast. The proof? Check out these insane results from our work with the brilliant team at @MotherDuck

🚀 18.9% lift in share of voice
🚀 3x content velocity

Here’s a quick look at how we did it 👇

[Manveer Chawla @manveerc]

@motherduck You don't need to compromise on speed and quality. You can achieve both.

Onward! 📈 #AEO #SEO #LLM #Marketing

[Manveer Chawla @manveerc]

@zeeg My biggest gripe, it keeps forgetting how I work and what i have already told it. Its the “We already talked about it, don’t you listen” syndrome

[Manveer Chawla @manveerc]

@zeeg Agreed with you @zeeg , what likely is going to happen is that volume of traffic for agents is going to explode and every website will have some way of supporting it. What’s not happening is users only relying on agents for everything.

[Manveer Chawla @manveerc]

@KaranVaidya6 @composio @lightspeedvp Congratulations @KaranVaidya6 and @GanatraSoham