Arun Sivashankaran

optimizing revenue funnels for the complex customer journey @funnelenvy

[Arun Sivashankaran @asivash]

@RhysSullivan lots of sharp edges working on this. currently trying Cloudflare Managed Agents and leaning into the Cloudflare primitives as much as possible

[Arun Sivashankaran @asivash]

I find it odd how little claude code knows about how it works - features, acceptable usage of Oauth vs keys, etc. Just went several rounds trying to figure out the best way to integrate claude-code-action and it had to revise the plan three times. Must be some reason for this?

[Arun Sivashankaran @asivash]

This is what we're working at https://t.co/LIdSBQDlJ8 (early days). The layer is the product. The tools on top are interchangeable.

[Arun Sivashankaran @asivash]

Karpathy is building LLM knowledge bases for research. We've independently landed on the same pattern for growth ops. Definitely similarities but the differences are more interesting. A few observations 🧵

[Andrej Karpathy @karpathy]

LLM Knowledge Bases

Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating knowledge (stored as markdown and images). The latest LLMs are quite good at it. So:

Data ingest:
I index source documents (articles, papers, repos, datasets, images, etc.) into a raw/ directory, then I use an LLM to incrementally "compile" a wiki, which is just a collection of .md files in a directory structure. The wiki includes summaries of all the data in raw/, backlinks, and then it categorizes data into concepts, writes articles for them, and links them all. To convert web articles into .md files I like to use the Obsidian Web Clipper extension, and then I also use a hotkey to download all the related images to local so that my LLM can easily reference them.

IDE:
I use Obsidian as the IDE "frontend" where I can view the raw data, the the compiled wiki, and the derived visualizations. Important to note that the LLM writes and maintains all of the data of the wiki, I rarely touch it directly. I've played with a few Obsidian plugins to render and view data in other ways (e.g. Marp for slides).

Q&A:
Where things get interesting is that once your wiki is big enough (e.g. mine on some recent research is ~100 articles and ~400K words), you can ask your LLM agent all kinds of complex questions against the wiki, and it will go off, research the answers, etc. I thought I had to reach for fancy RAG, but the LLM has been pretty good about auto-maintaining index files and brief summaries of all the documents and it reads all the important related data fairly easily at this ~small scale.

Output:
Instead of getting answers in text/terminal, I like to have it render markdown files for me, or slide shows (Marp format), or matplotlib images, all of which I then view again in Obsidian. You can imagine many other visual output formats depending on the query. Often, I end up "filing" the outputs back into the wiki to enhance it for further queries. So my own explorations and queries always "add up" in the knowledge base.

Linting:
I've run some LLM "health checks" over the wiki to e.g. find inconsistent data, impute missing data (with web searchers), find interesting connections for new article candidates, etc., to incrementally clean up the wiki and enhance its overall data integrity. The LLMs are quite good at suggesting further questions to ask and look into.

Extra tools:
I find myself developing additional tools to process the data, e.g. I vibe coded a small and naive search engine over the wiki, which I both use directly (in a web ui), but more often I want to hand it off to an LLM via CLI as a tool for larger queries.

Further explorations:
As the repo grows, the natural desire is to also think about synthetic data generation + finetuning to have your LLM "know" the data in its weights instead of just context windows.

TLDR: raw data from a given number of sources is collected, then compiled by an LLM into a .md wiki, then operated on by various CLIs by the LLM to do Q&A and to incrementally enhance the wiki, and all of it viewable in Obsidian. You rarely ever write or edit the wiki manually, it's the domain of the LLM. I think there is room here for an incredible new product instead of a hacky collection of scripts.

[Arun Sivashankaran @asivash]

His closing line hits - "there is room here for an incredible new product instead of a hacky collection of scripts." That's the gap in marketing/growth too. Tons of AI wrappers, no persistent knowledge layer underneath. Every agent interaction starts from zero.

[Arun Sivashankaran @asivash]

Glad to see this validated. I usually end up picking out the few concepts I like from the kitchen sinks and rolling them into my own stack.

[Matt Silverlock 🐀 @elithrar]

The number of issues I observe from people using these “kitchen sink” plugins for their coding agents — on top of context degradation + model guardrail errors — seems at odds with *any* improvement to output quality.

I don’t recommend + actively steer our people away from them.

[Arun Sivashankaran @asivash]

@thesamparr build static LPs with claude and embed HubSpot forms.

[Arun Sivashankaran @asivash]

@kurt @appenz @howard @meetgranola @firefliesai We’ve been fireflies customers for a while and I was trying Granola out today. Didn’t realize fireflies had a desktop app until I read this - thanks! fwiw getting markdown out of fireflies is quite easy

[Arun Sivashankaran @asivash]

@stephsmithio I loved the 8 week faster freestyle course by Effortless Swimming

[Arun Sivashankaran @asivash]

@tobi I'm going to be trying out Mercury (https://t.co/EwtCU03ZDC). Compelling design and seems to have been built with multi tenancy in mind.

[Arun Sivashankaran @asivash]

Autonomous experimentation wrapped around a self-reinforcing contextual layer

[ericosiu @ericosiu]

https://t.co/LGxhaM4TDn

[Arun Sivashankaran @asivash]

@coreyhainesco @ShivaSkanthan I'll burn tokens for (the offer of) stars.

Added a Google Tag Manager cli and did some cleanup https://t.co/dIqxtlSdZj

[Arun Sivashankaran @asivash]

@coreyhainesco that’s what it’s for :)

[Arun Sivashankaran @asivash]

@coreyhainesco all of these tools have APIs, it’s just that vendors only prioritize CLIs for developer audiences. For all of them you’ll need an access token and / or Oauth setup. And with the generator you (or anyone) can create more.

[Arun Sivashankaran @asivash]

A huge bottleneck for AI agents in marketing? No CLIs.

So I built open source ones for GA4, Ahrefs, Meta Ads, Mailchimp & Buffer. Skills teach the agent what to do, CLIs give it hands to do it.

https://t.co/ofBCPmrRFY