Kirk Marple

Founder/CEO of Graphlit (@graphlit): Operational Context Layer for AI Agents 🚀 @zine_ai @dossium 👋 ex-MSFT, PA born, Seattle bred. Dad to dogs/humans

[Kirk Marple @KirkMarple]

Very accurate, but still worth it. (Credit: @gapingvoid)

[Kirk Marple @KirkMarple]

@RaminNasibov Railroad Tycoon

First game that had me staying up too late, when I had work the next morning.

[Kirk Marple @KirkMarple]

@garrytan https://t.co/8Bgh2zboYX

[Kirk Marple @KirkMarple]

https://t.co/ZjnQdmjlqL

[Kirk Marple @KirkMarple]

@karrisaarinen @HarryStebbings It’s like no one knows how to open their laptop and sit on their own couch.

[Kirk Marple @KirkMarple]

@trashpandaemoji @SIGKITTEN @OpenAI @ajambrosino This you?

[Kirk Marple @KirkMarple]

@boardyai It’s the instant distribution channels that are worth it. Hard to recreate organically.

[Kirk Marple @KirkMarple]

Tagging a few of the companies and builders I referenced in the piece - lots of different angles here, but all touching the bigger question of how agents get durable context from the tools and knowledge we already use.

@microHQ @getindigo @linear @attio @intelligenceco @notion @Box @glean @Letta_AI @clickup @mem0 @cognee_ @TrustGraphAI @garrytan

[Kirk Marple @KirkMarple]

https://t.co/a8hU1jvmRa

[Kirk Marple @KirkMarple]

https://t.co/a8hU1jvmRa

[Kirk Marple @KirkMarple]

@TechCrunch Yep. The agent is the user now.

https://t.co/5U5GrSktyT

[Kirk Marple @KirkMarple]

https://t.co/a8hU1jvmRa

[Kirk Marple @KirkMarple]

@yoheinakajima Totally agree.

You’re reverse engineering the Graphlit architecture I built over the last 5yr, from first principles. 😀

[Kirk Marple @KirkMarple]

A file is for a person.
A block is for a person.
A page is for a person.

The agent needs its own primitive.

[Kirk Marple @KirkMarple]

https://t.co/a8hU1jvmRa

[Kirk Marple @KirkMarple]

@calcsam Great minds think alike 😄

We just shipped multi-channel agents in @dossium a few weeks ago.

Totally agree with this thesis.

[Kirk Marple @KirkMarple]

Your UX point is what I was getting at.

Do you want a file system or a wiki, is the simplest way to put it.

The data model underneath does matter, and that’s why one of Notion’s strengths is its prescriptive data model.

LLMs are obviously great at data wrangling but they still need another data layer to ingest to/retrieve from, and that’s where all these agent infra companies are focused. (Including us)

[Kirk Marple @KirkMarple]

@AnkMister We support Twitter/X ingestion via the X API, in our backend (@graphlit).

And then we treat embedded media like attachments on the tweet. Their API gives a downloadable path to the video so it wasn’t that hard.

[Kirk Marple @KirkMarple]

There's a big difference between Markdown-only 'company brains' and what a true context layer can support.

Multimodal ingestion from any source, along with LLM-driven entity and fact extraction.

Being able to paste a tweet, auto-extract the MP4 video, transcribe and analyze it - that's how you create the searchable, agent-native 'brain' for you or your organization.

Can even connect up to Claude or ChatGPT via MCP.

[Y Combinator @ycombinator]

Over the past year, we've been building our own internal agent infrastructure at YC: over 350 tools, self-improving skill loops, and a shared organizational brain that gets smarter overnight.

In this episode of the @LightconePod, we sat down with YC General Partner Pete @koomen to talk about how he led the effort from the ground up.

We cover how giving agents unrestricted access to one database was the key unlock, the self-improving skill loops that get smarter overnight, and why he thinks we've arrived at the personal computer moment for AI.

00:39 — YC's AI Stack
02:15 — The Finance Team Problem That Started It All
05:07 — SQL Access Changes Everything
07:20 — One Database to Rule Them All
09:14 — Jevons Paradox
10:07 — Denormalizing for Agents
12:15 — The Single-Player Era of Agents
14:16 — 350 Tools and a Shared Registry
16:24 — Skillify, DRY, and MECE Resolvers
18:23 — The Self-Improving Dream Cycle
20:26 — The Two-Sentence Pitch Skill
23:06 — How Super Intelligence Compounds
25:10 — Recording Everything as a Building Layer
27:10 — The Shared Organizational Brain
29:18 — Trust-Default Culture as a Requirement
30:44 — Raising the Floor for New Employees
32:35 — Horseless Carriages
34:24 — Why Chat Is the Best Interface for Agents
38:50 — Just-in-Time Software
40:49 — Centralizing vs. Decentralizing AI
43:32 — The Personal AI Revolution

[Kirk Marple @KirkMarple]

Just the difference between Gdrive and Notion, for example. Treating the file as the object vs everything having to live on a parent Notion page.

And I haven’t seen anywhere Notion AI is fully multimodal, such that it indexes an uploaded video and you can use Ask AI to chat about it. (Just tried this and it only uses pages as context not the video transcript)

But if it works for your use case, that’s all that matters.

[Kirk Marple @KirkMarple]

@nbaschez https://t.co/8cqFPI8Lmv

[Kirk Marple @KirkMarple]

There's a big difference between Markdown-only 'company brains' and what a true context layer can support.

Multimodal ingestion from any source, along with LLM-driven entity and fact extraction.

Being able to paste a tweet, auto-extract the MP4 video, transcribe and analyze it - that's how you create the searchable, agent-native 'brain' for you or your organization.

Can even connect up to Claude or ChatGPT via MCP.