/Tech2h ago

John Ssuh argues companies must rebuild around a single unified data timeline to maximize AI agent performance

Tech investors including Marc Andreessen have amplified the proposal.

1691.3K231.3K306.3K
Minh Nhat Nguyen
Dave Font
Jerry Liu#906
Original post
Marc Andreessen 🇺🇸@pmarca

Interesting.

John Suh@john_ssuh

Increasingly, I believe companies may need to be rebuilt from the ground up, where you have a single timeline of all observability + product metrics + file changes laid out in a retrievable system, like Datadog + Posthog + Google Drive + Slack (really unified filesystem of Claude Code chats + Codex chats). This might be the new data foundation for any and all companies to maximize AI. Needs to be rebuilt because keeping track of diffs on existing system basically impossible to produce longitudinal information on decisions and rollbacks, something coding agent storage companies are actively trying to figure out, but this should extend to businesses as a whole.

Highly skeptical existing businesses will adopt this though because it means overhauling everything about their instrumentation and business data, but I think businesses built on this foundation probably can execute 100x better and faster

5:37 PM · Jun 11, 2026 · 133.2K Views
Sentiment

Positive users endorse rebuilding companies around a unified AI data timeline for major execution gains while negative users call the idea bearish or long overdue.

Pos
81.6%
Neg
18.4%
17 comments with sentiment.
Cluster Engagement
Posts from X
Most Activity
Most Activity
VIEWS1K
Ambar Choudhury@ambarc

@john_ssuh This is what @bitboardhq aims to enable.

4hViews 1KLikes 2
BOOKMARKS2REPLIES4
Jerry Liu@jerryjliu0

There's got to be some major advances in the data ingestion, indexing, and retrieval layer to make this happen.

Agents are getting much more sophisticated in fetching diverse sources of context with simple tool interfaces (e.g. CLI + filesystem). But: * It is a known problem that agentic federated search through MCP isn't optimal for retrieval relevance * Agentic search is generally really slow without pre-indexing * When you interleave diverse sources of data, you need to weight each item properly * Different types of data require different query interfaces - from SQL to embedding search

Looking forward to seeing how this materializes. Our main focus today is unlocking document-based context and we'd be keen to see how this integrates with SaaS/structured data.

John Suh@john_ssuh

Increasingly, I believe companies may need to be rebuilt from the ground up, where you have a single timeline of all observability + product metrics + file changes laid out in a retrievable system, like Datadog + Posthog + Google Drive + Slack (really unified filesystem of Claude Code chats + Codex chats). This might be the new data foundation for any and all companies to maximize AI. Needs to be rebuilt because keeping track of diffs on existing system basically impossible to produce longitudinal information on decisions and rollbacks, something coding agent storage companies are actively trying to figure out, but this should extend to businesses as a whole.

Highly skeptical existing businesses will adopt this though because it means overhauling everything about their instrumentation and business data, but I think businesses built on this foundation probably can execute 100x better and faster

47mViews 779Likes 7Bookmarks 2
LIKES8
Gary Basin@garybasin

@john_ssuh @nptacek Great minds. Slacks successor will likely eat everything. People will learn how to use agents from watching their teammates use them

2hViews 993Likes 8Bookmarks 1
RETWEETS20
John Suh@john_ssuh

Increasingly, I believe companies may need to be rebuilt from the ground up, where you have a single timeline of all observability + product metrics + file changes laid out in a retrievable system, like Datadog + Posthog + Google Drive + Slack (really unified filesystem of Claude Code chats + Codex chats). This might be the new data foundation for any and all companies to maximize AI. Needs to be rebuilt because keeping track of diffs on existing system basically impossible to produce longitudinal information on decisions and rollbacks, something coding agent storage companies are actively trying to figure out, but this should extend to businesses as a whole.

Highly skeptical existing businesses will adopt this though because it means overhauling everything about their instrumentation and business data, but I think businesses built on this foundation probably can execute 100x better and faster

6hViews 191.4KLikes 907Bookmarks 1.1K
John Suh@john_ssuh

100%, though I think future of slack is more like shared sessions of codex + claude code + etc everything else. @sdand has an interesting side project that solves a portion of this. I think a cloud filesystem + capturing diffs on files, in addition to product and systems telemetry probably is a good foundation. But maybe this is just Snowflake, unclear

2hViews 317Likes 4Bookmarks 1
Zaki Hasan@zakihasan_

You don’t need to overhaul or rebuild a business from the ground up to get to an AI native state (generally speaking of course, the extent of change required depends on the complexity of the company as represented by data).

The business data can stay where it is, and the business itself can integrate any model/application layer into workflows. We’ve built this.

1hViews 48Likes 2Bookmarks 1
Nicholas Blanchard@corelumen

@john_ssuh @pmarca One day, a company will be built on an #Ainix foundation, and from day one, it will be built from top to bottom for inference.

https://www.github.com/syndicalt/ainix-public

32mViews 14Bookmarks 1
John Suh@john_ssuh

Yep. I got more pilled on what @ListenLabs is doing, essentially AI native qualtrics - high dimensional signal mining from consumers extremely high leverage. I’m not 100% sure about mining internal employee preference the way Meta is doing (though useful for training models in general)

1hViews 296Likes 1Bookmarks 1
Rick Trades@EricMarson14

@john_ssuh @pmarca How does a company like @Oracle do this? (Just one example)

2hViews 268Likes 1
Zaki Hasan@zakihasan_

@john_ssuh We’ve built this - it doesn’t require an overhaul or rebuild. Businesses can migrate to being AI native, using any model/application layer AI, whilst keeping their data exactly where it is.

1hViews 58Likes 2Bookmarks 1
Tony Rose@tonyrose023

We’re building it, a core CubiCube called Jade Planner - planning tools must be real time and on top of the systems of record, more so as agentic processes take more prominence , and we need to steer from higher levels of scope, and thereby abstraction,

What is even more cool about Jade Planner, we deploy with CubiCube - everyone gets their own, decoupling per seat complexity and maximizing per human personalization.

Check it out http://cubicube.com

37mViews 12Bookmarks 1
Gary Basin@garybasin

@john_ssuh @nptacek @sdand Yep shared sessions with broader workspace so teams can play with the artifacts together. I think this is an independent co from the labs because you’ll want to be model and harness agnostic; enterprises won’t want to be held hostage by a single model provider

2hViews 28Likes 2
Cooper Veit@CooperVeit3

@john_ssuh macrodata refining

4hViews 218Bookmarks 1
John Suh@john_ssuh

Having such system just allows agents to scale a lot more efficiently and make better decisions. It'll be trivial to ask questions about churn at scale, churn of a single user (tie it to engineering decisions AND product decisions), what experiments are working and not working. If you go down enough "whys", it ultimately comes down to not enough GPUs and electricity (we don't need this if Mythos costs 1 c, can do 10k tokens/s) + suboptimal data storage and access (we don't need this if data providers support 1000 queries/s for petabytes of data out of the box). A version control system + telemetry layer for an entire business, not just slices of the business arbitrarily defined by Business Analytics, OTEL, cloud file systems, communication apps.

4hViews 645Likes 2
John Suh@john_ssuh

@ambarc @bitboardhq This seems like a connector + Analytics dashboard?

4hViews 851
rf86@plein_soleil___

@john_ssuh @pmarca Or we can put everything in github

2hViews 90Likes 1
Seth BRKV@SethBRKV

@john_ssuh One fuck up (or a virus) will spread thru whole system. That's the main issue

2hViews 72Likes 1
John Suh@john_ssuh

@garybasin @nptacek @sdand Yeah. At the same time I think it’s beyond just cloud agents - it’s really a foundational data layer for the entirety of a business’s state change over time

2hViews 21Likes 2
Damir Wallener 🇭🇷🇨🇦…🚀🛰️…⚽️🥁…👨‍🍳@DamirWallener

That’s not really my question, though.

Companies can do those things now - but mostly don’t. It is very rare to see continuous improvement - mostly they lurch from one suboptimal static regime to another.

And it’s not because of technical limitations…it’s because that’s how big groups of humans organize.

How did that change, without getting rid of all the humans?

4hViews 63Likes 1
Load more posts