Someone needs to build a company around Customer Context Graph. Collect all the threads – emails, meeting transcripts, slack messages, contracts, deliverables, detail, info, and config – from your customers into context that can be explored and queried by agents. This info is scattered between CRMs, ticketing systems, note takers, product, landing pages – it's inherently cross platform information. You need a new solution. Kind of how Segment did it trad SaaS apps. With this context, you can fire up Claude Cowork or similar for ad-hoc work or build extremely powerful agent automation flows. Expose the context as skills, MCP, and file system. Even better if you build it as open-source with a hosted option so people can take it on-prem as needed. Create a connector ecosystem around it. This will power every single next-gen AI-native full-stack business. Sort of like the context graph (@ashugarg @JayaGup10 ) that has been discussed recently but I'm thinking something very concrete: "Get me all the context about this particular customer." A customer-level, cross-system context substrate that agents can explore and act on

Jan 18, 2026 · 7:48 PM UTC

143
22
402
117,353
Maybe Segment should just do this, would be huge. They already know how to do identity resolution on users really well. But feels a bit too heavy, would rather use a startup for this. Something light and easy to start with for devs
2
6
5,459
Sort replies: Relevant Recent Liked
Replying to @aikarjal
Agreed drew.tech/posts/knowledge-ba…

How We Built Knowledge Bases for AI Agents at Vercel

The biggest opportunity I see for an AI picks and shovels startup: batteries included knowledge bases.

drew.tech
3
1
12
3,664
Great post, spot on. We're thinking on the same lines here. To us it's also about servicing the customers directly on deliverables, not just about sales ops.
2
2,374
Replying to @aikarjal
Been thinking about this exact problem and building an internal tool around it. In order to productize it you need to build pretty robust integration coverage. I would buy vs build if I could :).
3
13
1,489
Same here, we're going to build this internally for our own use to power customer ops and deliveries. I have a company repo that has everything from strategy, product, sales, quarterly plans, etc. and I use it in Claude Code. Would love to do the same for customers
1
2
1,345
more replies
Replying to @aikarjal
We already do this today via @graphlit From Attio to Slack, from Fireflies to Google Workspace. Ingest all unstructured data sources. Index in multiple ways (time, vector, graph, geo). Model as relationships and entities. Retrieve via API or MCP.
1
6
1,160
Cool, I have been eyeballing Graphlit since your last article! How do you handle resolving customer identities across different sources? I admit that I haven't gone deep enough yet to fully explore Graphlit but it’s on top of my todo list. This need for customer-centric context graph sprang up from my work around our internal customer operations tooling.
1
1
968
more replies
Replying to @aikarjal
seems like many AI apps are trying to do this (but for use by their app). Wonder if more generally this should be "owned" by the customer and provisioned to Chatbot/apps as needed
1
6
972
I think it should be an infrastructure layer that can be plugged into apps and bridge the gap from CRM to your own product data, to email and others. Many apps seem to be building this on their own. But I think it's such a general concept that it should exist as infrastructure that can be plugged in. There might be multiple versions of this but the one I'm thinking about is centered around the customer. Customer data is in any case the most fragmented of them all.
2
490
Replying to @aikarjal
isn’t this just connectors → normalizer → graph db with summaries agents can query or pointer to info they can drill down grind would be connector maintenance which u *could* have agents just verify on a cron maybe but zapier could just launch a “customer graph” step and kill the startup unless u got vendor lock in by having a better agent layer if its open source mcp for Claude code, maintainer burden is probably too high If its paid mcp for Claude code, id predict a short life span
2
4
2,445
Customer identity resolution needs solving. And exposing conveniently to agent, MCP might do or a bit more complex skill. Connectors need to understand how to extract and process each source. Vector and full-text search should be there as well. Thinking out loud, would be cool to customize the form factor in which the context is exposed. Don't see that Zapier would kill this easily
3
2
1,722
more replies
Replying to @aikarjal
Have been working on this, and this timely discussion accelerated our thinking. Love to talk to more design partners. (Calendar link in the article)
Kirk Marple
@KirkMarple
Jan 20
Article

Introducing Dossium: The Customer Context Platform

You have a customer call in an hour. You need to know what happened in the last QBR, whether there are open support tickets, and what their VP said in that Slack thread last month. So you open five

1
2
360
Fantastic! This accurately captures what I was thinking. There are many nuances in there to be solved for sure, like access controls, data controls, custom integration hooks from product, etc. but it's directionally right there
1
199
Replying to @aikarjal
there is one and it does over 200M in ARR yall just dont even try
2
2
1,852
Care to share?
2
1,796
Replying to @aikarjal
we have @microHQ
1
1
1,234
Looks very cool
1
930
Replying to @aikarjal
Although it feels lucrative to provide identity resolution as a first-class-citizen, it is best to give freedom to the model to figure this stuff out (bitter lesson).
1
1
288
Interesting, elaborate. You think it could be better to give agent access to every system and let it figure it out?
1
167
more replies
Replying to @aikarjal
@usepylon
1
1
1,049
I've seen this round. Thought that it's a customer support thing more than anything else. Do they expose the context so I can use it elsewhere? Load it to Claude Agent SDK / Cowork / Code?
1,030
Load more