Arnab Seal @hermes_seal

Co-founder & CTO - @substrateco | Turbocharging platform for startups & giants: 0→1 builder for SaaS and DeFi, patent-holder, trusted to ship at scale, fast. getsubstrate.co Joined July 2025

Tweets

106
Followers

101
Following

697
Likes

230

Arnab Seal @hermes_seal

a day ago

At @substrateco, security is baked into our memory architecture, and not just an afterthought. 1. Auth: Every API call requires signed tokens, enforced by gateway policies. 2. Encryption: Both vector storage and transit use robust encryption. Sensitive content never lands in…

0 0 0 11 0

Brycent @brycent_

2 days ago

When you're vibe coding and the AI says "You're absolutely right!"

48 15 577 19K 22

Download Image

Arnab Seal @hermes_seal

2 days ago

And that's how we're piecing our RAG leveraged context suggestions at @substrateco

Arnab Seal @hermes_seal

2 days ago

And that's how we're piecing our RAG leveraged context suggestions at @substrateco

0 0 0 56 0

0 0 2 35 0

Arnab Seal @hermes_seal

2 days ago

We’ve been heads-down setting up our new RAG system at @substrateco , and I wanted to share a peek at how we’re piecing things together! 1. @OpenAI Embeddings to help us create our Vector Query from user prompts 2. @pinecone our vectorDB, searching mountains of docs and…

0 0 0 56 0

Klaas @forgebitz

3 days ago

if people don't care about your MVP they won't care about your V1 or V2

69 6 165 10K 12

Yohei @yoheinakajima

7 days ago

updated summary of responses (again with chatgpt, sorry for any misses or mistakes): (Part 1 of 2) Agent Platforms, OS & Memory Layers @christinetyip is building a shared memory layer so many agents can access a single, cohesive memory; @ankurdorroy’s @getsubstrate adds…

Yohei @yoheinakajima

a week ago

208 11 303 67K 184

16 14 64 15K 34

Substrate @substrateco

3 days ago

we didn’t need another model. we needed shared memory. substrate came from teams losing context between chats, tools, and model switches. one memory layer that travels with the work so answers stay consistent, handoffs don’t break. forget less, build more.

0 1 1 17 0

Download Video

Arnab Seal @hermes_seal

3 days ago

Agentic RAG in @substrateco uses autonomous agents to actively retrieve relevant information from various knowledge sources, including documents, past conversations, and data embeddings. By coordinating multiple retrieval systems, these agents find missing context and bring…

0 0 0 14 0

Arnab Seal @hermes_seal

4 days ago

Sharing how Agentic RAG is transforming memory in Substrate, check out our latest update for fresh AI breakthroughs at @substrateco! #AgenticRAG #AIcommunity #GenAI #Substrate #forgetlessbuildmore

Arnab Seal @hermes_seal

4 days ago

Sharing how Agentic RAG is transforming memory in Substrate, check out our latest update for fresh AI breakthroughs at @substrateco! #AgenticRAG #AIcommunity #GenAI #Substrate #forgetlessbuildmore

0 0 1 198 0

0 0 2 58 0

Arnab Seal @hermes_seal

4 days ago

Agentic RAG is now powering smarter memory suggestions at @substrateco! By leveraging multi-modal retrieval and context-aware generation, our system surfaces the right insights for you and your team, even from complex conversation lookups. It’s still experimental, and we’re…

0 0 1 198 0

Kr$na @krishdotdev

5 days ago

What if i fail ? What if you fly ?

7 2 26 2K 1

Arnab Seal @hermes_seal

4 days ago

For us at @substrateco, it’s never been about which AI platform wins. Let them battle it out. We believe infrastructure and memory should be a commodity, again. Here’s the problem: no model vendor wants you to move away from their ecosystem, so they won’t let you share memory…

0 0 0 66 0

Download Image

a16z speedrun 🧊 @speedrun

4 days ago

a startup is a bet against consensus reality. if everyone already agreed with you, it wouldn’t be a startup.

67 95 734 49K 112

Ankur Roy @ankurdorroy

4 days ago

nobody talks about the quiet lock-in here, context + memory + follow-ups means switching costs go up every day. that’s it.

0 1 2 25 0

Substrate @substrateco

4 days ago

context has landed. until now, every ai chat, tool, and model lived in its own silo. no shared memory. no continuity. substrate unifies memory across chats, tools, and llms — so your ai finally remembers what matters. forget less. build more. getsubstrate.co

1 1 1 119 0

Download Image

Arnab Seal @hermes_seal

5 days ago

Excited to see @substrateco gaining real momentum! Hit another 1k mark on the waitlist. We've just started rolling out workspace features so teams can collaborate seamlessly. Now you can easily import conversations from OpenAI and Perplexity, making it easier to bring your AI…