Hi HN! I'm Yiğit, founder of CoreBase. CoreBase is an infrastructure layer that lets products embed AI which answers from live data, scoped either to an end-users own data or to internal company data with guardrails.
I built CoreBase after I talked with local ISVs and some software service providers. The main problem they had is they want to integrate AI into their own platforms but either they dont have the time or they dont have the capacity. So I built CoreBase for them. I think the hard part is not the model anymore, its the infrastructure.
The problem for my situation is: Giving an agent access to the customers real data and connections without losing control of what it can see and do. CoreBase is that layer: agents answer from live data, take actions in your specified border and everything is scoped and audited. Today, CoreBase has these features, though it's still in the early stage:
- Every end-user can connect their own accounts with OAuth and only see their own data.
- The admin can set guardrails for the agents via visual builder.
- It supports REST, GraphQL APIs, cloud databases, vector databases and on-prem databases via CoreMCP.
- For API connections it support custom login flows and custom tools.
- BYOK LLMs
- Prompt injection and jailbreak protection
- Database schema annotations and glossary.
- Index files on the go, so agent can build context.
and some more.
Hi HN! I'm Yiğit, founder of CoreBase. CoreBase is an infrastructure layer that lets products embed AI which answers from live data, scoped either to an end-users own data or to internal company data with guardrails.
I built CoreBase after I talked with local ISVs and some software service providers. The main problem they had is they want to integrate AI into their own platforms but either they dont have the time or they dont have the capacity. So I built CoreBase for them. I think the hard part is not the model anymore, its the infrastructure.
The problem for my situation is: Giving an agent access to the customers real data and connections without losing control of what it can see and do. CoreBase is that layer: agents answer from live data, take actions in your specified border and everything is scoped and audited. Today, CoreBase has these features, though it's still in the early stage:
- Every end-user can connect their own accounts with OAuth and only see their own data. - The admin can set guardrails for the agents via visual builder. - It supports REST, GraphQL APIs, cloud databases, vector databases and on-prem databases via CoreMCP. - For API connections it support custom login flows and custom tools. - BYOK LLMs - Prompt injection and jailbreak protection - Database schema annotations and glossary. - Index files on the go, so agent can build context. and some more.
You can check the docs :) https://docs.corebasehq.com/
I'm 16 and building CoreBase full time for last couple of months. And I want to get some feedback to draw a path. Thanks!