Launching today

Dippr
A data platform served to your agents, with one prompt
164 followers
A data platform served to your agents, with one prompt
164 followers
It’s where your agents build and run your company’s software, tools and apps – with access to your company data across sources, the context to understand it, and the rules to use it safely.





Launch Team

AWS Activate CreditsCredits that fuel your startup’s next milestone
Promoted
i love this positioning. The hardest part of terms building apps with coding agents is the data access, Having federated queries, different permissions, Congrates on the launch...
@dizzzzzz Thanks! We kept hitting the same problem internally our teams wanted to build with AI, but couldn't safely access the data they needed. What would you like to see us build next?
@dizzzzzz You are right. Data access is the biggest pain today.
Hey PH, Michiel here, also one of the people building Dippr.
Short version of what it is: you connect your databases and warehouses, and then people, AI agents and small data apps can all query across them, each with their own access.
I wanted to say a bit about how we thought about the design, because most of it comes from things that annoyed us.
The first was copying data. Nearly every "let your agent talk to your data" setup starts by piping everything into yet another store. We didn't want to build that, so Dippr queries the sources where they are. Postgres, Snowflake, ClickHouse, BigQuery, Redshift, Iceberg, a Google Sheet somebody in finance maintains. You can join across them in plain SQL.
The second is the one I care about most. When an agent runs a query, it runs as you, with your grants, after you've consented to it. We don't hand it a service account that can see the whole warehouse. Same story for apps: if I open a dashboard and you open the same one, we each get the rows we're allowed to see. It made a lot of things harder to build, and I still think it was the right call.
One smaller thing: you can define certified metrics, so when an agent gets asked for revenue it uses your definition instead of making up its own SQL.
There's a CLI called dip that signs you in and hooks up Claude Code, Codex or whatever you use.
Honest question for anyone reading: how are you giving agents access to data today? I'm curious what people have duct-taped together, and which sources you'd want us to add next.
I'm around all day, ask me anything.
@michiel_de_smet Letting agents run queries without giving them more access than they need is a big one! Almost everyone we spoke to struggled with this.
Hoshang, I followed what you had suggested earlier, and it works as expected.
Thank you for building this, I'm going to talk about this at the next meetup I attend here.
@sahrizv Love it! Thanks! Did you also explore the semantic layer?
@hoshang_m Ah, not yet. I don't currently have a use case for that since our data vocabulary is limited, but I guess for our data team it might be a way to centralise the semantics. Will pass it on to them.
@sahrizv Got it. Yes, that would be awesome! Thanks again for spending time on Dippr and I'm glad you found it useful.
@hoshang_m Congratulations. And happy product launch.
@huisong_li Thanks so much for the launch wishes! If you give Dippr a try, I’d love to hear what you’d build first with your own data.
@huisong_li Thank you; it means a lot on launch day! How are you giving your agents access to data today?
Looks like a useful product! Will check it out. Congratulations to the team on launch!
@puneetar Thanks Puneet! What do you plan to build on Dippr?
Interesting that the origin was a ticket queue full of "can my agent access company data?" requests, that's a very real pain. My worry would be the rules layer. When an agent builds an internal tool on top of this, how do you stop it from quietly widening what a given user can see?