Here’s what I tested and liked
1. Easy configuration of hosts and switching between them. Added and modified some dbs and the tool was able to catch that.
2. Step by step resolution of queries. Tested this with the query you provided and it navigated well through the 109 tables and gave a detailed approach. I saw the agent encountering problems in the middle and resolving it during execution. The sql editor, chat window with descriptions around each query, and the final TLDR answer response is amazing.
3. Entire resolution took less than a minute. On checking each table that it went through, I’m fairly certain I would have spent a lot of time analyzing and writing queries.
4. The chat window has all the right tools to manage context. I liked how it detects and adds context automatically.
Hey 👋 everyone!
Shiva here, Co-Founder of Incerto. Think of Incerto like "Cursor for databases" — a copilot to interact and work on databases.
🚀 Why are we building this?
There’s no true AI-native database tool built from the ground up with AI Agents at its core — and the few that exist still lack the right context to generate correct queries.
Working with a database isn’t just about connecting AI — it’s about understanding schemas, tables, indexes, workloads, and business use cases.
Current tools either miss real context or suffer from context bloat.
🔧 What can you do?
Perform everyday database tasks — from writing and optimising queries to creating schemas, adding data, and managing indexes.
Automate changes e.g., split a column into a new table, validate data, and build the right indexes.
Troubleshoot issues e.g., investigate "Why is memory high between 9–10 AM every day?"
Speed up workflows, add business templates to auto-create schemas, seed data, or generate queries for APIs.
👩💻 Who is it for?
Developers, DBAs, Data Engineers — anyone who works with databases.
Teams that want to automate repetitive tasks (write/optimise queries, move data, change schemas etc).
Anyone who’s said things like:
"Why is this AI creating a `users` table? Our `users` table is completely different."
"I just wanted to split a column, why did it break my schema?"
"I copied this query from ChatGPT, but it doesn’t even run on my DB."
🤔 "Is this just another Text-2-SQL?"
Nope. SQL generation is only one small part. In fact, we like to call it "Text-to-Task":
You define a task in natural language.
Our agents gather the right context.
They collaborate to complete the task end-to-end.
🔒 Security Concerns?
Runs locally — no data sent to our servers.
Only outbound calls are to LLM providers.
Database credentials are stored securely (hashed locally).
✅ How to try it?
👉 Download from our website - https://incerto.in/
🎥 Watch the installation demo - https://youtu.be/AZx8VrAP838
💬 Ask us anything below, we’ll be here all day.
Got a use case in mind? Drop it in comments, we’ll help you build it.
🙏 Thank you, Product Hunt community!
@shivapundir Good luck
@mustapha_ajermou1 Thank you :D
@shivapundir Congrats Shiva, love the “text-to-task” framing. How do you see Incerto fitting into DB workflows—daily copilot or a power tool for troubleshooting?
@imnikhill10 Thank you ❤️
@shivapundir Congratulation on the launch.
@porush_puri Thank you ❤️
@shivapundir Congrats on the launch, Love the ‘Cursor for databases’ positioning — the focus on true context instead of just text-to-SQL makes a ton of sense. Super curious about how Incerto handles schema drift and query optimization in real-world messy DBs. Excited to try it out!
@nishant_dubey2 On schema drift, the idea is to keep the agent aware of changes as they happen, so it doesn’t get stuck on outdated snapshots (although this is a hard problem in itself, as changes to the schema can come from multiple sources). For query optimization, the agents can't automate everything. Just like Cursor you would be required to accept and handle changes.
Congrats on the launch, Shiva 🚀 Love how you’ve built Incerto with context-awareness at its core. I feel that’s usually the missing piece when AI meets databases.
We’ve seen something similar on the strategy side with Escape Velocity AI: without the right context, outputs feel shallow or generic. Once you layer in structure and domain-specific logic, it changes the game.
Do you see early adopters using Incerto more for day-to-day query writing, or for heavier schema-level work?
@andreitudor14 We see it majorly for day-to-day queries and generating database related code.
Schema level works are rarer in comparison.
Production issues are rare, but when that happens, our customers just use our tool to solve it. It is has all the context around firing issues (deterministically tracked)
@anurag_pandey19 That makes sense, as day-to-day queries are where the time savings really add up. Interesting that you’re also covering production issues when they come up, since context-tracking there feels like a real differentiator. Do you see teams trusting Incerto as a primary tool over time, or more as a companion alongside their existing workflows?
@andreitudor14
Its primary tool for all our clients, they are doing north of 20 Million tokens a day -- in just 3-4 people team.
There interaction is constantly through 10 am - 6 am, meaning they are sort of using it as we use VSCode or cursor. This is for generating queries for professionals who do this very often.
If a business head is using Incerto analysis they are not gong to spend more than 6 hours on product a week.
This is awesome, Love the “Cursor for databases” positioning, devs definitely need more than just Text-to-SQL. Curious, have early users been leaning more on Incerto for query optimization or for schema changes/management?
@musa_molla most used feature is text to task. To do ops. make queries, get code which depends on database.
thanks for the comment!
@musa_molla Thanks!
Feedback from early users has been a mix of use cases, but a lot of lean using agents to do ad-hoc tasks. You can even do parallel queries through the agent across different databases.
From highlighting slow queries, health issues, to solving the problems all the way Incerto has been able to help.
Incerto was also able to do a production issue RCA in 10 mins saving them a lot of downtime.
Whoa, this could be a game-changer!
An AI that can write queries and handle schema changes for databases with natural language?
That's insane. Could seriously save so much time
congratulations on your launch
@nookesh_karri thanks for the kind comment, yes we believe this can really pace up development and lot of analytical and ops tasks.
@nookesh_karri Thank you ❤️
Let's not forget - all of these BUT with the right context :D
What does AI-native mean? Isn't it more like being pregnant, either you are or you are not, there is no pregnant-native. 😅
@michacassola
I can tell you why we use "AI Native", can't debate the around literal meaning. Product has been built around the intelligence (AI) as it's core. Every decision we make is affected by that, because at the end the core flow is LLM powered. Every new feature we add, which even can be used without AI has context tagged along with it, so that its LLM comprehensible (formats, structure, core prompt for what a piece of context represents matters).
We would have just called it AI Database Copilot, but the need to differential from "Sprinkle of AI" vs "Built around AI" because around Dec 2024 our product was 90% deterministic with some AI powered feature, but later we re did the whole thing when Claude got so good, and we realized that AI making the decision and asking user for approval is better than deterministically suggesting and then use AI to modify.
Time spend on figuring out prompts, contexts, evaluating the outputs and worst case scenario is more than time writing code.
Hope it makes sense.
No comments on pregnant-native part.
@anurag_pandey19
What pun?
Ok: AI-native = Mostly AI
😀
@michacassola yeah, core is AI. AI gets better our products core value automatically improve.
Congrats on the launch, @shivapundir Shiva! Really like how you’re going beyond the usual text-to-SQL tools. The text-to-task angle makes it feel way more useful in work.
Quick question though, how good is Incerto right now with bigger, chained tasks (like tweaking schemas, validating data, and then optimising queries all in one go)? Can it already handle that smoothly, or is that still on the roadmap?
@lina_huchok
We have separate workflow just for optimizing queires for both postgres and clickhouse -- they work amazingly well.
It is designed to work on long tasks with help of different agents and smart context pruning.
It has successfully solved production issues in live environments, by detecting what is wrong and then also fixing it. It uses root cause agent + verified loop for detecting problems.
There will be a moment when context gets too big -- but that comes after hours of using same chat.
Also we have made it multi host, where you can execute multiple queries in different DBMS (and different workers of those) in one go. Making it suitable for production use cases.
Please do try out the product from incerto.in/download
If you want to join the group of folks using Incerto already, fill this : https://docs.google.com/forms/d/e/1FAIpQLSfjlDLP6CXul6KandpwgboJDaG-iNm4Ie_sqIR9vED55V4Lkg/viewform
@anurag_pandey19 Thanks for sharing, sounds really impressive!
Thank you ❤️ @lina_huchok . I hope Anurag answered your question.
Feel free to continue the conversation on LI as well :)
Loved the product. Had been struggling with my n8n agent which writes shitty queries, irrespective of how many times the schema is provided.
Question: What are you using for evals? How are you making sure of the performance of your AI copilot?
@additi thanks!
There are certain steps to it :
1. Understand what user is asking -- clarification agent does that.
2. text to SQL accuracy -- this is achieved with schema linking, keeping only relevant context and giving lot of context (around deployement, past queries, table schema, values of fields, semantic layer on data)
3. Doing specialized tasks -- we have different agent for notably useful tasks. E.g. Root Cause agent which is used to find problems just hypothesises and verifier agent tests then approves or disapproves.
4. We also list out the assumptions at the end of the conversation and a accuracy score (though LLM generated only)
5. Complex tasks [still not GA] : We will split complex tasks into mermaid dependency graph and execute smaller tasks with human feedback.
6. There ability for users to add more contexts (through templates), ability to change the system prompt (for specialized niche tasks)
Hope that answers, looking forward to your feedback on product!
@anurag_pandey19 Great!
You can check out https://noveum.ai/en. This is our product, we help companies run evals and make sure that their AI agents are reliable. Let me know if you would want to understand more.
@additi
Sure that actually sounds useful, we spend quite sometime evaluating the output. Would love to connect!
Thank you ❤️ @additi
I hope you have connected with Anurag on LI to take the conversation forward :)