Launch HN: Outerbase (YC W23) – A new UI and editor for your database
Accessing data is a challenge to team members who aren’t data analysts or engineers. Databases are usually locked down to a few team members, and everybody else has to rely on them to get access. Most non-engineers can’t (and don't want to) use developer tools, and developers don't want to write SQL for teammates all day. Technical employees end up being bottlenecks for access to data. In some cases this can be extreme—we’ve seen publicly traded companies with only 2 data scientists for the whole org!
Our goal is to make data accessible to everyone who needs it. We have an intuitive spreadsheet-like editor that sits on top of your databases, as well as the capability to save and share queries. You can take those queries to create charts and dashboards for your team. You can also query your data using EZQL, our natural-language-to-SQL conversion. We use OpenAI to power the natural language process, and we pass the relational schema on top so we can easily know the relationships between your tables.
Prior to starting Outerbase, I (Brandon) was a product designer at DigitalOcean and noticed that while DO did a good job making it simple to create databases, there wasn't a modern solution to manage them afterwards. Often users had to use PHPMyAdmin, psql, or $insertDBGUIHere, and to be honest most of them do not provide the best user experience. They’re for a very technical audience, and fall short of making data accessible for everyone. We saw a need to do for data what DigitalOcean did with the droplet.
Brayden led an engineering team at Walmart and dealt with data at a completely different scale. He led the iOS, Android, and web teams for their amends experience and a lot of time was spent pulling, querying, and generating reports on that data. So when we talked about building this he was immediately in.
How it works: We have a React-based frontend that uses a combination of Sequelize and some native libraries to normalize the underlying SQL, which allows us to query and connect to different relational databases. Currently we support Postgres, MySQL, Snowflake, BigQuery, and Redshift. We don't store any of your end data—everything else is encrypted and all credentials are stored in KMS.
Tools like Outerbase make it possible for people to do their jobs more directly. One of our larger customers uses us as a way to moderate what gets posted to their app. Users submit data and our customer will actually go in and mark a column approved if the content is ok for their audience.
Outerbase is available to use today. You can try it for free with 1 user and then if you want to collaborate or use additional features you can upgrade to our pro tier or the obligatory “call us” enterprise tier.
We would love to hear your thoughts on the product, you can sign up today for free, use the sandbox database or connect your own! We know the space isn’t exactly uncrowded, but we hope our approach to building something that is intuitive and collaborative will make it easier for everyone to access their data. We know some HN users are not our target audience because they’re technical and already have tools they’re comfortable with—but even then you might want a tool so your team doesn’t have to bug you as much with data requests! We let you simply give them read access to their db and enable them to do their own queries.
We’d love to hear your views, opinions, experiences about this. What would you want to see f...
81 comments
[ 3.7 ms ] story [ 165 ms ] threadWhat is the difference between Retool, Basedash and Outerbase other than they are all funded by YC?
I'm a big fan of Retool and Basedash and love what they have done, but there's still some initial learnings there to build those experiences, where we want to enable users to come in and immediately see their data and start interacting with it.
I see a future where we end up replacing more traditional BI tools rather than internal tools, if that makes sense?
retool is a rad platfrom, i found it when i was looking for an MS lightswitch replacement, so its clear what it is
also outerbase is a bit strange, its a cloud db interface and the download link is for mac (not windows, mac) , the use cases seem very particular (niche) to me
I'm curious why does use case sound niche? We love to think that giving everyone access to data is a pretty big challenge to solve. Is it in our messaging or do you think the overall vision is off? Would love to understand how we could be more clear in what we are providing.
for internal apps, moving to the cloud is a big step for most places i worked in, its a huge step
I just find it striking, that its considered as a casual thing or the norm, nowadays
I think maybe you might want to explain on your site, why puting your internal db on the cloud is safe, easy ... and not a big deal
But then I saw it's subscriptionware. Nope. I don't use subscriptionware for my personal computing. And the billion-dollar healthcare company I work for doesn't do software subscription unless you're Microsoft, or similarly indispensable. Startups need not apply.
At least the pricing is clear and mostly visible, except for the enterprise tier. Much better than most subscription offerings these days that have even the lowest tiers locked behind a phone call.
I hear you, you can use us for free if you are a personal user. Meaning one connection, 5 saved queries, and a one dashboard. However, we want to be able to support users as they scale with their queries our expenses add up too. I hope you understand!
How would someone who makes software to help you interact with your database and write queries using AI buy food and pay rent, if they didn’t already have money?
Are you implying that if their software was open source, with a self-hostable version, you would use that, and contribute back to the community, which would provide them some service value?
I don’t think the GP comment mentioned open source in any way.
Just FYI, many companies sell software at less than $600/user/year.
[1] https://app.outerbase.com/databases
[2] https://app.outerbase.com/settings
We are working on a resolution to this issue right now. This is an issue with our logout function on the frontend not removing the auth token. There is no risk of scope access outside of your machine.
We will push out a fix shortly.
EDIT: Fix has been deployed. Thanks again!
It now just redirects to the auth page. Clicked on the back button in my browser and voilà
https://imgur.com/a/EvMaVu9
From the screenshot that was provided, they're seeing the client render a page, but it's failing to acquire any data from the API. If they opened the network inspector they'd likely see that the requests are 401'ing after logging out.
I'm not pretending that this is good UX -- it's not -- but it's not evidence of a security issue. That said, we have every intention of nailing down fantastic UX as quickly as possible. (I'm a recent addition to the company but) it pains me personally for anyone to see any mistakes and I hope to impress you soon.
For internal databases we do offer VPN and on-prem (Dockerized) solutions for our enterprise customers.
Private databases that are on localhost we don't _currently_ support on our native Mac app but it's on our roadmap!
Turns out it's just Electron
Disappointed. Will stick to TablePlus
We thought Electron would be a nice interim until we go down the route further.
I know Electron doesn't get a lot of love, is it the file size, the UX, or what about it makes it less desirable for you?
In the comment above you told you have a "native Mac app". I almost got excited only to discover it's Electron, argh
Maybe next time you shouldn't use "native" when it's clearly not native. Just to set expectations right
HN Post: https://news.ycombinator.com/item?id=26853339
As an open-source solution you can a have higher chance of succeeding and a higher retention rate than proprietary ones. You can raise a seed round just as they did
Feel free to e-mail me, I have sold a open-source startup in the past and co-founded another last year
Completely understand the running on own-infra it's something we offer our enterprise customers as it's a bigger lift on our end as well.
Happy to hop on a call or zoom or whatever to chat through all the challenges we ran into while building this, would love to hear more about aidmin as well!
SQLite should work, I just want to test it more before launching. If you'd be interested in beta testing that with me would love to chat more!
For myself I purchase only non-expiring licenses with a one-time fee. It is absolutely fine if I have to pay again for upgrades from time to time, but I want to be able to decide to not upgrade and still use my work for another 20 years, when the vendor is already forgotten. And provided, I still have compatible hardware ;-)
A good example would be Softmaker, which I am happily paying for (but they also nag me every other day, if I wouldn't rather like to switch to subscription).
Here's my use-case. I am fine paying $50 a month for someone to make dashboards, create queries, etc.
However, that's too much (and significantly above other tools) for the "users" of the product: for example, salespeople who I would like to enable to see what products their customers have used. I'd expect something in the 10-20 range per head per month for them, in exchange for more limited functionality.
Some companies solve for this by charging for "creators" (dashboard designers, the people who set up the database connection, or run more custom queries) versus viewers. This encourages broader adoption within a company. $50/mo for any type of user is pretty steep.
We are looking into building out a view-only mode that would be much more affordable. Potentially even embedded experiences so you could share reports internally without needing to actually log into Outerbase.
Would that solve your use case?
Not OP, but this reminds me of Heroku Dataclips and I think you should strongly consider such a product feature.
https://blog.heroku.com/how-dataclips-power-insights-at-hero...
EDIT: Tremendous! Thanks for the reply!
But disallowing them from modifying the DB would be fine.
Maybe limiting it to tables established by the Creator, and with a certain View/Dashboard arrangement, would work.
How would that work, does Outerbase run locally on my server instance?
If it had some functionality for backups and a nice Auth story it could really be a good side project secret weapon.
From a data security perspective, I would be very wary about granting access to our company databases to an online service such as this, but much more comfortable hosting it internally.
Completely understand the concerns around data security, that's not something we take lightly at all. We have other ways of accessing VPCs without having to poke holes in your network as well, including private links, VPN connections, and VPC-to-VPC peering.
Mongo Atlas has a really great database editor / UI built into their web UI but ofc it’s Mongo only.
Curious if you have plans to support connecting to a database over SSH? I have a database running on a VM and I can connect to via an SSH tunnel. Avoids having to expose the database to the internet.
We soon have covered all kinds of flavours on SQL
Mathesar ( https://news.ycombinator.com/item?id=34999774 ) was the last one I tried out but it couldn't support more than 250k rows without crashing and required foreign keys in order to infer relationships. Otherwise it's exactly what I want. An Excel/Airtable interface around a raw database connection.
How does Outerbase compare?
Edit:
Mistypes
To be entirely candid and honest, returning a lot of rows presents problems on our end as well at the moment. It's something we are actively working on improving for large data sets to create the best user experience without having to concern yourself about what you're returning.
For our EZQL (natural language to SQL) you don't necessarily need to have foreign keys to infer relationships between tables. We try to infer those relationships for you.
Also, relationships without foreign keys is on our roadmap: https://github.com/centerofci/mathesar/discussions/2276
I would even take this further and hide the SQL altogether, and make this accessible via Slack/Teams bot.
I need answers from queries in the DB but I don't have DB access and having me write sql queries is a waste of time.
We do have a Slack integration that isn't launched YET but is in internal testing to be released soon. Glad to hear this feedback and know we're on the right track with thoughts here!
Today's large tools require in-depth specialized knowledge on the data as well as how to apply it to create dashboards. Our mission is to create data tools that non-developers can understand, use, and create impact with on day 1.
One feature we're really excited about is how to hold a conversation with your database to fine tune the chart you want in natural language, and have "verified" sources come in to double check what has been produced meets the expectations of the business. Certainly a lot more things on the way as well.
Not true in my experience. True at very large / very mature companies, but at any <5-year-old company the DB will be widely accessible, perhaps universally so. Certainly every engineer, every analyst, and every manager will have full access. (I'm sure someone will pipe in to disagree but this has been true for 10 out of the last 10 companies I've worked with). Coming from DigitalOcean and Walmart it's not surprising you would make this mistake, not trying to dunk just want to suggest this might not be as big of a problem as you think.
In that same vein, in a small company environment you don't harass engineers with query requests, you learn SQL. I've seen hundreds of non-technical users do this (and helped many of them), it's not unusual. Visual query-building tools never help. Even good interfaces like Looker, on top of carefully-crafted data warehouses (which most won't have) are still vastly inferior to SQL and so analysts don't use them. Again, just trying to suggest the problem may not be what you think.
Generating queries with a language model won't work. Real data is far messier than you expect. The table names and field names will not be as literal as you need them to be, and half the database will come with "special instructions" (like "oh you need to divide that field by 100 because...") that will not be inferrable from the names or the data. Many DBs will completely lack foreign keys or consistent key naming. There will often be epochal "eras" in the data where everything before YYYY-MM-DD worked like X, and after that it's Y, and the value of that date is recorded nowhere.
"Currently we support Postgres, MySQL, Snowflake, BigQuery, and Redshift."
I know Java isn't hip but consider relying on JDBC for DB access. There are more DBs out there than you can imagine, and 100% of them support JDBC. Otherwise there will always be some DB you're not supporting, and adding that support will be costly.
Anyway just some food for thought. Good luck. (about me: 18 years in data & analytics)
I agree that users should learn SQL, even when you use EZQL it returns the actual SQL in a fully-editable IDE that can be modified in whatever way you would want.
Do you think that the existing tools out there provide a good environment for those non-technical teams to learn SQL?
Maybe I'm biased but I have never been a huge fan of accessing data from the majority of the existing tools as they feel outdated and like tools built mostly for engineers.
However, it does make sense for their technical team members to launch this for them and give them access to the data.
Some of our features that really help us stand out from this particular pack are the fact you can use EZQL in Outerbase to ask natural language questions on your database without knowing how to query (we'll even expose the SQL if you're interested in what was produced, great for learning SQL). Very helpful for larger databases. In addition to that you can take these queries and create data visualization right in our tool.
meh, i don’t want to allow inbound connections from the internet.
i absolutely hate mode, but i liked their bridge connector.
For the SQL stuff being displayed, we like that it gives non-technical users an opportunity to learn SQL along the way. Now providing it as an optional UI is a worthy consideration for us to perhaps make.
We do have a self-hosted version on our enterprise offering currently but I understand it might not be the most desirable for non-enterprise customers.
I've checked and double checked firewalls to make sure the IP is whitelisted. I'm able to connect with the same setup from my personal computer via psql (with the only difference being the ip whitelisted)by whitelisting it.
Hope you get this working or add some improved connection diagnosist mechanism.
Edit:
If it helps, a POST request to
https://app.outerbase.com/api/v1/workspace/39b72940-083d-4aa...
Fails with a 503 response of > The server does not support SSL connections
... But it's not an SSL connection. I provided a plain ip address.