Aampe builds agent infrastructure for CRM teams. It's the plumbing that lets marketers send everyone their own message at the same time. To deliver on that promise, Aampe's customers need to land Aampe's data in their own warehouse, on their own terms, and use it with their own tools. We sat down with Ben Tengelsen, Data Scientist and Technical Product Lead at Aampe. He walked us through why data sharing is core to the platform, how the team thought about build versus buy, and where analytics is headed next.

Aampe provides agent infrastructure for CRM teams. Tell us a little about the platform and your role on it.

Hi, my name is Ben. I work at Aampe. Aampe is a company that provides agent infrastructure for CRM teams. My background is in economics and data science, and I work at Aampe as a data scientist and also a technical product lead.

For an AI platform like Aampe, getting data into the hands of your customers’ analysts and models is existential. Before joining Aampe, you were on the receiving end of data shares yourself. What did that experience teach you?

In my previous job, when I was leading a data team, we had data shared with us, and it was easily my favorite way to receive data because it was just kind of no effort. It just shows up. I go into some admin UI and I click accept, and then the data is just there.

It is like a glorious gift when a company is willing to share data with you in that way. And so of course, that’s what we wanted to do. When we were standing up this data share, we just wanted the data to just be there, and we didn’t want to have our customers burn compute to bring it in, to manage connectors or other things that could potentially break.

When you started evaluating how to deliver data to customers at scale, what were the non-negotiables?

What we were looking for was a solution that was secure, affordable, and would also allow us to comply with really tricky data residency requirements — because we have customers all over the globe and need to comply with all the agreements that we have in place with them.

Every engineering team wrestles with build-versus-buy on a problem like this. Walk us through that decision.

The other question we wrestled with was, should we just build this ourselves? Because it’s just pipes. We’re just sending data from one place to another. Maybe it’s not so tricky.

But the surface area of this problem grows and grows as you start getting into the details. How do I want to manage the data as it’s changing and growing, and I’m adding new tables and I’m taking tables away? How do I manage resync? How do I make sure that all of this is done securely? How do I make sure that there’s no downtime?

And then there’s also this long list of places that you can send this data. I can send it to Snowflake, I can send it to Databricks, I can send it to buckets, I can send it to Redshift. There are so many different places that I can send this data.

You hold up this list to your data engineers and they’re like, this is going to take us — we will be doing nothing else this year if we build out infrastructure to do this. And as a product person, I know all the other things I’d love our engineers to be building, and I’m acutely aware of how precious their time is. So it was an easy decision. It was a 100% no-brainer that this is a buy instead of a build.

Talk a bit about what you’re building with Bobsled and what the experience is like for Aampe’s customers.

My initial impression of Bobsled was extremely positive. Their docs are excellent. Their Slack channel is extremely responsive. And then the product itself just has everything that you need, and it’s easy to find what you need. I was able to stand up our first data share within minutes of logging into Bobsled for the first time — and that was just like, yes, this is how it’s supposed to be.

It is an awesome feeling when a customer says, “Hey, we’d like to set up a data share.” They tell me the three or four things that I need to know, and then within minutes of them asking for a data share, we have it turned on — and they have it. There’s no better way to blow the socks off your customer than to be so immediately responsive to something that they want.

What’s really cool to see is when customers use our data share and start using the information in the data that we’re sending over in cool ways. Sometimes they just send us screenshots of reports or other things that they set up. What we’re really helping our CRM teams do is navigate this new world where you can send everyone their own message at the same time.

Zooming out, analytics is changing fast, especially inside AI platforms. How is that shaping what Aampe delivers?

Analytics is changing really quickly right now. Of course, within the Aampe product we have a set of charts that are just part of the Aampe UI. But the charts that you put inside your tool are increasingly small in what a customer has access to. They can now create their own charts. They can bring their own tools, they can bring their own machine learning models. They can bring their own — just a growing and growing list of tools to understand this data.

Once that data foundation is in place, what part of the customer experience is your team focused on unlocking next?

Once the data foundation is really set, then the question is, well, what tools can you provide to make it very easy for analysts and data scientists and other citizens of the world to be successful in tackling this data?

This comes down to a public GitHub repo with examples, with AI files and context files, with potentially libraries. Certainly lots of example queries. Basically anything that you would want to feed into an AI agent, anything that you would want to use on your own — that then really unlocks the value of this data.

Interested in how Bobsled helps AI platforms like Aampe deliver data to their customers? Get in touch to learn more.