Consensus raises $30M Series B to expand from academic search into a full research workspace

Speaker 1: Yeah.

Speaker 11: And so there are a million homes missing.

Speaker 1: Yeah. It's rough. Well, thank you so much for taking the time.

Speaker 5: Thanks, guys. Great great to

Speaker 6: finally have you on

Speaker 3: the show.

Speaker 10: Thank you.

Speaker 11: You all. Thanks.

Speaker 5: Come back soon.

Speaker 1: Our next guest is Eric Olson from Consensus back on the show with an exciting fundraising announcement. He's in the waiting room. So we'll bring him in to the TBPN Ultradome.

Speaker 8: Eric, how

Speaker 1: are you doing?

Speaker 9: Well, it's fantastic. How are you guys doing?

Speaker 5: We're doing great.

Speaker 1: Please reintroduce yourself and the company for the for the listeners and then we can go into the news today.

Speaker 9: Love it. So I'm Eric, the cofounder of Consensus. Yeah. We're building AI for academic and scientific researchers.

Speaker 5: Yeah.

Speaker 9: To date, the product's been very search focused, focused on the literature review use case, people finding references for whatever work they're doing, and we are announcing a $30,000,000 series b today led by Great Point Ventures. Take the product beyond search, move into more of a workspace for researchers.

Speaker 1: How how has the actual go to market been? Obviously, you're raising more money, so I imagine there's been traction. What's the response been like? What's the customer acquisition strategy?

Speaker 9: Yeah. I mean, that has been far and away the best part of this entire process is the amazing work we get to see our users doing. I mean, these are people who are genuinely changing the world from the research they're doing. Like most AI products, you know, we've gone mostly straight to consumer. Mhmm. There's so much on the ground demand for tools like this that we're really growing mostly organically. You know, we're in the business where so many people are doing work in a lab or at a university, and there's tons of networks affect sharing. So we've leaned into that above all else. And then more and more recently over the last year, we now work with over a 100 universities where we sell usually directly to the library, and then they'll distribute the product to students themselves.

Speaker 1: Interesting. Smart. Where are the models or the existing tools sort of falling down? Because it feels like anything that has a very, very short reinforcement, you know, reinforcement loop like research or coding can be automated like very quickly. You can get really great results. But stuff like fill up that test tube, throw it in the centrifuge, even if that just takes a few minutes, it's harder to actually close the loop there quickly. And so that's always been my thesis for why we're seeing such rapid advancement in cyber security because you can run it in simulation, video games. But stuff in the offline world, anything that needs IRL feedback is slowed. Do you have an opinion about where all this goes?

Speaker 9: Yeah. I mean, there are people who are going after full autonomous scientists. Yeah. You push a button, discoveries come out. We're kind of betting against that. Mhmm. You know, I think, like you said, there are many parts of a research process that can be automated. Mhmm. We want to focus on those parts.

Speaker 7: Mhmm.

Speaker 2: But there

Speaker 9: are many things that happen outside of that that are very uniquely human. Mhmm. And it isn't just the putting the atoms together or putting the thing in the test tube. There's lots of, you know, taking things from different domains, understanding the connections between them, talking to people, working through ideas and coming up with these new ideas. That is the essence of science. It isn't just what's in the test tube. And, you know, despite how good these models have gotten, still not great at drawing connections between things and coming up with a new idea. So we wanna focus on the things that are actually automatable, like searching for papers and running many iterative searches and finding the right materials, and let humans and scientists focus on the things that are still core to science in that collaboration, in that inquiry, in that discussion.

Speaker 5: Yeah. Is the bar being raised on college campuses? Like, our our students and researchers, now that they have AI just expected to do two, three times as much work in a given period?

Speaker 9: I mean, to some degree, absolutely. At the same time, you also will see lots of professors that will put up lots of, you know, blockers to people using AI and forcing them to do, you know, proctor tests and making sure that they're doing things that are not AI enabled. So I think that there are certain parts of it that are sped up immensely, then there's other parts of it where professors are putting lots of safeguards up to ensure that everything is being done in the classroom exactly the way that they want. So I think it's kind of a little bit of a mix of both.

Speaker 1: How do you think about harnesses, wrappers, just SaaS that can accelerate different pieces of the research puzzle to sort of create that operating system? Like, are you drawing from the analogy of like the IDE, the coding agent, know, Silicon Valley has moved really aggressively towards like coding agents are all you need, but cursor is still doing really well. So, you know, a lot of code review will still happen even in a world where AI is writing a lot of the code. So walk me through sort of how you're thinking about the future of research where there's a huge AI portion but the researcher still remains in the cockpit.

Speaker 9: I think IDE is a really great corollary because you want there to be these things where you have this harness where it can do these airative loops. You can give it feedback to say, you know, go down that path or that path and pull in the things you're looking for. I think the one difference between us and maybe a coding use case is there are very discrete steps that happen sequentially. So maybe you're in a searching use case, but typically then you're moving to like, oh, now I'm writing this paper. Now I'm deeply analyzing this data. So it's kind of like two actually different surfaces than just one ID.

Speaker 1: Yeah.

Speaker 9: So you wanna be able to build a workspace that makes it a really easy transition from, I'm in this search, I'm in this learn discovery mode, but now I need to transition to this new task. Yeah. That same thing is still present. You wanna be able to iterate with a model and say, hey, take me down this path. I wanna explore this idea. I wanna write this. I wanna edit it. Okay. But now I wanna go back to search because I just wrote this paragraph, but now I need to find references that support that paragraph. Maybe you're taking that paper and then you're going to the next step, which could be you're running some real life experiment. Maybe you're getting off of your surface you're in. So, like, I think it is each sequence is very similar. I think the difference between us and a lot of other use cases is that there are these very distinct surfaces you need to build for each part of a research process.

Speaker 1: How are you thinking about integration with lab notebooks? Is that gonna be important at some point? Is it already important? I've been, there's been a big push for, like, cursor for bio, but I've always had struggle I've always struggled to understand that pitch because there is no Versus code for bio. All of the electronic lab notebooks, the ELN market is closed source typically. And so but I imagine that with the right API integration, the right partnership, you can still partner with those. Is that in the roadmap? Is that in the sweet spot or something maybe down the road?

Speaker 9: Absolutely. But I think you're bringing up another great point that even in each one of those, like, broad buckets, so now if you're moving to, like, some sort of writing note taking use case

Speaker 1: Yep.

Speaker 9: There are different surfaces for different types of research.

Speaker 1: Sure.

Speaker 9: Like, there's your bench links for biotech Mhmm. And then there's, you know, just word in Google Docs if you're a graduate researcher or something. Yeah. We wanna both create APIs that integrate with those so you can do it directly in those services. We also think we can build our own dedicated services too.

Speaker 1: What about for What about for scientists that use sort of like hosted iPython notebooks that might be on like Google Collab or something like that. Some light hosted. Do you want to build that stack so you can sort of provide that or or do you want to partner with the existing tools?

Speaker 9: Why not both?

Speaker 1: Why not both? Okay. Well, you have the money to do both. So good luck and congratulations and thank you for stopping by the show.