Demo • Amoeba AI — From Dashboards to Decision Agents

1858
05:22:07.290 –> 05:22:08.869
Julia Nimchinski: Go to the show again.

1859
05:22:09.130 –> 05:22:10.600
Tooba Durraze: Thanks for having me.

1860
05:22:11.380 –> 05:22:15.330
Julia Nimchinski: Our pleasure. Are we seeing your product release?

1861
05:22:15.480 –> 05:22:16.010
Julia Nimchinski: Today.

1862
05:22:16.010 –> 05:22:24.589
Tooba Durraze: Yeah. Yes, you are. And we happen to time our product releases, through this great summit, always, because we feel like the audience is

1863
05:22:24.610 –> 05:22:36.000
Tooba Durraze: So great, and learned so well. So, yes, here’s another exclusive peek into our product. But let me start with giving a bit of background into myself. So again, name’s Tuba, I’m the CEO and founder of Amoeba.

1864
05:22:36.000 –> 05:22:52.659
Tooba Durraze: I think one of the things that is really unique about Amoeba, we’re building a brain for your go-to-market business. So think about anything related to your data that’s related to go-to-market, we essentially provide you the intelligence layer that extracts the right outputs for you.

1865
05:22:52.660 –> 05:23:06.580
Tooba Durraze: But we’re a little bit different in the sense we’re not an LLM-based platform, so if you think about large language models or transformers, the thing that they’re really good at are things like predicting the next word, so a lot of, like, content-based use cases.

1866
05:23:06.580 –> 05:23:19.790
Tooba Durraze: But the things that they are sometimes really bad at are the reliability of output, and, like, we talked about that earlier in the panel, like, deterministic outputs, etc. So we’ve developed a neurosymbolic AI system

1867
05:23:19.790 –> 05:23:36.269
Tooba Durraze: And essentially, it’s like adding a reasoning layer on top of your large language model without you actually having to do it. So it’s learning from your data, it’s creating that reasoning, on top of your data, so for you as a business owner, it’s a very turnkey solution.

1868
05:23:36.440 –> 05:23:50.219
Tooba Durraze: As Julia knows, I’ve transitioned from talking about these things as, like, neural nets, because it’s a very geeky product, into what more… how the product works practically. So, I’m going to dive into the product really quickly.

1869
05:23:52.910 –> 05:24:04.909
Tooba Durraze: And follow along with me here. One of the things that I want everyone to think about is, how you consume data, right? That is at the center of everything. How you consume data dictates, sort of, how you end up

1870
05:24:04.910 –> 05:24:24.509
Tooba Durraze: running your business or what you extract out of consuming your data. So the way we’ve designed this platform, it’s equal parts to emulate you having a data scientist or data analyst on your team, but it mostly transitions into, what we would call as an observability layer for your business, which means, like, at some point, you can walk off

1871
05:24:24.510 –> 05:24:32.090
Tooba Durraze: Like, hands off, and, like, let the system run itself, and the system’s responsible for telling you what’s important and recommending what you should do against it.

1872
05:24:32.370 –> 05:24:42.919
Tooba Durraze: So if we talk about, kind of, again, back to how we consume data, one of the biggest ways we consume data is, you know, business data, if you look at spreadsheets right now, or you look at reports.

1873
05:24:42.940 –> 05:24:58.840
Tooba Durraze: And our answer to that is, obviously, spreadsheets and reports are really good for structured data, and sometimes you still have a hard time extracting insights, but they’re not great for unstructured data, which are things like your Zoom recordings, which are things like your content, etc.

1874
05:24:58.840 –> 05:25:04.930
Tooba Durraze: So Amya has this concept called explorations, where you can come in and essentially look at

1875
05:25:04.930 –> 05:25:11.999
Tooba Durraze: I’m… for… in this… for the sake of time, I’m actually not going to ask the question, because big data products obviously have, like, a latency issue, so I’ll walk you through some…

1876
05:25:12.000 –> 05:25:35.839
Tooba Durraze: responses, but you can imagine this is, like, your deep researcher, in essence, that can go in and look at you. I’ve created one here for high-value MQLs, and what it did was essentially first gave you insight into how it’s doing it behind the scenes, but then went in, looked at everything that was relevant and interesting related to that query that I had asked, gave you a summary of, like, what you should change, what you would change.

1877
05:25:35.840 –> 05:25:41.019
Tooba Durraze: But it’s the equivalent of if I was to tap my analyst and say, hey, can you analyze this for me?

1878
05:25:41.020 –> 05:25:43.590
Tooba Durraze: Right? So that’s one way you consume data.

1879
05:25:43.590 –> 05:25:58.570
Tooba Durraze: Now, another way we consume data is my favorite, which is, dashboards. And, you know, we all are responsible, for… have been responsible for setting up dashboards, maintaining dashboards, or taking our dashboards into our executive meetings, etc.

1880
05:25:58.570 –> 05:26:10.489
Tooba Durraze: One of the things that’s unique about Amoebus Agents is we ask you, essentially, what is a goal that you’re trying to achieve? And your goal could be you want your pipeline to grow by 20%, you want your business to grow by 30%.

1881
05:26:10.490 –> 05:26:13.009
Tooba Durraze: So think at the level that’s that macro.

1882
05:26:13.010 –> 05:26:34.730
Tooba Durraze: And we have something interesting, which is UX on the fly, so essentially based on what your goal is, is what you see represented on your screen. So if your goal is, like, a growth-based goal, essentially you’ll see different things, or different compositions of things, but if your goal is, like, a maintenance goal, you’ll see different compositions of things, and that’s because our

1883
05:26:34.730 –> 05:26:44.050
Tooba Durraze: Agents are smart enough to pick what are the right pieces to show you, instead of showing you the same footprint for every single goal that you set, which is one of the problems with dashboards.

1884
05:26:44.060 –> 05:26:56.169
Tooba Durraze: So, and the other thing we do is, like, once you set a goal, your neural net is on and observing continuously, right? So again, you set a goal and you back off. So if I look at what’s in the composition of a goal, you see a couple of different things.

1885
05:26:56.170 –> 05:27:16.169
Tooba Durraze: One, you see, obviously, how you’re progressing against the goal. Now, again, these summaries are generated… the data is generated via the neurosymbolic system, and then we use LLM to synthesize, right? So these summaries, in part, are meant to tell you, hey, these things are changing day over day, week over week, so it doesn’t look the same every time you come back to the platform.

1886
05:27:16.180 –> 05:27:29.010
Tooba Durraze: But then the first thing we do is reveal to you, like, what are the insights based on the data that you’ve connected to this related to your goal, and these are ranked in order of, like, preference or priority, based on what’s most relevant to that goal.

1887
05:27:29.010 –> 05:27:40.709
Tooba Durraze: And they are multivariate in nature, so sometimes they will… might include a single variable, sometimes they might include multiple variables. A great example of that is, one of our customers had

1888
05:27:40.710 –> 05:27:52.199
Tooba Durraze: a particular blog that converted really, really well, with a single color of a CTA. So again, it’s as expansive and as inclusive of any and all data that you can connect to it.

1889
05:27:52.280 –> 05:28:05.829
Tooba Durraze: And as you can see with this little squiggly line here, that if the… if the footprint changes, meaning if this relationship of this variable to the data changes, it moves up and down. So, if I’m a marketer, and I’m coming in here, and…

1890
05:28:05.830 –> 05:28:20.840
Tooba Durraze: either my team or my C-suite execs are asking me, you know, what’s working, which is, like, the first question we get asked all the time. You no longer have to go digging across your entire data warehouse. You can just show them, against this goal, these are the activities.

1891
05:28:20.840 –> 05:28:28.020
Tooba Durraze: that are actually working. This is a new way of looking at what was traditionally called attribution. Now, we don’t just stop there.

1892
05:28:28.370 –> 05:28:52.589
Tooba Durraze: Essentially, the next thing we do is obviously give you recommendations. And we’re very precise about our recommendations, because sometimes you get recommendations and it gets overwhelming, because there are a bajillion recommendations to take. So for us, our concept of recommendation is, based on all the insights shared, in order for you to achieve this goal, what are some things you can do based on the systems that are connected? And now again, these systems could be

1893
05:28:52.590 –> 05:28:58.710
Tooba Durraze: like Lisa was showing, metadata as an example. These systems could be outbounding systems. These systems could be…

1894
05:28:58.710 –> 05:29:03.259
Tooba Durraze: more human-centric systems as well. So any and all systems that are connected.

1895
05:29:03.260 –> 05:29:19.750
Tooba Durraze: And we not only give you high-level recommendations, but when you click on them, we basically give you multiple ways in which you can go ahead and actually try this out and implement it. And the reason we do that is because we value human participation where intuition and strategy is concerned.

1896
05:29:19.750 –> 05:29:27.010
Tooba Durraze: So this tool is giving you everything at your disposal to be able to take and apply a strategy without, you know.

1897
05:29:27.010 –> 05:29:46.429
Tooba Durraze: doing it for you autonomously. So again, you are human in the loop, so you’re picking the simulation or the scenario that you feel like might work best to you. And then say you pick a scenario that works best for you, you can obviously go ahead and generate tasks, which is basically a project plan in order for you to implement it. Now, behind the scenes, for us at Amoeba, because we use our own product.

1898
05:29:46.430 –> 05:29:54.519
Tooba Durraze: we automated this, so it not just generates tasks, but it actually, like, goes into those systems and starts orchestrating, so that’s always an option. But…

1899
05:29:54.520 –> 05:30:07.699
Tooba Durraze: as you remember from dashboards, it’s basically bridging the gap from, like, how you’re used to consuming static dashboards into a more living thing that you’re consuming. And then last but not least, I would say the way, execs

1900
05:30:08.020 –> 05:30:17.530
Tooba Durraze: basically take in information to consume data is in the form of long-form analysis. So, as an example, you’re going into a board meeting, or you’re going into, like, your QBRs.

1901
05:30:17.530 –> 05:30:42.459
Tooba Durraze: or any kind of business review, etc, and you want to understand, like, for this quarter, like, what actually happened. And these are longer, these are denser, these are meant to give you a more strategic output on, like, what happened and what you should change. So there’s a bit of a historical outlook. So you can go ahead and kind of create that here, and to give you an example, this is a quarterly pipeline intelligence report, essentially. So, obviously, there’s a bit of a summary.

1902
05:30:42.960 –> 05:30:56.800
Tooba Durraze: First and foremost, it gives you a causal analysis, so telling you why something is happening, not just it’s happening. And as you can see, if I’m scrolling through it, it gets very dense and deep, so maybe not all these pieces are relevant for all the stakeholders within your organization.

1903
05:30:56.800 –> 05:31:06.559
Tooba Durraze: But obviously, some of this is relevant for your… a good summary is, like, relevant for your executive team, a good summary is, like, relevant for your, for your board.

1904
05:31:06.560 –> 05:31:23.460
Tooba Durraze: Now, I talked through a bunch of different products, and one of the reasons why we’re dialing in on how you consume data is we want everyone in the company to have a unified data language, essentially. That was one of the problems with go-to-market data, is that everyone interprets data differently.

1905
05:31:23.460 –> 05:31:42.460
Tooba Durraze: Right? So we’ve designed this thing called our monitoring feature as well, where you can go ahead and say that this topic, this goal, or this exploration is interesting to me, and once you turn monitoring on, then it is the responsibility of the system to, within your systems of action and systems of record.

1906
05:31:42.460 –> 05:31:55.299
Tooba Durraze: to alert you when something is going wrong, not happening, any kind of anomalies, any kind of trends and spikes. The reason the topic thing is really interesting, because if you imagine a neural net, which is a node with a bunch of nodes around it.

1907
05:31:55.300 –> 05:32:15.190
Tooba Durraze: You can basically assign the same tag or the same topic across all these different features. So, say I’m tagging this as channel performance, then I might have a channel performance here, an exploration, a goal, and then I might have a report that’s also channel performance limited. So, if I now turn on notifications or alerts for that.

1908
05:32:15.190 –> 05:32:28.369
Tooba Durraze: then that system is now monitoring not just the things that you were interested in, like, pulling the thread on, but the goal that you were trying to achieve related to it, and then the briefings that you were pulling. So every output from here on out, after you start monitoring it.

1909
05:32:28.370 –> 05:32:36.359
Tooba Durraze: Even, there’s this passive output, which is what we push to you, but active output, which is that you get from the system, is now adhering to

1910
05:32:36.360 –> 05:32:45.549
Tooba Durraze: what your business brain is. So it’s like a… this entire thing is a business brain, and then you get to individually create, like, end user-based business brains.

1911
05:32:45.610 –> 05:32:53.259
Tooba Durraze: I’ll pause here. Obviously, we do all the things of connecting to a bunch of different kinds of data, like any kind of structured or unstructured data.

1912
05:32:53.260 –> 05:33:16.819
Tooba Durraze: But I think the idea here is, like, over time, we’re planning on changing your entire data interaction model and transitioning you from, you know, having to ask your data questions and pull information out to letting the system then give you that information, and then eventually progressing into agent-to-agent orchestration as well, where you can just say, I approve this, and then it goes ahead and orchestrates that amongst many platforms.

1913
05:33:17.950 –> 05:33:36.809
Julia Nimchinski: That’s phenomenal, Tuba. I am just biased, so I’m gonna leave my comments aside, you know, the admiration to your platform, and thinking, and the philosophy, and even, you know, the small details, that it starts with a goal, and not just a simple question. Love the way you’re building it.

1914
05:33:36.950 –> 05:33:46.550
Julia Nimchinski: We are seeing a couple of questions here, and people are asking, how does FALS integrate with existing data systems?

1915
05:33:47.350 –> 05:34:00.370
Tooba Durraze: So, we integrate with… we can integrate with any and every kind of data source. So, we mostly integrate with data warehouses or any kind of, like, MarTech solutions right now, advertising platforms, CRMs, etc.

1916
05:34:00.370 –> 05:34:10.520
Tooba Durraze: And if there is a connector that we’re not connected into, then we generally build, like, a custom connector. The advantage of where we are with AI, like MCPs and all of these tools are at our disposal to easily connect.

1917
05:34:10.520 –> 05:34:29.799
Tooba Durraze: But I think one of the other unique advantages is you can port in structured and unstructured. So, for some of our customers, like, bringing in some, like, Gong data, as an example, alongside this data has been really helpful, or bringing in what’s on your website alongside some of this, like, qualitative data and quantitative data has been really helpful.

1918
05:34:31.140 –> 05:34:34.459
Julia Nimchinski: And can you speak more about,

1919
05:34:34.570 –> 05:34:39.630
Julia Nimchinski: The algorithm or customization of insights for different teams.

1920
05:34:40.340 –> 05:34:56.850
Tooba Durraze: Yeah, so we’ve designed this system to be, again, the brain of your business. So we anticipate, as an example, like, a marketing team, if you divide a marketing team up, there are sub-goals under a marketing team, and then we envision the sales team also has sub-goals. So the idea behind

1921
05:34:56.850 –> 05:35:07.209
Tooba Durraze: Kind of aligning to, like, a universal data language means the neural net is capable of segmenting and understanding these are marketing-related goals, these are sales-related goals.

1922
05:35:07.240 –> 05:35:16.979
Tooba Durraze: But the cool thing about a neural net or a platform of this sort is sometimes you set a marketing goal, and you might meet your marketing goal, but your sales team doesn’t meet their goal.

1923
05:35:16.980 –> 05:35:29.950
Tooba Durraze: Over here, you set a goal at the level of how much, do you want to grow your business, as an example, overall. So it’s constantly checking each of those goals against each other to make sure you’re not doing something that’s at a detriment

1924
05:35:29.950 –> 05:35:44.629
Tooba Durraze: to another goal that your company or your business has. So the insights and recommendations will tailor to allow you to meet the lowest common denominator, the breadth of all the goals that you have in the system, instead of just optimizing for, like, a singular goal.

1925
05:35:45.770 –> 05:35:50.400
Julia Nimchinski: And people are asking, what data privacy protection do you offer?

1926
05:35:50.900 –> 05:35:56.669
Tooba Durraze: Yeah, we actually… I come from a data science background, so we…

1927
05:35:56.670 –> 05:36:13.850
Tooba Durraze: we were small as a startup, but from the beginning, we started with our, obviously our SOC 2 GDPR compliance, our ISO compliance, and then on top of that, we have a couple of, vendors that we work alongside, like Private AI as an example, where if you wanted to anonymize your data, because a lot of times.

1928
05:36:13.850 –> 05:36:24.759
Tooba Durraze: Customers will, before they sign on as a customer, will send us, like, a slice of their data to understand what kinds of outputs they could expect as a way to prove that this is valuable to them.

1929
05:36:24.760 –> 05:36:45.740
Tooba Durraze: So we employ other services to help us kind of anonymize the data. Now, you can also try this platform completely with synthetic data. So if you were in favor of kind of trying this, or you wanted to see, like, how this would perform in scenarios that are similar to, like, what you face in your business, reach out to us, and we’re happy to create those simulated environments for you as well.

1930
05:36:47.190 –> 05:36:53.789
Julia Nimchinski: That’s amazing, and what’s the best next step to, should I just reach out to you on LinkedIn?

1931
05:36:53.790 –> 05:36:59.689
Tooba Durraze: Yeah, I’m… reach out to me on LinkedIn, and again, it doesn’t have to be necessarily even about

1932
05:36:59.690 –> 05:37:23.839
Tooba Durraze: what we do at Amoeba, but if you want to know more about neural symbolic AI, or you’re curious to play around with agents, I’m always happy to geek out. And then as a special offer, obviously, to the participants of this conference, we are offering a free analysis. I know a lot of board meetings are coming up, I know a lot of end-of-year reviews are coming up, so we’re offering free analyses via Amoeba to you, for a short period of time, so reach out to me, and I’m happy to kind of facilitate that.

1933
05:37:24.610 –> 05:37:27.099
Julia Nimchinski: Fantastic session. Thank you so much, and…

1934
05:37:28.010 –> 05:37:44.799
Julia Nimchinski: That wraps up day one of the GenTech Distribution Summit. Thank you for all the incredible insights, speakers, panelists, demo leaders, and everybody who’s watching, and we’ll be back tomorrow at 8.30 a.m. Pacific.

1935
05:37:45.220 –> 05:37:53.879
Julia Nimchinski: Lots of exciting sessions tomorrow, many Medina, Mark Roberge, two CRO panels, so join us.

1936
05:37:53.920 –> 05:38:08.480
Julia Nimchinski: And if you’d like to book a session, any one-on-one type of consultation, coaching, mentorship, that’s all available with the majority of our speakers here, so you can just check our marketplace, and yeah.

1937
05:38:08.610 –> 05:38:11.670
Julia Nimchinski: hard-scale.exchange, and see you tomorrow!