When Meetings Become Agentic Systems — Fireside Chat with Sam Liang & Blake Williams

Julia Nimchinski:
Welcome to the show, Sam Liang, CEO and founder of Otter and Black Williams, AI and PE operating partner. Sam, Blake, how are you doing?
Blake Williams:
Hey, what’s up, Julie? How have you been?
Julia Nimchinski:
Good, excited for this one. I believe Sam is here. But, yeah, what’s the latest and greatest with you, Blake?
Blake Williams:
Not much. Demandi was, acquired a little bit ago, and I’ve taken, like, the past four and a half months off, and just kind of tied up with some, mid-market funds, and working kind of attached to portcos and helping them deploy AI, and Really quantify and build frameworks around how they should do that.
Julia Nimchinski:
Very cool.
Blake Williams:
Yeah. Sam…
Julia Nimchinski:
Simon Blank, before I get started, let’s hear your production about AI and GTM for 2026. Sam. Can you hear us, Sam?
Sam Liang, Otter.ai:
Oh, yeah, yeah, sorry, my screen looks a little strange. Okay, so we are seeing this.
Julia Nimchinski:
Announcement.
Sam Liang, Otter.ai:
An awesome, okay, awesome.
Julia Nimchinski:
Yeah, welcome to the show, super stoked to have you here. What’s your top GTM AI prediction for next year?
Sam Liang, Otter.ai:
Well, we’re transforming order from a AI meeting note-taker, to a corporate meeting-centric knowledge base with Agentico workflows. So, we’re focusing on building a enterprise-wide central repository of meeting content, and then use Agentic solutions to optimize people’s daily workflows.
Julia Nimchinski:
Love it. We use our product, in a community, myself, for, I don’t know, like, more than 5 years so far, so… Ichfad’s here. Blake, the stage is yours. Take it away.
Blake Williams:
Okay, very good. Good to see you again, Sam. It’s only been a few minutes. But you’re in this process, like you just said, of transforming Otter, you know, from transcriptions, from just a transcription tool, been working on it for about 7 years, and you’re moving it to this full enterprise knowledge base platform. There’s a couple things that I want to get into. You mentioned that there’s a billion dollars in annual ROI, and I want to unpack that. You’re delivering that for the customers. It’s a major proof point. That… that really should just accelerate how you guys grow and expand. So, Sam, you’re the CEO of Honor.ai. If you want to tell us just a little bit about yourself, so everybody listening knows, and then we’ll jump right into the, the three pillars.
Sam Liang, Otter.ai:
Yeah, of course. And Sam, again, We, with my co-founder Yunfu, we started Otter, back in 2016, before AI became a household buzzword. We started, with the vision that, hey, we want to… capture all the conversations in the world, digitize everything, make everything searchable and shareable. Then we basically created the AI meeting assistant category, rapidly growing, we have processed billions of meetings, the, now, you know, there are a bunch of meeting note-taker apps on the market now. However, we are advancing and transforming Otter from a… just a meeting note-taker to a enterprise meeting-centric knowledge base with Agentica workflows. So, we see, the tremendous value, tremendous knowledge varied. in meeting content, which, are mostly lost in most enterprises, or even if they capture some of that information, it’s usually siloed. There’s so… We see just tremendous opportunities, tremendous, value that we can unlock.
Blake Williams:
Okay. So what was the moment that you… like, where were you, and what was the moment that you realized that… Otter has to become something more than just transcription. And, you know, how did it transform into enterprise knowledge base? There’s companies like Recall.ai that let almost anybody With… with 500 bucks, build a meeting agent. Where were you when you realized that the ground was kind of shifting and that you needed to change direction?
Sam Liang, Otter.ai:
I would… I wouldn’t say we suddenly, realized this and decide to, change direction. Actually, I wouldn’t say we’re changing direction at all, because this was the, vision, from day one. We just, you know, understand it will take some time for enterprises to adopt this new concept, adopt this new mindset. So when we started back in 2016, we focused on building the, a proprietary voice AI technologies to transcribe, to recognize, speakers. By the way, we are… I believe we are almost, probably the only one who can do real-time speaker dialization and real-time speaker identification, especially when you have two or multiple speakers in the same Zoom room. Zoom couldn’t do it, Microsoft couldn’t do it, and we have our patent technology that can separate speakers and label everyone. independently. So we built that technology ourself. We realized that for any new Disruptive, product like this is hard to sell to enterprises top-down, because people are worried about, privacy, worried about, security. And it just requires such a big, culture change. So we decided to go bottom-up first. So this is very similar to, a number of, revolutionary products, Figma, Dropbox, Slack, they all went bottom-up first. So, we rely on that for several years, and we have over, 30 million users now. And every month we are getting millions of new users. organically, mostly organically, people. There’s tons of words of mouths. And then we look at our database, and we actually see, like, tremendous enterprise people already using it on their own, or any, I wouldn’t name names, but, any major, companies will look at their, the domain name. you know, XYZ.com, right? There are usually hundreds, if not thousands of users who are already using in that enterprise. And then many of them actually start to contact us and say, hey, we found a lot of users using Otter. They ask us about security, they ask about, like, potential enterprise, contracts. So, that, that’s how we… Realize that, okay, the time is right to focus on building enterprise knowledge base now.

Blake Williams:
Definitely is. It definitely is, and thank you for that. And for everybody listening, we’re gonna get into… AI agentic workflows, and we’re going to talk about approachable AI transformation, so just keep sitting… sitting tight. We’re going to get there to some tactical value that you could implement today. But before we do. You know, Sam, you said, you guys are the only ones that can do real-time speaker identification, and with 30 million users, my next question is around AI agentic workflows. Most people are toying around with kicking off their own personalized workflows to you know, capture whatever productivity they’re trying to, but meetings are this artifact of corporate culture that we think kind of slow us down, and… the vision that I’m… I’m getting when I hear what you say is that this is… this is a new opportunity for shared context to develop, and AI-agentic workflows to kick off from there. What types of, How do you see that playing out in different organizations? from there.
Sam Liang, Otter.ai:
Yeah, if we think about, enterprise workflows, almost all of this revolve around meetings. revolves around meeting. Meeting is almost the, you know, the center of every workflow. you start with planning, like the corporate planning, team planning, departmental planning. You use meetings to discuss your objectives. plans. You break down a big plan into smaller plans, you know, assign action items to each team, assign action projects or, action items to each team member. A two-leg otter actually sit in all these meetings. It actually hears everything he wants to do. Do you want to increase revenue, right? Do you need to launch a new product? Do you need to fix, some, product issues? Or, it also hears, what the customers need. It hears, you know, in sales meetings, if you use a tool like Otter, it hears everything what the customers talk about, what new features they need, what pain points they have, what use cases they need to, to address. So, a tool like this, and when you build a knowledge base, all this knowledge base has actually, understand what you need to accomplish. And understand who needs to do what. Then, this gave our AI the opportunity to figure out, okay, what agents can help you do what?
Blake Williams:
We can start with some…
Sam Liang, Otter.ai:
simple actions, like, you know, after a customer call, right? We know you need to put that data into CRM. then the agent can go ahead and just extract the insights, extract data, extract the user story, put that into CRM automatically. The sales rep don’t need to do anything. Right? After an interview, a recruiter interview, the data is automatically put into the applicant tracking system. Then we also need to know that, okay, after a call, you probably need to write a follow-up email. Okay, then we just use agent to write an email draft for you based on the conversation. It’s personalized, it’s customized based on a specific meeting. You just review it, you know, within a minute, you can hit the send button. for, right, after a project management meeting, you know, okay, we know there’s a certain bug, needs to be worked on, or it’s a new bug, then we can go ahead and maybe file the bug in Jira. Right? So these are some simple tasks. We start with that, but gradually it will be more and more sophisticated. So, a tool like Otter is sit in meeting. It can almost, can, already do something in real time while you talk. it can, you know, either start to drop a message, start to drop a document for you, put data into whatever database you need, CRM, applicant tracking system, or other data storage system.
Blake Williams:
Yeah.
Sam Liang, Otter.ai:
So, this is… this is the start of a genteic revolution. You know, we can start with a simple task, but gradually, you know, order can become an intelligent AI teammate. It can do more and more.
Blake Williams:
Yeah, that’s… this is, I think it’s going to be truly transformative, especially as most companies… they say 99… 95% of, AI deployments fail. And when you think about the simplicity that Otter is bringing, for just a CFO to think about which AI projects am I going to underwrite in this business. Moving forward, or, you know, quarter by quarter. having access to that kind of one context, being able to identify friction or acceleration is, truly powerful. When we think about tying in that… that ROI. If you’re generating right now, I think it’s a billion dollars of ROI, and how do you go about tactically calculating what that looks like? On the surface, it’s like plain as day, right? We can subjectively see that we don’t have to do a couple different follow-up tasks, etc. Is it that… is it just that simple?
Sam Liang, Otter.ai:
Ugh, The reason we come up with that $1 billion ROI is actually based on customers’ ROI calculation. These are real enterprise customers. They calculate that, for every 20 seats, they purchased from Otter. They see it generated one full-time employee’s value. One full-time employee’s value. And, and they’re based on their contract, it’s actually generate, you know, if they pay us, let’s say $10,000 or $100,000, they… they estimate that it generates 10X, at least 10X. Actually, they’re, the, some say it’s 12, it’s 13. But we just rounded to 10x. Because we actually are, generating over a $100 million in revenue today, so we estimate that 10X of $100 million is a $1 billion, over $1 billion. return on investment. So that our customers invested $100 million in our product, we generated, you know, 10x, value for them. I think this is still at the low end today. The ROI multiple can only grow. Yes. Because the AI will continue to become more and more powerful, it, you know, the identical workflow can accomplish more and more sophisticated tasks. So, our vision is, in a few years. You will rarely use keyboard to do anything. This is my keyboard.
Blake Williams:
of…
Sam Liang, Otter.ai:
Because people actually don’t like writing. And, there are already data that claims that 50% of new written document on the internet is generated by AI. So that percentage will only grow. In a few years, probably 90, 95% of the written document you see on the internet will be generated by AI. However, people won’t stop talking. Talking is, you know, generally the original, authentic data right out of your brain.
Blake Williams:
Yes.
Sam Liang, Otter.ai:
So this is why talking and meeting become even more dominant in terms of knowledge generation. And then almost all written documents, your Google Doc, your Confluence, your Notion Doc, most of them will be generated by AI pretty soon. However, right, people will continue to talk, they continue to brainstorm with voice, they just continue to meet with colleagues. So, what we envision is that pretty soon, you you just… park, then our AI will do all the busy work for you, will make your life easier. So, voice, it will be the ultimate, user interface.

Blake Williams:
Yes. Yes. Now, you know, as you said that, I feel like, with the port codes that I’ve been talking with. The one thing I notice slows 100% of them down, it seems pretty pervasive, is the culture and the education level around AI workflows is what slows the adoption and really the materialization of that AI ROI. That we think we’re going after. It’ll… I… you know, Julia asked for a prediction. I predicted in 2026 that that… that education gap is going to close very, very fast, and many, many more, AI companies are going to achieve the… and expand, right, in your case, the ROI they’re able to capture. But on that note, let’s… let’s get tactical and start talking about, you know, you addressed the planning. I feel like FP&A is going to have a great opportunity to tap into this knowledge base, and expand what they’re doing. But sales and marketing, those are, you know, this is a GTM show. Let’s start with, you touched on it briefly, what’s the seller? Going to be able to do, or not have to do anymore, so that they could focus on keeping sales about I’m just about selling.
Sam Liang, Otter.ai:
Yeah, for sales and marketing, we see that they actually adopt AI much faster than some other departments. Part of the reason is that for a sales department, the pressure is high, right? Hey, they have this pressure to talk to tons of customers, generate, you know, very aggressive revenue goals. So, they adopt AI pretty fast. Our own sales team, although it’s still small, use our own tool as well. You know, we also sell the tool to other, sales departments. So, the way they use it is, they, you know, they have, our AI note-taker join the meeting. It… right now, it sits there silently, right? It takes… transcribes, summarizes in real time, allows you to chat with the meeting content in real time. And as I mentioned earlier, after every call, the sales rep actually doesn’t need to even open Salesforce, because within a minute, the data is already pushed automatically into Salesforce or HubSpot. And then the, the follow-up email is automatically generated. The sales manager, this is also very, valuable for sales managers. Several ways sales managers can use this. One is that for a really critical calls, even if the sales manager cannot join the call, the sales manager can actually have a dashboard and monitor multiple live sales calls in real time. Right? You could monitor 4 or 5 important sales calls, you know, like maybe a million dollar deal. And then you can give live Tips, live, advice to the sales tribe in real time. In addition, if the sales manager is not there, we have a live coach feature. Live sales coach. Actually, the live sales coach is trained on thousands or ten thousands of sales calls, and trained using your own company’s playbook, you know, how do you pitch your product, how do you handle customer objections? So the sales coach actually sits there silently, doesn’t say anything, but whenever it hears something tough from the customer, it can pop up the answer, for the sales rep to read back to the customer. It’s almost cheating in the exam. Right? And, and, pretty soon we’ll add this feature and also give, you know, every, every sales call, the sales manager can, create a plan first. Okay, these are the five, you know, topics you need to touch on, or five critical questions. So it will remind the sales rep to cover everyone, and if you forget that it it will pop up a reminder and say, hey, you haven’t asked this question about this particular issue. The sales rep can, you know, hey, oh, okay, then it will ask the customer that question. Also, on the company level. We allow you to create, you know, channels, other, meeting channels. This is very similar to Slack channels. You can organize the channels based on clients, you know, you can have one channel for every client, or you can, in our company, we also create a giant, channel for all the sales calls. We have thousands of sales calls there. And the… We allow that… we make that public inside our team, so that the product team, the product managers can review any sales calls. And query any sales calls, or, you know, ask, hey, what are the common, customer feature requests in the last week? It can give you a really nice, summary of the, the feature request, or you can ask any, any question. Allows, so it allows the, it promotes, and make it, cross-functional collaboration much easier. In the past, right, the sales team needs to write a memo, maybe after 2 or 3 months, and share that with the product team, and say, hey, these are what the customers are asking for. But right now, everything can be real-time. You know, the key user feedback, the PMs, the engineers, the customer success team can be notified, you know, almost instantly. So, really accelerate how people operate and accelerate all the workflows.
Blake Williams:
So that’s available right now. Like, you guys are doing that right now.
Sam Liang, Otter.ai:
Yeah, it’s already available. It’s ready.
Blake Williams:
So, when I hear that, I just see the breakdowns and the gaps between silos almost disappear instantly, as you start to pull in all of that between just the collaboration between sales, marketing, product, the entire company. Do you feel like this is… I mean, I know this is not your intent, and you may not even care about CRMs. at the end of the day, is this the nail in the coffin for any CRM that stays a database that just has stages? You know what I mean? It’s almost… you could almost do better and be more flexible with a Postgres database that Claude spins up for you.
Sam Liang, Otter.ai:
I think we’ll, order today actually work with CRM, together, because there’s a lot of, legacy data already stored in there. I don’t see we replace that instantly. However, for new data, most of the data came in through, voice. is unstructured. It takes effort to put them into a structured database. As I mentioned, right, that process can already be, automated. But even… and other also provide a, interface to query the Salesforce or other CRM. By the way, I found the Salesforce UI so hard to use, you know, I came from engineering background, it’s really hard for me to learn to use Salesforce. Now I just actually use objects’ own UI. I can chat against Salesforce, I can chat against other databases without even opening Salesforce. So, but the longer term, I think for, you know, all the new data, I, I don’t even think Salesforce is necessary. Because AI, and with a large language model, can organize the data even better, and actually avoid the limitation of that database’s schema.
Blake Williams:
Yeah, so if, there’s people listening and watching, today that… There’s two use cases. One, I’m a user of Otter. How do I bring this to my company and say, hey, look at what they’ve done, look at what’s possible today, we should push this into a pilot. And then, two, if I’m not a current user, how do I… how do I make this, how do I kick off a trial, or something like that?
Sam Liang, Otter.ai:
Yeah, you know, we have been going bottom-up, so everyone can actually sign up for free and start using from today. But if you are a team lead, if you’re VP, you know, we’re happy to talk to you, you know, give you a demo, and very likely, there are already some users in your team who are using Otter today. You, you may or may not even know. We actually see people, like, Using other, even with their, personal credit card. because they found that it already optimized their personal workflow, but now, hey, we are, you know, evolving into enterprise workflow, so it does require the buy-in from, the, the VPs, the, the CIOs, the… the CTOs. So, you know, it’s just tremendous knowledge and intelligence that, either lost, not captured, or it’s, even if it’s captured, it’s, siloed, you know. Our goal is to break down that information silo, really. promote info… free information flow, free information sharing. Of course, you can configure the permission system access a scope for each, meeting, so that you can still control, the data retention, control security, so we built that, SOC 2 Type 2 compliance, HIPAA compliance, so, it is ready for enterprise adoption now.
Blake Williams:
Yeah, I think, you know, based on everything that we’ve talked about, you guys might have one of the most underwritable AI opportunities for any business today, if you’re thinking about investing in pushing AI across the company. I appreciate the time, Sam. We’ve got a minute left. you know, I said my prediction for 2025, or sorry, 2026. Do you have a prediction or a partnership that you’re excited about, coming up in 2026 that, You wanna break the news on?
Sam Liang, Otter.ai:
Yeah, I already alluded to that, my prediction that, people, will continue to talk, and voice continues to become the dominant, communication channel and content generator. People will write less and less.
Blake Williams:
Yes.
Sam Liang, Otter.ai:
Eventually, I don’t know how soon, you know, you will rarely use your keyboard, so you can accomplish almost everything using voice.
Blake Williams:
Yeah, that’s an amazing day. I’m ready for it. I’m ready for it. Love to write, but love to think and talk much better. All right. Thanks, Sam.
Sam Liang, Otter.ai:
Thank you, Blake. Great to chat with you.
Julia Nimchinski:
Thank you so much, Sam. Thank you, Blake. What’s the best way to support you? One more second. Sam, how about yourself? Where should our people go? Otter.ai?
Sam Liang, Otter.ai:
Yeah, just go to auditor.ai, either, you know, sign up for free, or, you know, contact us, and we’ll, be happy to give you a demo, just share, user stories, users’ ROI. use cases.
Julia Nimchinski:
Amazing. Blake?
Blake Williams:
Yeah, the best way to support me is, just add me on LinkedIn, and if you are thinking about building frameworks around how to adopt AI across the organization and underwrite them. So that you keep your job and don’t lose it, on running pilots, just give me a shout. Happy to chat.
Julia Nimchinski:
Again, thanks.
Blake Williams:
Alright, now. Bye-bye.