---
title: "Why Slack Is the Future of B2B Support"
episode: 40
podcast: "The LeanScale Podcast"
publisher: "LeanScale"
guest: "Tony Tom"
guest_title: "Founder & CEO of Orca"
date_published: 2025-10-29
date_modified: 2026-07-22
duration: 00:34:40
word_count: 5513
topics: ["ai-in-gtm", "gtm-strategy", "revenue-operations", "brand-positioning"]
canonical_url: https://leanscale-knowledge-hub.netlify.app/podcast/tony-tom-slack-b2b-support/
source: "LeanScale Podcast Knowledge Hub — https://leanscale-knowledge-hub.netlify.app"
license: "Free to quote and cite with attribution to The LeanScale Podcast."
---

# Why Slack Is the Future of B2B Support — Full Transcript

> Episode 40 of The LeanScale Podcast, with Tony Tom.
> Published October 29, 2025 · 00:34:40 · 5,513 words.
> Machine-transcribed and **not diarized** — speaker attribution is inferred, so verify
> attribution against the audio before quoting a specific person.
> Structured breakdown: https://leanscale-knowledge-hub.netlify.app/podcast/tony-tom-slack-b2b-support/

## 00:00 — Meet Tony Tom and the origin of Orca

**[0:00]** Today we got Tony Tom, founder, CEO of Orca. Tony, so excited to have you here. Really, really pumped to dive in to what you're doing and bringing an AI agentic experience to supporting B2B customers in Slack, where 99.99% of us are living all day long. So Tony, we'd love to hear how you got the inspiration to start Orca and gave you the courage to get it started. Yeah, thanks. I'm super glad to be here. Great to chat with you folks, being a fan of the podcast for a long time. To tell the founder story, it's always a long story for every founder, but long story short, I've been into startups for a very long time,

**[0:50]** especially in the early days when startups was not cool, being in product management, specifically for B2B companies for the last decade or so. And as we were building product for small companies, and I've been in a few big companies as well, like Zoom info, what I've noticed was the customer data that, or a better way to say this is probably voice of the customer, is not getting translated back into the product teams as much as we wanted it to be. We had to hunt for that. We had to go and work for that. That was the way that it was set up. Not necessarily

**[1:34]** because the organization wanted it that way, but it's how it was. That's the basic answer for that. So we started there with our first product. When we started the company, we thought about how do we bridge this gap between the go-to-market segment of customers, prospects, everybody's talking about what they want, et cetera, et cetera. And then we would bridge this gap to the research and development team, the team that's actually building the product. That's where we started with the first one. But as we went through the journey of building out that product, talking to customers and giving

## 02:08 — Why B2B support data is fragmented across channels

**[2:08]** it to customers to try it out, what we realized was even the customer-facing teams itself themselves don't have the ability to understand what exactly is happening with the customers. The scale of customer support, success, all those post-sales functions is too big to manage even. And especially when B2B is starting to scale up or go to the maturity of what we see today, probably scaled it up super fast after the 2020 era. What we're seeing today is an influx of customers going through Slack, email, Discord, five different channels. And then all the feedback is getting

**[2:51]** distributed, all the data is distributed across multiple channels. Nobody can manage that. Even the communication support teams cannot manage that. So instead of solving the larger problem initially, we thought, why don't we start solving the problem for the GTM teams itself or the post-sales teams itself? That's how we landed on what we have today, an AI-first customer support platform for B2B companies specifically built with account as the centerpiece of it. That's an interesting part that we figured is built because most of the ticketing platforms are built for horizontal use cases. If you take anything out there, it's more ticket is the center

## 03:38 — Account-first vs. ticket-first

**[3:38]** of that platform. But we see pre-sales or anything that until sales closure is always account, account, account. You have to keep account in your hands to cater them, pamper them throughout. But when we get into post-sales, that gets stopped. Account is nowhere to be seen. So we want to bring that back in as a positioning or as a way to do things in B2B. So that's probably a shorter version of how we landed here, the nuances still. Yeah, makes sense. And Tony on that, so current market when you started to that point, they were focused a lot on each individual ticket. When you're saying that, that's the centerpiece of their

**[4:23]** product. They were looking at a ticketing page and making sure the ticket was integrated with the right systems and all of that. You're saying that you're going account first so that you can pull in all these channels to one place. And again, kind of back to the data issue, is that the gap that you think AI is going to help with? And yeah, what advantage do you think I guess coming in with the account first approach will give you when it comes to AI? Yeah, it's when people start talking about support by itself, the default thought process goes into like, how do I solve a question

## 05:02 — Context over answers: what AI should actually do

**[5:02]** that's coming from a customer? How do I answer that question? So because it's all textual based and AI has the ability to look through a ton of data and then come up with an answer for a question. That's what you see in chat GPT. So by default, every AI company starts there, how to answer this question. But when it comes to B2B, what we're seeing is like, it's very contextual. You as a customer support or a customer success, anybody who's touching the customer post-sale has a ton of context or needs a ton of context. So AI can come in and look at all this data in the backend,

**[5:39]** maybe pre-sales data, call recordings, and all the other data points that are there in your CRM systems. Even the question answers that your humans have already delivered to the customers and got a thumbs up or got a five start from the customer, all that can be used to treat the account better, not just treat the ticket better. So what I mean by that is, is this account happy with all the tickets across different platforms. It could be happy with just having a good happy stakeholder. We see a lot of that. People buy stuff just because they like what they see,

**[6:19]** but the users might not love it. So how do you analyze that sentiment and bring it back into the stakeholders visibility? A champion might not be happy, but the users could always be happy. That's another scenario, specifically looking at that. A champion could change jobs. The new champion coming in might not be having enough context about how this buying process happened, where did we start, how is this product being used, et cetera. A lot of other scenarios like how is the customer actually aware of 80% of all the things that you have shipped so far. We've

**[6:54]** seen 80/20 rule happen a lot in B2B. 20% of the features is used by mostly 80% of the customers. So how do you make sure everyone is aware? All of this happens through various touch points, mostly through QBRs and quarterly reviews that happen from a CSM standpoint these days. But once every three months is not good enough for your customer to be like, hey, four times in a year, you're paying $20,000, $50,000 for a product. You talk to them four times a year, you have 30 minutes. How do you convey all of this? How do you get all of that within a limited 30 minute time

**[7:32]** period? So it's not happening as it should be. This is where AI comes in and looks at the scale across different platforms and help you communicate. Even your support teams get empowered to say, hey, we are seeing this pattern. We are going to send this to the product team and give them this pattern. We'll keep you posted on the update. That's a better way of supporting an account rather than just saying, hey, here's the answer for a question. We do not support this today. That's how we look at it. No, that's great. I think distributing the information to the right

## 08:12 — Product tour: Slack tickets and message grouping

**[8:12]** people at the right time, organizing the right information, especially in the B2B context, that account is the persona you're really trying to serve. So it's difficult to do that with the current solutions. I'm really excited to see how you pull this all together in Orca and show us how it's different than typical ticketing systems. Yeah, definitely. So what you're seeing on the screen is the Orca platform. I'll walk you through what it means. On the left side, what you're seeing is basically the account. So we'll get to the account. That's our centerpiece. This is

**[8:51]** the ticketing system. So we already have a ticketing built into the system where you can see and manage tickets within the system as well. We also have a piece where documentation is automated for most of the cases. To communicate back to customers broadcast and general settings, we also have the AI agent where if you want to dig deeper into specific scenarios, specific patterns that you want to identify, we can enable that too. So I'll start with Slack. What we are starting to see is a lot of B2B companies are starting to see or starting to bring customers onto a Slack Connect

**[9:34]** channel. So what you're seeing on the screen, I'm just simulating a customer here. So I'm just going to send message as a customer from here to my customer Connect channel with Orca. So it's just a problem that I'm seeing and hoping to see if this bug is removed. So what happens is since Orca is already present in this Slack channel, it keeps looking at all the messages that is coming in and keeps track of that as tickets. So if we go back to Orca, you would see this ticket is coming in from Slack here. And you can start responding to the ticket from here itself. You have basic

**[10:17]** description that's AI generated, what is happening exactly in the ticket. It's auto assigned to Jiren, one of our teammates, and the status is like all of the ticketing is managed there. Now, one pattern that we usually see is customers don't always go and respond in threads when they start looking at Slack. So they could be coming and saying, hey, somebody else could be coming in and saying, app updates could have a problem. They'll just send it as a message itself. This is one of the smallest applications that we have seen as AI is we are looking through a lot of messages and we

**[10:58]** are figuring out what is the context, what is the actual context from this customer started talking about this problem. We keep grouping that into one single place where we can keep track of things that are related to this problem where you would go and say, hey, we have similar messages coming in, AI would auto categorize them, auto group them, and then start looking at it. And you can start answering here, looking into it. This is a basic way to start a conversation and then you can send this across to your team and start looking into it. This is where our second part of the system

## 11:42 — The triage workflow

**[11:42]** comes in where you start sending this across to your team. It's a little bit difficult to convey what the customer is going through, especially when this conversation goes beyond your control, right? Customer keeps talking about it. These are the problems, et cetera, et cetera. It keeps coming in a lot. That's where we have a workflow within Slack itself where you can automate the triaging. So what you're seeing on screen right now is the Orca platform. So you might have seen Jaren is assigned for that particular ticket, ticket 232, but what Orca does is it auto triages the ticket. It brings in the account, which account is about

**[12:27]** what is a priority, what is a type of it, and with an original message context into a single Slack triage channel where your engineering team, your product team can come and see the whole context without you sharing anything, without you being able to take all the context that you want. You want to probably give everything to it. Whatever that happens in the customer conversations is always triaged in a Slack channel as well. This is a workflow that we have in our system. So your product team don't miss out anything. Your engineering team don't miss out anything, or even

**[13:03]** your leadership don't miss out anything. But that's where we start looking at B2B as a customer support. Let me pause there. It's a complex workflow to start with. I'll take a step back as you go through this and then explain how this is set up as well. I really appreciate the little things and intentionality, and it shows when you've actually been in the field solving these, just grouping another DM into the same ticket and having the context of knowing that it's grouped together. Because people communicate all different ways, and it always bothers me when people are

## 13:42 — Team threads and replying to customers via emoji

**[13:42]** slacking outside of threads. But sometimes you don't know if it should be a thread yet. So I think that is a huge, huge advantage. I know it's a small thing, but those little things go a long way. Yeah. Another thing that we facilitate from here itself. So this is our Slack, and we want our team to be able to discuss things here as well. So whenever a ticket comes in, ticket 232, this becomes a discussion thread with my team. So immediately I can tag anyone in my team, and how are we doing here? Any updates? This thread becomes my team's Bible on how to manage this

**[14:24]** ticket. So this is like becoming a longer thread. I can talk to my team. I can just simply tag people. None of this goes back to the customer. But let's say I'm just going to simulate myself here. But let's say my team member has a question. Hey, I have something to ask to my customer. Now, your team member has to be in the support platform or to be in the loop, regardless of whether it's email, Slack, whatever. You've got to be in a touch point with your customer to interface. So whenever my team wants to communicate with the customer, what they can do is they can

**[14:58]** come here. They can use a pre-determined emoji. This is again a setting in our platform. And Kira, do you mind sharing a quick video about this? So this particular message that you've seen, pre-labeled, Google can send this message or anybody in my team can send this message from here. And it goes back to Kira. So it goes back to the original thread. And it goes back as a question. And that enables my team to easily communicate with my customers. And also, whenever that happens, the major problem that we hear from customers is like, oh, do I know what my team is talking about?

**[15:42]** Do I know what is exactly that they're talking about? Because the ticket's volume can be high. You don't have to keep track of things in Slack. It always comes back here in the system. So you don't have to worry about what your team is talking about. You will always have that track in the system. So you never worry about what's missing there or what's happening there. And whenever you want to go back to the original message, this is where the source comes in. The source clicking on the original message directly takes you to the customer's channel and exactly where the customer is talking about this problem. So it is all linked together so

## 16:21 — Multi-channel: email, Discord, and suggested responses

**[16:21]** that your Slack is never getting missed. Now at this point, all our customers can mute their customer Slack channels and just listen to either one Slack channel or just look at Orca to be in the whole spectrum of listening to customers and understanding this. That's just the Slack part of things where we start. This is one of the focus points for our product as well. Because we see a lot of customers bringing their customers onto Slack. Slack has become a high-touch way of doing customer support for B2B specifically. That's where we started working on this. But as we scaled up,

**[17:04]** as we started seeing more and more patterns, we saw, hey, email is another way of looking at customers. So this is coming in from emails. And we started seeing Discord as a channel as well because you see a lot of tickets coming in from community, a lot of tickets coming in from people who really want to understand. So I'm just going to jump into Discord community here quickly. So this ticket 231 came in from one of the communities that we manage. And you can see it's fairly technical. It has a code snippet in it. It has some of the errors in it, et cetera, et cetera.

**[17:41]** And you also probably are seeing suggested response here. So what we do here, especially for technical tickets, is whenever we get a technical ticket, your support team might not be empowered to kind of have the context. And what we also say is it needs a lot of context on the implementation B2B end. Dig deeper into all of this. As we could dig deeper into this, just as I shown you, the technical person or non-technical support guy can just simply use this AI response. Hey, here is the documentation. And here is what I found from documentation. Can you check this as we go through this internally? This is a simple message that my technical,

**[18:33]** non-technical support person can send across and assign this to the CTO or somebody who's in charge of the technical stuff and enable them to look through this very fast. And meanwhile, the customer would be happy that they've gotten a response. That is where our AI comes in. We believe in terms of enabling AI to suggest answers rather than just auto answering, especially in the B2B, very technical, very contextual questions. That's where our AI suggestions come in as well. So this is the starting part of ticket management and we still have emails, all of that. But the core part, as I mentioned, in the starting point

## 19:21 — The account timeline and sentiment view

**[19:21]** is account base, right? All of this is tickets, regular, some things that you see in any platform out there. But the difference comes in when you start looking at the account section of things. So this is where we come in as a system and we track everything. So the timeline that you're seeing here tracks everything. So the Slack messages that you have, the tickets that are getting created, the responses, all of this. And this enables you to track the engagement in a much deeper, much horizontal level, the sentiment of the customer on a more horizontal level. And whenever you want to jump into specifics, you always have the sources here itself at an

**[20:08]** account level. So you never have to worry about whether I'm keeping track of things beyond tickets. It's always account level. Everything that your account is communicating with, you can keep track of that, go through the history of the account, understand where we lost or where we won. If I want to look at March specifically, I know the communication is like in the early days and probably towards the middle, I haven't touched base with them towards the end of the quarter, which probably is a mistake. Then if somebody comes and asks me, "Hey, what is happening with

**[20:42]** this particular customer?" I can always say, "Oh, this communication is a little bit broken." I can simply trigger a communication from here itself by going into their Slack channel and say, "Hey, how are things going? How is everything happening on your end?" Yeah, that was really a really cool view on the account level.

## 21:03 — How AI responses are generated (GitHub + docs)

**[21:03]** Sort of aligned with this and what you showed before on the AI responses. How are you generating the responses? What information are you looking for right now to give that type of suggestion? So there are two ways of... It's a fairly high touch setup that we do when we want to enable AI responses because what we have seen a lot of companies do is they'll just give you, "Okay, go and tag in your documentation and we'll take it from there." That's where I think the quality gets dropped a lot and the responses that you get is not really relevant for the context. So we analyze the customer initially. So if it's a technical customer, we're going to...

**[21:53]** There's some open source code base. So if they can give us access to their internal GitHub controlled way, we also understand their code bases. We go in technical. Our AI agents go in and understand the folder structure of those, how it's been getting implemented. That is the starting point for us. We also look at documentation. So that's where our documentation piece comes in. We would take in any documentation that you have today, but what we're seeing is people don't have all the complete documentations. So whenever you start using AI, you would see almost all the

## 22:30 — Loom-to-docs: closing the data loop

**[22:30]** documentations are like, "Okay, this is not great." But what we have figured is, okay, if we can enable somebody to create documentations within a few seconds with AI, some things that we do specifically just for our system. If you want to bring your documentation to our system, we have that, but that's not a mandate for us to work on top of documentation. But if you do not have a documentation and just let's say you have only a loom video, I'm just going to quickly take this demo video here, you can just simply go and enable this loom video to create a documentation for you

**[23:08]** out of the blue. So it's going to come back. I'm just going to quickly move into one of the created ones. It's going to come back after creating to me. So something that I've created here is it's this documentation was created totally from a loom video, including the screenshots, no human touch at all. It has step-by-step instructions. It has 11 screenshots. It has what do you do with a screenshot, how to configure things in the basic configuration. So this is fully created from a loom video. So we enable two things. One, go into deep technical stuff whenever you

**[23:44]** want to. That's a hand-holding stuff. When you think documentation is not sufficient enough for AI to answer, or when you see low quality, you can come here and just paste your loom video, just paste your demo videos and enable your documentation to be much better, much more comprehensive and go deeper into each of these use cases because loom is something that every company kind of like uses today or one of the video platforms where you have videos recorded everywhere, especially for customer communication. You have caught a lot of videos is what I see. Now, if you were to take all

**[24:20]** of that video and then convert into documentation, the cycle gets completed, the data becomes better. So instead of garbage in garbage out, it will become quality in quality out. That's our way of disclosing the loop and in data as well as giving the best answers from AI agents. That's a huge feature. I think even if we're thinking about us at LeanScale, we loom every single thing we build for a customer. We film a loom for it. And a lot of that is just kind of living in that video. It's not put into documentation that could train our own models as well.

**[24:59]** So I think this is huge and just baking that into the workflow to continually train the model based off of probably a lot of people are sending these looms to customers, right? So just closing that loop and then having it feed the model, that's huge. Yeah, sending loom, one of the biggest use cases of like quick video recordings like loom is to send it across to customers, right? I mean, quick recording, quick take, copy the link, send it across to customers. Now, you can just use that for improving your documents itself. You don't have to worry about the problem with documentation

**[25:35]** that we see very often is nobody owns documentation piece in a B2B lifecycle. So is the product person owning that? Is the customer person owning that? The customer success manager owning that? Nobody has ownership. If you ask like five people, all of them say, "I don't own it." So the problem is nobody wants to write this thing. And it's where we come in with our unique angle and you'd see a lot of loom videos getting converted into documentation as you go forward. Wow. That's awesome. And we talked about this a little bit, but this is sparking some ideas

## 26:14 — AI beyond answering: reporting and the in-house agent

**[26:14]** for what could be next. How are you thinking about leveraging AI and the products with already a lot of these foundational things you've already built out? Yeah. One of the things that we really believe in is whenever you start looking at this, I think this is going to keep repeating in my conversation across, people start thinking about AI just answering questions. That's not the way to like properly use AI or like, I wouldn't say properly. There's more to it. There's more to what you can do with AI. And when you look at support or customer post sales, especially in

**[26:57]** B2B, there's a lot of backend that is happening. So the backend includes understanding what the stakeholder needs. Like I need to report back to my manager, my leadership team, my customers. Everyone needs what exactly are they looking into? What exactly are the problems that we are seeing across? Probably like email as a channel, Discord as a channel, Slack as a channel, and maybe chat as a channel. Where is my documentation lacking clarity? All of those questions come up from leadership. Now what happens is the customer support people or whoever is like manning the ground

**[27:36]** gets loaded with these questions and they cannot have enough context on what to do. They don't have the time to manage this. That's one of the places where we see AI. And this is a starting point for us. We have an agent in house as well. This enables just an agent. So what we are looking at is like, I just asked a question, list all the accounts based on tickets and group them by ticket count. Now for you to get this in the past, you had to go and figure out like how to do analytics on top of the data that is reported back to Snowflake and bring that data back into Tableau and

**[28:11]** like generate a report. This table is like taking three seconds. Any question that you have on top of data, like give me, show me, I'm just going to ask this, show me all the requests from Slack messages in the last two weeks. To find this data and give this to your product team or to your leadership team and seeing, hey, this is what the pattern that I'm seeing today for a support person or a support manager used to at least take like half a day or even maybe one full day of work. This is where a lot of backend stuff gets automated. You can write workflows to automate,

**[28:50]** send this across every two weeks to my leadership, my product team, my engineering team, just so that I'm clear on my end. I've communicated properly. You have kept track of all of these things. This is where I think AI is going to shine the most, where you can have visibility across a lot of data and you can see what are the things that are happening across different accounts as AI comes back with all of that. I have a ton of history on my questions, everything that's happening there. I keep asking AI, hey, what do I look at for my priorities in what are my customers talking about?

**[29:28]** This is one major aspect where people are not talking enough, especially when you go into support. I hear this a lot because when we say, hey, we are a support platform, people are, oh, do you want to answer questions? Do you have AI to give you suggestions on answers? Even people get to suggestions, but they think support always relates to answering questions, which is where we come in and bring in a differentiation. Yeah. 100%. 100%. That's awesome. Yeah. That's awesome. One, another thing that I think it's really relevant is the workflow system. All of the system defaults that you've seen or the base layer that you've seen, it has an underlying

## 30:18 — Workflows + AI for B2B

**[30:18]** workflows engine that's working out for you. Some of the things that we hear from a lot of B2B customers is now workflows plus AI is the key in B2B is what I see. Workflows plus AI using the components to deliver what really, really matters. One of the workflows that I've shown in this demo initially is the triage workflow, where you can enable your team to respond, enable your team to be access to the customer conversations, et cetera. Another one interesting part is the routing workflow, where we have very specific routing actions based on conditions that is time zones,

**[31:01]** conditions that is tickets, et cetera, to build a detailed workflow system for routing. We are also bringing in AI specific workflow. So let's say if your customer is not happy, sentiment is really bad and you are seeing a lot of bad patterns across your customers, escalate this to a manager is a routing workflow that AI can bring in. It is again, the backend stuff where AI can help a lot. SLA being breached for a bunch of times and customer is not happy, alert me in this particular channel in a particular way that I always look at the customer when SLA is getting praised. So these kinds of workflows is what I think builds the B2B system of the

**[31:48]** ticketing management or like customer support for B2B rather than just answering questions that I think is where the future is going. I love what you showed with the workflows. I think orchestrating the SLA's assignments, creating everything automatically is really is what's going to help a B2B company scale. And I really appreciate everything you shared. I think you've shown a new way of thinking about not just support, but getting your product data, leveraging what you've already done before. So the looms, the documentation, building responses, but then using those responses and sentiment to feed valuable data to the product team. So you

**[32:36]** can do what every company is trying to do. You're trying to build the best product possible to serve your customers in the best way. And sometimes it's so hard to get that data. But I think you've built a platform that does this in a really elegant way that helps integrate the support with the product, helps take care of your customers while also feeding that data back and then automatically building a lot of the components you need so you can actually scale. Tony, I'm really, really excited for people to take a look at this. And I'm excited. I know this is just the beginning

## 33:13 — How to get started with Orca

**[33:13]** of Orca too. So I'm excited of what's coming next too. That's very kind of you. And we're just getting started. That long way to go. And if anyone wants to get started, you can just go sign up directly on Orca. And it's a really slick process. But is that the best way to get things going for anyone that's interested? Yes. One, since you've already mentioned, so what happens there is that's one workflow that we want to ship out to customers as well. We're testing it on our end that if there is somebody who's signing up at get Orca.ai, they would by default get a Slack notification to join the customer channel. You don't have to wait for

**[33:57]** us to reach out. You don't have to come back and ask us, can we have a Slack channel? By default, you get a Slack invite to the email that you're signing up with. And then you just join that channel. There'll always be me and my people there. Our team will always be there in Slack. That's one workflow that we are hoping that we can ship to customers as well. That is where we start. So just sign up. We'll see you on Slack. Perfect. Perfect. Awesome. Love it. Well, thank you, Tony, so much. Appreciate it. And can't wait to see what you guys build next. Yeah, great. Appreciate you guys having me here. Thanks.