Latest / AI for Founders with Ryan Estes / AI Just Closed 40% of Your Tickets Without You

Transcript

AI Just Closed 40% of Your Tickets Without You · AI for Founders with Ryan Estes

• 0:00
  Your IT team is drowning, maybe like mine, and every time you
• 0:03
  ping them with how do I plug this cable in, you're making it
• 0:06
  worse. So AI isn't coming for the
• 0:08
  C-Suite first, it's coming for the help desk, the ticket queue,
• 0:11
  the Jira dashboard you've been ignoring for three weeks hoping
• 0:14
  it disappears. And honestly, maybe that's
• 0:16
  terrifying, because if AI can order your laptop, reset your
• 0:19
  password, and route your HR request before a human even sees
• 0:22
  it, what happens to the people whose entire job was being human
• 0:25
  in the loop? We spent decades telling
• 0:27
  workers, learn tech, you'll always have a job.
• 0:29
  Now the tech is learning them. Every Slack message you've ever
• 0:32
  sent asking for help is training data for the thing, replacing
• 0:35
  the person who answered you. This is a transfer of power.
• 0:38
  And the scariest part? Nobody's asking who owns the
• 0:40
  automation when the automation owns the workflow.
• 0:43
  And maybe I'm being hyperbolic. We're going to need somebody,
• 0:45
  obviously. But if AI took every small task
• 0:48
  off your plate tomorrow, would you finally get to do the work
• 0:50
  you've been avoiding? Or would you realize the small
• 0:53
  tasks were the work? Welcome to AI for founders.
• 0:55
  I'm Ryan Estes. Today I'm sitting down with Tom,
• 0:58
  founder of Unthread, and here's a few things you're going to
• 1:01
  learn today. One, why Slack is secretly the
• 1:03
  most expensive black hole in your company and how to plug it
• 1:06
  up to the YC playbook for landing your first 10 customers
• 1:10
  without spending a dollar on ads.
• 1:12
  And three, the one e-mail structure that gets a 90% reply
• 1:15
  rate from founders who normally ghost everyone.
• 1:18
  Now do you want more founder secrets like this delivered
• 1:20
  straight to your inbox? Head to AI for founders.co,
• 1:23
  where we serve up weekly AI plays hotter than your last
• 1:26
  pitch deck. These are the moves that
• 1:27
  actually move the needle. Sign up for free and join
• 1:29
  thousands of founders already eating good every Tuesday
• 1:32
  morning. And if this episode hits, drop
• 1:34
  us a five star review. So the algorithm gods bless your
• 1:37
  startup with traction it deserves.
• 1:39
  Let's go. All right, Tom, welcome to the
• 1:51
  show. Thank you so much for joining
• 1:52
  me. Yeah.
• 1:53
  Thanks for having me, Ryan. Yeah, because I need to talk to
• 1:56
  you. But I I want to start it off
• 1:57
  with some stuff I was reading. Why do you hate cars?
• 2:01
  Well, where do I even start? I've been a a cyclist for a long
• 2:06
  time, and once you try to get around on just a bike in a world
• 2:11
  dominated by cars, you have no choice but to rebel against them
• 2:15
  since they're basically always out trying to kill you.
• 2:18
  So yeah, I've been working on advocating for public transit,
• 2:21
  better bike infrastructure, more pedestrianized, you know,
• 2:24
  streets and walkways. It's a tough uphill battle,
• 2:28
  though. Yeah, I love it and I, I totally
• 2:30
  get it. You know, I if I get on my my
• 2:35
  trusty steed every now and again, it's like all just doing
• 2:37
  road work in the in the country since I moved out of the city,
• 2:40
  but I was living downtown Denver for many, many years.
• 2:43
  I was a fixed gear rider and the amount of times that like like
• 2:48
  literally people would swipe me, you know, like not not
• 2:52
  acknowledging that there is sconced in steel and that I'm
• 2:55
  just made of butter. I mean, it's incredible.
• 2:57
  There is definitely, there is definitely a divide between
• 3:01
  cyclists and cars. And if I'm going to be critical
• 3:04
  about cyclists, a lot of times their aggression is actually
• 3:08
  comes out on me, you know, so like cyclists be really
• 3:10
  aggressive towards cars, flipping them off and stuff.
• 3:13
  I'm like, please do not do that because they're not coming back
• 3:15
  for you. They're coming for me when I'm,
• 3:17
  you know, going to the grocery store.
• 3:19
  Yeah, I know cyclists themselves don't don't have a great
• 3:22
  reputation. I think it's OK to flip off
• 3:25
  cars, actually. Maybe that's a controversial
• 3:27
  take, but if they're trying to kill you, I think that's a
• 3:29
  pretty fair exchange. Attempted murder and a small
• 3:33
  hand gesture, I feel like it's not so bad.
• 3:36
  But I think cyclists also could be annoying to pedestrians where
• 3:39
  they don't respect the rules and that gives them a bad name.
• 3:43
  But yeah, I think once someone tries to run you over, I think
• 3:45
  all bets are off on politeness. Yeah, Wait, you really are a
• 3:48
  cyclist. That's good to hear, mostly.
• 3:50
  Utilitarian. I just try to get around.
• 3:52
  I'm not. I'm not actually in a full
• 3:53
  spandex suit but and I respect that life too.
• 3:56
  You never go full spandex, you know?
• 3:59
  You never do. The only my concern is like, I'm
• 4:01
  not going to flip off anybody because I'm not sure where
• 4:03
  you're calling from. I'd love to know, but I'm in
• 4:05
  Colorado and I just assume everybody's armed in the teeth
• 4:08
  here, you know? Yeah, well, I'm in Brooklyn, so
• 4:12
  not as many people armed, but definitely some people with who
• 4:16
  aren't afraid to confront you, I guess is the way to put it.
• 4:20
  I've had quite a few serious confrontations.
• 4:23
  Yeah, you won't get shot. You might get stabbed.
• 4:25
  Yeah, exactly. I don't know which one's worse,
• 4:28
  but. Amazing it's it's cool and you
• 4:32
  know reading about you it seems like you created kind of this
• 4:35
  like guerrilla marketing campaign to abolish cars, which
• 4:39
  I I think is a a bold offer you know in Brooklyn you probably
• 4:42
  have the luxury that you don't really need a car you know I'm.
• 4:45
  Trying to own a car in over a decade.
• 4:48
  Amazing. I love it.
• 4:49
  It's so cool. I'll I'll link the the details
• 4:52
  for that in the show notes so people can check out what you're
• 4:54
  working on there in Brooklyn. But what I really want to talk
• 4:57
  to you about, oddly enough, is not cycling.
• 5:02
  So I could talk about this all day, but sure, yeah.
• 5:04
  We can move on. I could tell you some stories,
• 5:06
  no doubt. I'm sure you got some, but I
• 5:09
  want to talk about your career. I want to talk about you've had
• 5:12
  an incredible arc already. You know, Forbes 30 under 30
• 5:15
  guy, you know, had some successes, had some LS, you
• 5:19
  know, lived to tell the tale. And now you've got this really
• 5:21
  cool company called Unthread dot IO.
• 5:23
  So give us kind of a overview and explain to us what the tool
• 5:27
  does. Yeah, so Unthread's goal is to
• 5:30
  automate support for ITHR and other operations teams.
• 5:35
  So the way we do this is by embedding natively in tools like
• 5:38
  Slack and Microsoft Teams. So instead of filing a ticket,
• 5:42
  Unthread automatically detects when you need an issue and knows
• 5:45
  how to triage that, collect the right information from you.
• 5:48
  And then once it has that information, it can run a series
• 5:51
  of automations to complete those requests end to end.
• 5:54
  So a pretty common one is getting a new laptop from your
• 5:57
  IT team. You know the old school ways,
• 5:59
  you go to some random Jira dashboard and you file a bunch
• 6:01
  of information into a ticket and then you kind of hope that
• 6:04
  something happens someday. Through on thread, you just say
• 6:06
  I need a laptop. We already know the policies
• 6:09
  around laptop renewals, which one you're allowed to get.
• 6:11
  Once we collect that, you know, request from you, we actually
• 6:14
  create a purchase order, we ship a laptop, we get the tracking
• 6:17
  number, we send you the tracking number, all in real time.
• 6:20
  So the idea is that people don't need to be involved in this and
• 6:23
  we work with a lot of IT teams who are saying like this is the
• 6:25
  worst part of our jobs, we need a tool to do this for us.
• 6:29
  Oh yeah, it's a nightmare, right?
• 6:30
  This is the last domain of, of spreadsheets of like, hey, I
• 6:35
  need XYZ or hey, I mean, maybe another one would be like, I
• 6:40
  didn't associate my hours properly last week with the
• 6:43
  proper client. I need to fix it.
• 6:44
  Is this something that in Slack it's going to walk you through
• 6:46
  the whole thing and complete it right there?
• 6:48
  Yeah, exactly. And so for the workflows that
• 6:51
  need to be done to finish these different types of requests,
• 6:53
  people can use our like visual workflow Builder, but pretty
• 6:57
  often people are actually vibe coding their own workflow like
• 7:02
  automations to complete those steps.
• 7:04
  So, you know, everyone's process a little bit different,
• 7:06
  everyone's set of tools a little bit different.
• 7:08
  So they can just describe what they want to have happen via AI.
• 7:12
  And that can work either through these kind of visual building
• 7:14
  blocks like a xapier, or actually write code for them to,
• 7:17
  you know, make more advanced actions.
• 7:20
  Cool. Is it what's?
• 7:22
  What's the LLM that's powering it?
• 7:25
  So we let companies bring their own LOM of choice and this has
• 7:31
  been based on some customer feedback.
• 7:32
  So I mean, we use Clod and Chachi BT, you know, by default
• 7:36
  we can have a managed instance. So it really doesn't matter.
• 7:39
  But we see a lot more companies who are, they care a lot about
• 7:42
  privacy and they have, you know, custom agreements with these,
• 7:46
  you know, AI companies. And so we actually let them plug
• 7:49
  in their own, you know, API key and specific model they want to
• 7:52
  use and have that power, all the functionality as a way to
• 7:55
  guarantee that all the data is staying within their same
• 7:58
  account. That's just been an interesting
• 7:59
  trend we've seen with a lot of enterprise companies in the past
• 8:03
  maybe 12 months is they say, yeah, sure, you have your own
• 8:08
  managed ChatGPT instance. But we, our legal team worked
• 8:11
  really hard to get an agreement with our, you know, ChatGPT
• 8:15
  license. So we want to use that.
• 8:17
  Cool, so if I'm the customer, am I spending my tokens?
• 8:21
  You can't. Yeah, that's that's, yeah.
• 8:23
  And that that allows on thread to not really try to do
• 8:26
  something where we like mark up the cost of tokens.
• 8:28
  That's not something we're really interested in.
• 8:30
  We really want to just charge based on like the value that we
• 8:32
  provide, not just like marking up AI costs.
• 8:35
  Cool. So within Slack, is this like a
• 8:37
  specific channel for these needs or is there like another Slack
• 8:41
  member that's actually the agent that you would talk to and put
• 8:44
  in a personal message? It's both.
• 8:47
  And that's important because different teams use it in
• 8:50
  different ways. So even an IT team, they might
• 8:53
  have like a hashtag IT help channel where anyone can reach
• 8:56
  out with questions or requests. And so the bot will just
• 9:00
  automatically jump in there and take care of it.
• 9:02
  But maybe for someone like an HR team, these are more private
• 9:05
  requests. Like, you don't want to be
• 9:06
  posting in a Slack channel somewhere about your, you know,
• 9:09
  health issues. So we also have a bot that
• 9:12
  people can DM and that does all the work of, you know, talking
• 9:15
  them through the issues, collecting the right information
• 9:16
  and then file a ticket. Quick side note for founders who
• 9:20
  are using AI to write content, you've probably noticed this.
• 9:24
  It's incredibly fast, but it doesn't really Dr. meaningful
• 9:27
  traffic. And honestly, that's because
• 9:29
  most AI generated content is just average.
• 9:32
  There's no real insight, no depth, no structure, missing
• 9:35
  things like internal linking, and Google's getting very good
• 9:38
  at filtering that out. So instead of trying to just
• 9:41
  prompt better, I actually came across something interesting
• 9:44
  called Neena. It's a system that combines
• 9:46
  human strategy and review with AI writing to create high
• 9:50
  quality SEO optimized blog content that's actually designed
• 9:54
  to rank and keep ranking. What I like about it is that
• 9:57
  it's done for me. Their team handles the topic
• 9:59
  strategy and the content review and data markups for Google and
• 10:03
  turns blog posts into actual long term traffic assets.
• 10:08
  So if you're a founder thinking seriously about organic growth,
• 10:10
  this is one of those things worth paying attention to.
• 10:13
  You can check it out at trineena.co.
• 10:15
  That's TRYNIN a.co trineena.co. In three years, AI fluency will
• 10:24
  be a baseline expectation for every college grad, not a
• 10:27
  differentiator. It'll be a requirement.
• 10:30
  The students getting ahead are the ones building that skill
• 10:32
  right now in high school and college before it becomes table
• 10:35
  stakes. ainativestudent.com is the program doing that weekly
• 10:40
  one-on-one coaching, live cohort sessions, on demand courses and
• 10:45
  the fastest growing AI chat for students all tailored to your
• 10:48
  students specific career path. Pre Med they learn how AI is
• 10:52
  transforming diagnostics. Pre law they learn legal AI
• 10:55
  tools business they learn to think like an AI power operator.
• 10:59
  1 flat membership less than a private tutor.
• 11:02
  Cancel anytime. And if you're a founder
• 11:04
  listening to the show, you already know the cost of falling
• 11:06
  behind. Don't let your kid pay that
• 11:09
  price. Go to ainativestudent.com, apply
• 11:12
  in a couple minutes and get your student into the cohort that
• 11:15
  changes their trajectory. AI native student.com and send
• 11:18
  them to the chat. It's totally free.
• 11:22
  Cool. I love that it's this is
• 11:25
  interesting. So does it have functionality?
• 11:27
  So if I'm just thinking of IT opportunities, if I was like,
• 11:30
  how do I plug in this cable to the back of my monitor, I can
• 11:34
  take a picture of it and then put it into the chat.
• 11:36
  Would it be able to describe to me back how to do it?
• 11:39
  Yeah, totally multimodal. So images, text, video and then
• 11:43
  models themselves already have a preset, you know, corpus of
• 11:47
  knowledge that they know. Like they know how, you know,
• 11:49
  iPhones work and all this sort of stuff.
• 11:51
  But more often people will connect their proprietary
• 11:53
  company knowledge to unthread, degenerate some of these
• 11:56
  answers. So they might have some special
• 11:58
  equipment that they have documented in some confluence
• 12:01
  page deep in a sea of thousands of confluence pages.
• 12:05
  And so Andre will also surface those docs and be able to
• 12:07
  connect to whatever your internal knowledge repository
• 12:10
  is. Nice, I've got a spin off
• 12:13
  product idea for you here. We can go 5050 on this.
• 12:16
  OK, this is this is family ID. I don't know if it if it's like
• 12:22
  in your family, but the second you have a start up or something
• 12:24
  and you're like in tech, quote UN quote, basically anybody who
• 12:28
  has some kind of technical question ends up emailing or
• 12:30
  texting you of like, how does my phone work?
• 12:33
  How does my camera work? All these things like Hey you
• 12:35
  need to come check out my hunt thread because we got IT support
• 12:38
  all day there for you. That's a good point.
• 12:41
  I've been fixing family printers since I was like 15, so yeah, it
• 12:45
  might be nice to offload that to Unthread at some point.
• 12:47
  Everyone's like out just to ask Tom, He knows everything about
• 12:51
  computers. Yeah, my whole inbox for my
• 12:53
  family just wraps to unthread at this point.
• 12:56
  Amazing. Cool.
• 12:58
  What what a great utility and I could see the use cases
• 13:02
  immediately. So talk about the inception of
• 13:04
  the product, because I think that's kind of an interesting
• 13:06
  story. Yeah.
• 13:07
  So for us, the first problem that we identified was the idea
• 13:12
  that Slack was taking over the world and it was just losing
• 13:17
  important information and a sea of threads and channels.
• 13:20
  And it wasn't good at actually getting work done.
• 13:22
  It was good at just like chatting with people.
• 13:24
  And so one of the first products that we shipped was actually
• 13:28
  focused on people who support their customers on Slack, which
• 13:31
  is still a big part of the unread platform today.
• 13:33
  We were actually building a different app at the time.
• 13:36
  We had a Slack channel with our customers and they would reach
• 13:39
  out with bugs and feature requests in there.
• 13:40
  And then we would just like forget about it or you'd try to
• 13:42
  mark unread or, you know, save for later.
• 13:45
  These kind of weird tricks to try to keep track of things.
• 13:47
  And we said, like, I know as a human that I need to track this
• 13:52
  thread as work. Why can't AI know the same
• 13:56
  thing? Like it knows what I need to do.
• 13:59
  It knows how to generate answers for people.
• 14:00
  And that's almost like kind of the bare bones.
• 14:02
  That's obviously easy to do now that AI just like answer
• 14:05
  people's questions. But the next level thing was
• 14:08
  people are asking you to do something and we can assign that
• 14:12
  task to a person to do it. And that was, you know, an early
• 14:15
  prototype of fun. Thought it was just building
• 14:17
  like a ticketing infrastructure layer where I can manage my
• 14:20
  queue of things I need to do. And then the the next layer
• 14:23
  beyond that was naturally OK, well, which of these things do I
• 14:26
  not have to do anymore? Like now that agentic AI and you
• 14:29
  know, MCPS are connected to every internal tool that you
• 14:31
  have, I can look at my set of tasks that I've done and support
• 14:35
  request I've had to click around in a dashboard and do manual
• 14:38
  work for and say, let me just ask AI to automate this and see
• 14:42
  how far it gets. And that's a key metric of ours
• 14:45
  that we track and we call that our deflection rate, which is a
• 14:48
  request comes in through one of these channels.
• 14:51
  How often can AI just understand it, take action and close it end
• 14:56
  to end? Amazing, cool.
• 14:58
  And you know, the, seems like there's a couple of novel use
• 15:01
  cases here too, like accessing a knowledge base and, and things
• 15:04
  like that is almost just like an MCP, but having an MCP available
• 15:07
  in Slack is amazing, you know, because you're there already.
• 15:10
  I, I really feel like the winners in AI right now are
• 15:13
  people that are not making me go to another dashboard.
• 15:16
  It's like, come to me, you know, and then I, I'm very welcoming
• 15:20
  for you, you know, but if it's like actually then submitting a
• 15:24
  ticket, it's ordering a computer in that case and sending it to
• 15:28
  you. That seems like a little bit
• 15:30
  extra where it's actually maybe is it working in the browser and
• 15:33
  completing tasks kind of like, you know, the agentic promise
• 15:37
  kind of thing. Well, behind the scenes, yeah,
• 15:41
  AI can spin up a browser, it can do web search for most of these
• 15:45
  requests, though you'd be surprised at how often these are
• 15:47
  are automated. So we work with a couple
• 15:49
  different vendors on the hardware side, like CDW, and
• 15:53
  there are a few others where they have an API and you can
• 15:57
  connect that API to a company's kind of list of SKUs that they
• 16:00
  purchase and actually do this intend without needing to do
• 16:03
  something as hacky as like tell an AI to interact with a
• 16:06
  browser. Which again, still works for a
• 16:07
  lot of use cases. But yeah, you'd be surprised how
• 16:10
  much can be done through an API or an MCP at this point.
• 16:13
  I know it's incredible. You know, so these would be like
• 16:17
  vendor requests where basically the agents aren't responsible
• 16:20
  for like completing A transaction.
• 16:22
  That seems to be kind of the bottleneck where a lot of things
• 16:24
  fail. But if you if they have a vendor
• 16:26
  account, then you're just kind of invoiced at the end of the
• 16:28
  month in some kind of fashion so that the the agent doesn't have
• 16:32
  to complete a purchase. Yeah, and a lot of these
• 16:35
  requests still require a human in the loop.
• 16:38
  And that's something that we support as like a first class
• 16:40
  concept. So something that's maybe a bit
• 16:42
  more sensitive, like resetting your, you know, multi factor
• 16:47
  auth token. You know, that that maybe is
• 16:50
  something you want to protect with a human in the loop with an
• 16:52
  approval step. And we can also route that
• 16:54
  through this request. So say, all right, these are the
• 16:55
  sensitive ones where an approval needs to go to their manager or,
• 17:00
  you know, go through a series of executive approvals for, you
• 17:02
  know, contract reviews. And so that that's a really
• 17:05
  important part. Like there will be these times
• 17:07
  where AI is going to take a destructive action and you want
• 17:10
  to have a person take a look at that first and just kind of one
• 17:14
  click it to resolve. And yeah, you've probably seen
• 17:16
  that. You've probably seen the story
• 17:18
  with that company who got their whole production database wiped
• 17:21
  by Claude. That's where having a nice human
• 17:23
  approval step before some of these destructive actions can
• 17:26
  really help. Yeah, 100% cool.
• 17:29
  And you took this product through YC.
• 17:31
  What was that like? Yeah, YC was an amazing
• 17:34
  experience. Yeah.
• 17:36
  Yeah, I started my first company back in 2013.
• 17:40
  It's crazy. Yeah, like 13 years ago.
• 17:43
  And you know, I felt like I'd been through the ringer and kind
• 17:46
  of knew the insurance and outs and everything.
• 17:47
  But yeah, going through YC is just an amazing community.
• 17:51
  Everyone there is just super talented, like some of the
• 17:53
  smartest founders I met there. And they also really gave us a
• 17:57
  great jumping off point in getting UN threat off the
• 18:00
  ground. So yeah, whether that was
• 18:02
  introductions to customers or employees or investors, the
• 18:06
  network was huge. And yeah, would highly recommend
• 18:09
  anyone who's interested just apply.
• 18:11
  And we weren't sure we were gonna get in.
• 18:13
  We didn't actually have a working version of Unthread
• 18:15
  ready by the time we got into YC.
• 18:18
  So yeah, we definitely recommend just giving it a shot and seeing
• 18:20
  what happens. It's it.
• 18:22
  It helps probably on the application to have successfully
• 18:25
  exited a company. Probably, yeah.
• 18:28
  But, you know, they they also just look for hungry Hustler
• 18:31
  founders. Yeah, totally.
• 18:33
  You know, maybe for, you know, a post exited founder, I'm really
• 18:37
  interested in kind of the value gain from YC like investment
• 18:40
  opportunities, maybe design partners, staff, the network of
• 18:43
  course is is absolutely money. Was there anything surprising
• 18:47
  maybe about like go to market or or anything else that like
• 18:51
  really kind of aided in the launch of the product?
• 18:55
  Yeah, I think that the way YC thinks about certain things like
• 19:00
  fundraising, like hiring, like your initial go to market
• 19:03
  strategy is best in class. Basically it's something that
• 19:08
  they're, you know, it's not just some random wiki that is static
• 19:12
  and lives somewhere that you just read every now and then.
• 19:14
  Like this is a continually updated thing based on, you
• 19:16
  know, real founders experiences. So for me, it was sort of like,
• 19:20
  OK, yeah, I've been through it and I've done it my own way,
• 19:22
  just kind of like trial and error and seeing what works.
• 19:26
  But it was a good opportunity to talk to real people.
• 19:28
  Like, you know, they introduce you to the founders of Airbnb
• 19:31
  and the founder of Coinbase and, you know, all these successful
• 19:35
  companies and you can really just reach out to them for
• 19:38
  advice and they will answer. So yeah, I think that that
• 19:41
  network was really the the big unlock. 100%, you know, and I
• 19:45
  mean, why see, it's his voice 7% equity, you know, so it's
• 19:51
  expensive, but I, I would love to kind of get your take on like
• 19:54
  the start up game because it definitely seems like you're
• 19:57
  like spinning, hustling, flipping things, you know, are
• 20:00
  you seeing that's just like, hey, some percent here might
• 20:03
  turn into, you know, $100 million company down the road.
• 20:06
  Like what, What's your kind of approach to the the expense I
• 20:09
  guess of YC? Yeah, it's sort of like the
• 20:12
  benefit of having someone in your corner and what they can
• 20:16
  offer you. So one thing is just getting you
• 20:19
  off the ground faster. I mean, the the time to get to
• 20:22
  product market fit is always, you know, a race and a slog.
• 20:26
  People call it the abyss where you're just kind of navigating
• 20:29
  around trying to figure out who are your customers, how do you
• 20:32
  reach them, how do you price, what do you build?
• 20:35
  I think going through YC is an accelerant to save you a lot of
• 20:39
  time in the abyss. And so that's valuable in a lot
• 20:43
  of ways. Like your own time is obviously
• 20:44
  valuable, but you also have to be, you know, early to market.
• 20:47
  You can't be sitting around in the abyss for years.
• 20:49
  You've got to get that product out there and and, you know,
• 20:52
  beat the other competitors to market.
• 20:54
  Yeah. And you've got to have traction
• 20:55
  quick because, I mean, you know, we're kind of redefining what a
• 20:58
  potential mode is in, in startups and like, the
• 21:01
  technology, ain't it, you know, So, you know, having to keep
• 21:05
  what's that? Toad is cheap now.
• 21:08
  Yeah, free essentially, you know, so like having that
• 21:13
  network, you know, for potential distribution and being on the
• 21:15
  tip of the right person's tongue is like, I mean that that might
• 21:20
  be defensible. And maybe for software, the only
• 21:22
  defensible component you know, which is which is interesting.
• 21:27
  So graduate YC, let's talk about the first couple of customers.
• 21:31
  So did you were you able to pull like design partners and stuff
• 21:34
  from YC and kind of like how, how did you get your first
• 21:37
  couple of customers? Yeah.
• 21:39
  So the the first were definitely YC companies, mostly companies
• 21:44
  that were in previous batches a little older than us, but, you
• 21:47
  know, we we thought kind of experienced this pain point a
• 21:49
  little bit more. And so, yeah, again, the benefit
• 21:52
  is, you know, I could just cold reach to a a founder to their
• 21:57
  e-mail address and be like, hey, I've got this thing I'm working
• 21:59
  on, you know, do you want to check it out and see if this
• 22:01
  works for you? At this point, you know, the YC
• 22:03
  network is pretty big. And I actually don't even know
• 22:06
  if that strategy works as well as it did just three or four
• 22:09
  years ago, just because now that the community is so big.
• 22:12
  But back then it was as simple as, yeah, DM ING a founder with
• 22:15
  a real ask, like not a, you know, spam e-mail, not kind of a
• 22:19
  sales call masked in a feedback call, like a really transparent,
• 22:24
  straightforward ask of how they can help you and why you chose
• 22:27
  them over all the other people. And the reply rate is, is just
• 22:31
  super high for that sort of thing.
• 22:32
  And we got a lot of people to come on and tell us what they
• 22:34
  didn't like about our product, what challenges they had that we
• 22:37
  didn't think about what what challenges we thought they had
• 22:39
  that they didn't have. So really just getting on those
• 22:41
  initial calls with people who are within our target, you know,
• 22:44
  ICP was huge and just YC made that a lot easier.
• 22:48
  Amazing. It's kind of like the secret
• 22:50
  handshake. Yeah, yeah, exactly.
• 22:53
  And I hope that's not going away again with the the flood of of
• 22:56
  founders coming in now there's just the the community's so
• 22:58
  massive, but there are other ways too.
• 23:01
  I mean some of our other customers actually came in
• 23:03
  through friends of ours who introduced us to a vendor that
• 23:08
  they had that was one of our first customers that was kind of
• 23:10
  funny. We're like, hey, I really want
• 23:11
  to use, we were really want to talk to this company and we
• 23:14
  noticed that you use them as a customer.
• 23:17
  Could you connect us to their like CSM that you work with and
• 23:20
  and go from there? And that was an inroad with a
• 23:22
  customer. So you can get creative with
• 23:23
  these sorts of things. I mean, as long as, again,
• 23:25
  you're really clear with why you're asking for something and
• 23:27
  people are generally down to be helpful.
• 23:30
  I love it. So you know that that e-mail
• 23:32
  structure then is like quick introduction, very specific ask,
• 23:36
  and then why you've decided to ask them instead of somebody
• 23:39
  else. Yeah, yeah.
• 23:40
  And I respond well if people come to me with that.
• 23:43
  Yeah, So do I, you know, that's The thing is like when you, when
• 23:48
  you need something, people actually kind of line up to help
• 23:51
  you, you know, it, it, it doesn't need to be so
• 23:54
  transactional. It's just like, hey, I'm working
• 23:56
  on this, I can I get XYZ. It's like direct is always best,
• 24:01
  yeah? Then, you know, if you can
• 24:02
  actually be helpful, it's like, hey, I'm looking for feedback on
• 24:03
  XYZ and I'm like, well, I only really, I'm familiar with ABC so
• 24:07
  I wouldn't be a good fit, but I know this other founder who's
• 24:10
  really good at XYZ, so I'll send you his way.
• 24:13
  Yeah, totally. I love it.
• 24:15
  I mean, ultimately that's why I love this podcast is you become
• 24:17
  kind of a hub and talk to founders all day and it's like
• 24:20
  you need to talk to this guy, you know, which is a fun place
• 24:23
  to be. Yeah, cool.
• 24:26
  Looking at the pricing page, looks like it, you know, it's
• 24:28
  basically 50 bucks per agent per month with a minimum of five
• 24:33
  seats. And now are we, when we're
• 24:35
  talking about seats, are we talking about individual unique
• 24:38
  agents? Yeah, those are humans using the
• 24:41
  platform and yeah, we priced by a couple different things.
• 24:44
  Yeah, per human support Rep who's in the system, you know,
• 24:49
  managing a ticket queue or building these automations or
• 24:51
  trying to solve, you know, their own support problems.
• 24:54
  We also do, you know, for people who don't want to bring their
• 24:57
  own AI, you know, there are some AI related costs so we can scale
• 25:01
  up with, you know, the volume and sort of the cogs that we
• 25:03
  would face around that. So yeah, there's a couple
• 25:05
  different models and and it also depends like that that price
• 25:08
  point is sort of the introductory point to Unthrow
• 25:10
  and that if you have, you know, more advanced usage, you know,
• 25:12
  it goes up from there. Yeah, you bet.
• 25:14
  So if I have 100 people in my Slack channel, would I want all
• 25:19
  or would I want 100 seats? I mean, I know you'd want me to
• 25:21
  have 100 seats, but is that how people are setting it up?
• 25:25
  No, no, we don't try to charge per user in the chat, like per
• 25:28
  person who's just there communicating with each other.
• 25:31
  It's really about the people who are tasked with solving the
• 25:33
  problem and who care about building the workflows.
• 25:35
  So yeah, I mean, for a typical, like large company, that's
• 25:39
  either their, their IT operations team or their
• 25:42
  customer support team. So yeah, like an average person
• 25:45
  who's just in there to ask for help, we wouldn't, you know,
• 25:48
  they they don't need a seat. Nice, I love this because it's
• 25:51
  like helping the besieged IT department it sounds like.
• 25:56
  Oh, they are besieged, let me tell.
• 25:57
  You no, I know I get the spillover because they're so
• 26:00
  grisly. They, you know, people ask me
• 26:02
  stuff. I'm like, hey, I say IT question
• 26:04
  like, oh, I don't want to. I don't want to bother them.
• 26:07
  Yeah, yeah, yeah, exactly. So I mean, that's where the AI
• 26:10
  can comes in, comes into play is you can ask that any question.
• 26:13
  Sometimes people actually don't know whether to go to IT or to
• 26:16
  go to HR or to go to their manager to ask a question.
• 26:20
  And so I think news actually come through on thread as this
• 26:22
  kind of unified intake layer. And you know, if this is an HR
• 26:26
  ticket, we'll route to the HR team for you.
• 26:27
  If it's for IT, we'll route to IT for you.
• 26:29
  You don't need to know, you know where their help desks are,
• 26:32
  which page to go to. You kind of have one intake
• 26:34
  source and just let AI figure out the rest.
• 26:37
  Incredible. And I know we're working up on a
• 26:39
  timetable here, so I want to make sure that every use case is
• 26:42
  kind of represented. What are some other great use
• 26:44
  cases for unthread? So once you build a foundation
• 26:49
  around conversational AI that lends itself to support within
• 26:54
  these chat platforms like Slack and teams that were previously
• 26:57
  left untracked. And so ITHR, these internal OPS
• 27:01
  teams are one case. Another case is actual external
• 27:04
  customer support, which I mentioned was like one of our
• 27:07
  earlier products we launched and we use Unthread for that for all
• 27:10
  of our customers. When they talk to us, every
• 27:12
  request that they have goes through Unthread.
• 27:14
  Now, the automations that you build around that and the
• 27:16
  workflows and how you set up SLAS is different.
• 27:20
  And Unthread allows you to configure these different teams
• 27:23
  with their own kind of workspaces and their own
• 27:25
  different ways of operating. So the IT team can have their
• 27:27
  own policy of how they do things.
• 27:30
  Similarly, the customer support team can connect, you know,
• 27:32
  their Slack channels, their e-mail address, maybe even a
• 27:35
  Discord that they use for support.
• 27:37
  And they can set up a totally different pipeline where they
• 27:39
  have different integrations. Maybe they're integrating with
• 27:41
  Stripe and, you know, with your CRM and we'll route it to the
• 27:45
  right, you know, dedicated CSM for that account.
• 27:48
  That's another great use case. And again, it's just sort of
• 27:50
  like once you've built a good foundation around how to have,
• 27:55
  you know, a conversation with a user that seamlessly works in AI
• 27:58
  when it's needed, and they can also have AI get out and loop in
• 28:02
  a person at the right time. Really any sort of, you know,
• 28:05
  support team is eligible for it. Incredible.
• 28:09
  I would love to have you back and kind of see a live demo of
• 28:11
  it in action. You know, do something for
• 28:12
  YouTube when we got a little bit more that would, that would be
• 28:15
  really cool. Yeah, I would be more than happy
• 28:18
  to do that. Fantastic.
• 28:19
  It's great to meet you. The the tool is unthread dot IO.
• 28:23
  Tom, where's the best place for people to connect with you
• 28:25
  personally online? Follow me on LinkedIn.
• 28:29
  I post a bunch of horrible posts from time to time trying to be
• 28:33
  funny, and sometimes they're helpful, sometimes they're just
• 28:35
  dad jokes. Yeah, just Tom Bashant on
• 28:38
  LinkedIn. I might be the only one, I don't
• 28:40
  know, But yeah, find me. Follow me there.
• 28:42
  And then check out unthread dot IO.
• 28:44
  You can reach me through there. Nice, you ever do any downhill
• 28:48
  mountain biking? You know I have not, but that's
• 28:51
  been on my list for a long time. I tell you the place to do it is
• 28:54
  Cresta Bute, Colorado. They have the most incredible
• 28:59
  downhill courses. I I've been a few times.
• 29:02
  I'm not going to go again because I kind of felt like I
• 29:04
  got out of it unscathed and I don't want to push my luck.
• 29:07
  But yeah, I hear it's kind of like you're improvising with
• 29:10
  your life on the line. You got to know like which turns
• 29:12
  to make and. You'll definitely think you're
• 29:15
  about to die, but you know, the, the suspension is so deep that
• 29:18
  you're just like, Oh my gosh, you're just, it's kind of like,
• 29:21
  it's kind of like, you know, rolling down the hill on a
• 29:24
  marshmallow. It's it's so fun.
• 29:26
  It's my favorite kind of cycling, although I'll never do
• 29:28
  it again. Yeah, I mean, I feel like I'm
• 29:30
  going to die when I bike around Brooklyn, So I mean, I'm already
• 29:32
  used to that, so I might as well just enjoy it a little bit more.
• 29:35
  That's right, Tom, great to meet you.
• 29:37
  Thanks for joining me. Yeah, thanks for having me.
• 29:40
  Ryan Here's what separates the founders who become icons from
• 29:42
  the ones who stay invisible. A weekly newsletter.
• 29:45
  Not because it's trendy. Because inbox is where business
• 29:48
  actually gets done. It's where investors read your
• 29:51
  thinking before they take the call, where future hires decide
• 29:54
  they want to work for you, where customers remember why they
• 29:57
  chose you. Your social posts disappear in a
• 29:59
  couple hours. Your newsletter compounds for
• 30:02
  years. Inbox Alchemy builds your entire
• 30:04
  founder newsletter operation from scratch.
• 30:07
  The writing. The design designed the
• 30:08
  beautiful branded website and the growth engine that takes you
• 30:11
  from zero to 25,000 subscribers in your first six months.
• 30:16
  They manage everything, so your personal brand is working for
• 30:18
  you around the clock, building reputation, deepening
• 30:21
  relationships and driving real pipeline, all while you run your
• 30:25
  company. The founders building e-mail
• 30:27
  audiences right now are playing a game the algorithm can never
• 30:30
  touch. That's Inbox alchemy.co and go
• 30:33
  build yours. While you were recording your
• 30:37
  last episode, Casey Taylor was out there pitching your podcast
• 30:40
  to brands and securing sponsorship deals.
• 30:43
  Well, she's not yet, but she could be.
• 30:45
  Casey is Howdy's AI partnership agent and she does something
• 30:49
  pretty wild. She actually secures
• 30:51
  sponsorships for your show. Not a match list that you have
• 30:54
  to work through yourself, not a template you have to send.
• 30:56
  She handles the outreach to follow up the whole thing and
• 30:59
  brings back partners. You started your podcast because
• 31:02
  you had something worth saying. Casey makes sure that to get
• 31:05
  paid for sayingitgomeetcasey@gethowdy.com.
• 31:09
  That's GETHOW di.com. 2 minutes to get started, then Casey takes
• 31:14
  it from there. That's howdy
• 31:16
  partnershipsgethowdy.com. In three years, AI fluency will
• 31:22
  be a baseline expectation for every college grad, not a
• 31:25
  differentiator. It'll be a requirement.
• 31:27
  The students getting ahead are the ones building that skill
• 31:30
  right now in high school and college before it becomes table
• 31:33
  stakes. ainativestudent.com is the program doing that weekly
• 31:37
  one-on-one coaching, live cohort sessions, on demand courses and
• 31:42
  the fastest growing AI chat for students all tailored to your
• 31:46
  students specific career path. Pre Med they learn how AI is
• 31:50
  transforming diagnostics. Pre law they learn legal AI
• 31:53
  tools business they learn to think like an AI power operator.
• 31:57
  1 flat membership less than a private tutor.
• 31:59
  Cancel anytime. And if you're a founder
• 32:01
  listening to the show, you already know the cost of falling
• 32:04
  behind. Don't let your kid pay that
• 32:06
  price. Go to ainativestudent.com, apply
• 32:09
  in a couple minutes and get your student into the cohort that
• 32:12
  changes their trajectory. AI native student.com and send
• 32:16
  them to the chat. It's totally free.