00:00:06.240 Thank you for being here today.
00:00:08.800 I am Ali Merchant, the CEO of CallBotics.
00:00:13.759 And
00:00:15.839 today I'm going to share with you guys a
00:00:18.560 practical and simple blueprint called
00:00:22.720 the ACE loop, which will help you deploy
00:00:00:26.160 AI agents
00:00:28.400 without the chaos, without the overwhelm
00:00:32.320 and the long timelines that leaders
00:00:35.360 generally in contact center experience.
00:00:39.840 But before we get into that, let me
00:00:42.640 share a little bit about myself about
00:00:45.600 CallBotics. So you guys know that
00:00:50.960 this blueprint is not just a theory, a
00:00:55.360 philosophy or something that was created
00:00:58.239 out of a whiteboard, but instead it was
00:01:02.079 built by experiencing
00:01:06.320 the realities of contact center for 17
00:01:10.080 years firsthand.
00:01:13.680 So
00:01:15.280 17 years ago I founded a company called
00:01:18.960 AM Infoweb.
00:01:21.040 It basically helps outsourcing companies
00:01:24.640 and enterprises with with in-house
00:01:27.600 contact centers set up their operations
00:01:32.400 and scale it.
00:01:35.840 And whatever you guys do on a regular
00:01:38.640 basis, me and my team, we have done it
00:01:42.240 for these 17 years now. Whether that be
00:01:45.840 opening up offices, setting up floors,
00:01:49.520 hiring, training, managing large teams,
00:01:52.159 the KPI pressures that come with them.
00:01:55.600 And just like you guys, I have also had
00:01:58.320 those long nights where I've been on
00:02:00.799 calls, not
00:02:03.680 um been not been on calls, you know, on
00:02:06.960 the floor, not being able to reach to a
00:02:10.000 family function or my kids events or a
00:02:13.040 date night. And I have been through all
00:02:16.160 of that. So when I say that what what
00:02:20.319 works and what doesn't work in a contact
00:02:22.239 center, it's because I have lived it
00:02:26.480 firsthand with my team just like you
00:02:29.520 guys have.
00:02:34.560 So and
00:02:38.080 while we did all of this right, one
00:02:42.640 thing
00:02:44.239 uh that we that we observed was that you
00:02:47.519 know all these years that there are
00:02:51.519 multiple challenges but the biggest
00:02:53.920 challenge that we faced as a contact
00:02:56.560 center leader was you know year on year
00:02:59.680 since 2008 I saw that technology was
00:03:02.959 changing and the frustration or the
00:03:07.280 challenges that came with deploying it.
00:03:09.599 Understanding what works, what doesn't
00:03:11.360 work, how to kind of really make or
00:03:14.159 create value out of that for my
00:03:15.840 organization was the biggest challenge
00:03:20.000 and in one particular project everything
00:03:24.159 kind of you know changed for us. What
00:03:27.519 happened was one of our
00:03:30.959 customers they needed a CRM and this was
00:03:35.840 back in 2015.
00:03:38.080 So we went to the best reputed vendor
00:03:41.360 that was out there. They came in gave a
00:03:44.720 great presentation. The verbiage, the
00:03:47.440 terminology was very impressive. And
00:03:49.680 then the proposal arrived
00:03:52.560 and what the proposal said was that it
00:03:55.680 will take 6 months and £95,000
00:03:59.599 just for the blueprint. Not even the
00:04:01.840 product. This is 10 years back, right?
00:04:04.720 I'm sure you guys have also experienced
00:04:06.799 this. And for the final product, what
00:04:09.599 they said was 18 months to build it and
00:04:12.640 half a million pounds.
00:04:15.439 Sure, the budget was a little heavy,
00:04:19.279 which was fine with us, our client. But
00:04:22.000 the bigger bigger challenge was the
00:04:23.840 timeline.
00:04:25.840 And we knew that instantly we knew that
00:04:29.360 this is not going to work for us or our
00:04:31.520 customers rather
00:04:33.840 because our customers they needed
00:04:35.840 agility and every future change meant
00:04:40.400 that we had to go back in the same
00:04:42.720 expensive but more importantly very very
00:04:45.360 very long timelines. Right? So that's
00:04:48.800 when we understood that
00:04:51.759 we need to build a team that is meant
00:04:56.160 for contact center leaders not a tech
00:04:59.840 team which is built by contact center
00:05:02.080 leaders so it has real impact but also
00:05:04.800 understands their challenges. Hence we
00:05:07.039 set up a team internally. Long story
00:05:09.600 short we built that same CRM just for
00:05:12.800 £15,000. It's been a decade, 10 years
00:05:15.840 now and still that CRM works and it
00:05:18.800 works really really well
00:05:22.240 and once we built that internal
00:05:24.080 capability
00:05:25.600 everything changed for us
00:05:28.160 and for the past 10 years
00:05:31.440 that tech team has been building these
00:05:34.160 multiple products like CRM, dialers,
00:05:36.880 workflow tools for our customers
00:05:40.400 and they provided not only the
00:05:43.199 reliability but the most important thing
00:05:45.600 which is applicable in a contact center
00:05:48.560 environment is the speed to deploy it.
00:05:52.160 And then obviously came the era of AI.
00:05:55.680 And while we did multiple projects, we
00:05:58.479 set up a dedicated good team just to
00:06:02.400 build
00:06:04.319 voice automation and conversational AI.
00:06:06.880 We started
00:06:08.960 to source the experts globally to build
00:06:12.639 that platform but we had a duty towards
00:06:15.759 our clients right we had to provide the
00:06:18.800 best that was out there not what just we
00:06:21.120 made. So constantly while building that
00:06:24.800 platform we were also
00:06:28.080 we also sourced technology from outside
00:06:31.840 and
00:06:34.080 we basically you know we we were
00:06:36.400 basically benchmarking this technology
00:06:38.960 how it is working how things are
00:06:41.360 happening and what we saw during those
00:06:44.000 tests was something what surprised us as
00:06:46.960 well that the platform which we called
00:06:50.160 CallBotics that we had built internally
00:06:52.080 was performing on use cases 20 30% times
00:06:55.520 better meaning it was resolving
00:06:58.880 you you know calls 20 30 times more and
00:07:02.400 the CSATs were also higher on the queues
00:07:04.800 where we had CallBotics and since the
00:07:09.039 goal was always been you know that we
00:07:12.720 wanted to build products that supported
00:07:14.720 contact center leaders we've understood
00:07:16.479 that it is time that we launch this as a
00:07:19.199 separate or a different company because
00:07:22.160 the other company that we had was more
00:07:23.919 like a consulting operations company and
00:07:26.479 this is like a pure SaaS company.
00:07:29.680 What I mean to say by or convey by all
00:07:33.919 of this is that when we say that
00:07:39.120 our resolution rates are higher, it's
00:07:42.319 because it's not the technology or the
00:07:45.520 platform is not just built by tech
00:07:49.759 leadership but more of the leadership
00:07:53.440 that we have within the organization are
00:07:55.599 contact center leaders and hence the
00:07:57.759 technology or the platform form is built
00:08:00.240 on small small nuances that you and me
00:08:04.240 we have seen for so many years which
00:08:06.639 makes the resolution rates so high
00:08:10.160 and that's why I'm here today to share
00:08:13.759 with you this blueprint because while
00:08:16.560 building that tech we built that
00:08:18.400 technology and we deployed these
00:08:20.400 multiple external platforms voice
00:08:22.720 automation as well we kind of made any
00:08:26.479 number of mistakes and finally we
00:08:28.720 figured out what was the best
00:08:30.960 way to deploy voice automation and I'm
00:08:34.159 sure you know a lot of you will resonate
00:08:36.958 with this at least I did for many many
00:08:39.200 years that every time a new technology
00:08:42.320 came in or a new platform came in there
00:08:45.120 was always this pressure on me to
00:08:47.040 implement it and also at the end of the
00:08:50.000 day I had to prove the value to the
00:08:51.760 leaders to the rest of the team to my
00:08:54.080 colleagues everyone
00:08:57.360 so enters
00:09:00.160 the ACE loop. This is the blueprint that
00:09:03.279 we created which has three major
00:09:07.519 parts to it. The first is assess, second
00:09:10.880 is configure and third is elevate. This
00:09:13.760 is the framework that we created or the
00:09:15.600 blueprint that we created. Assess
00:09:19.920 is basically what we talk about here in
00:09:22.959 assess is about
00:09:25.519 you assess the simplest of the workflows
00:09:29.200 first and then you know you put them up
00:09:31.279 for voice automation not the most
00:09:33.680 complex ones. The second part of it is
00:09:37.760 configure, right? Meaning generally when
00:09:41.040 we talk about configure, we think about
00:09:44.959 integrations and you know other
00:09:47.120 technology related heavy lifts. But what
00:09:49.519 we mean to say about configuration
00:09:51.200 configure is you have to configure your
00:09:53.360 AI agents and iterate them in real time.
00:09:58.240 The third thing is elevate which is
00:10:01.839 basically now that you have assessed
00:10:06.000 which workflow to automate and you have
00:10:09.519 configured your AI agents it's time to
00:10:14.160 expand those to other workflows
00:10:17.680 integrate it with your current workflow
00:10:19.920 system or current workflows and tech stack
00:10:24.079 and most importantly now that we know
00:10:27.040 what the AI agents are able to do
00:10:30.320 reliably is when we re-engineer the
00:10:33.760 processes around that
00:10:37.519 and
00:10:40.240 let's first talk a little about assess.
00:10:44.160 So with assess
00:10:47.360 as I mentioned we need to first
00:10:49.839 understand what are the easiest
00:10:52.640 workflows
00:10:54.640 which are very very much defined and on
00:10:58.240 top of that where value can be proved
00:11:02.079 immediately.
00:11:03.600 So
00:11:05.200 the thing that we have observed is a lot
00:11:07.040 of time AI doesn't fail but the
00:11:10.000 processes around it are so complex
00:11:12.959 that
00:11:14.880 the AI in the first go we take
00:11:19.040 up the most most complex use cases. So
00:11:21.279 this is what we figured with all the
00:11:23.360 mistakes that we made that we should
00:11:26.399 actually take the high volume if
00:11:28.399 possible but if not low judgment and
00:11:31.920 repeatable processes the easiest ones
00:11:35.200 where there will be the least amount of
00:11:37.360 friction. And if that is not possible
00:11:41.200 the second best option would be those
00:11:44.000 backlog or surge calls that our current
00:11:47.279 teams are not able to handle. And then
00:11:50.240 if even that is not available go for the
00:11:54.320 after hours or weekend queues where you
00:11:57.519 are understaffed or don't have staff
00:11:59.600 at all and then you go for you know side
00:12:04.079 queues if even that is not available
00:12:08.720 and what we have observed is that
00:12:11.440 in any contact center this
00:12:14.720 kind of workflows are easily available.
00:12:17.839 some of the examples if I have to share
00:12:20.320 and probably already identified some
00:12:22.560 within your workflows.
00:12:24.800 On the inbound side, we saw that these
00:12:26.959 simple lookups when people call and they
00:12:29.200 just need some information or balance
00:12:31.680 check or payment checks. On the outbound
00:12:35.279 side, confirmations, callbacks, you
00:12:38.639 know, voicemail handling. These are the
00:12:40.720 queues where we saw that not only did
00:12:43.920 the internal teams not mind it. But
00:12:47.360 with that the customers also were not
00:12:50.480 looking for too much
00:12:52.800 judgment or emotions. They were just
00:12:55.120 looking for updates. And just think of
00:12:58.639 it this way.
00:13:00.720 We a lot of times think that maybe this
00:13:03.600 will not be you know they will not be
00:13:05.440 happy with it. But what we have seen is
00:13:07.519 just that feeling for 10 years. You and
00:13:10.160 me whenever we have called a queue
00:13:12.720 either we expect an IVR a robotic IVR or
00:13:16.560 we are expecting
00:13:18.560 long wait at least 5 to 10 minutes.
00:13:23.279 With this what happens is suddenly a
00:13:25.839 customer calls and boom here you know
00:13:28.560 they are already talking to someone
00:13:30.560 immediately and because the agent has
00:13:34.079 all the information integrated within
00:13:36.320 itself it doesn't have to even read
00:13:38.399 through those documents they just have
00:13:40.399 the answers so your AHT also drops
00:13:43.120 tremendously
00:13:45.120 so that is what assess is all about
00:13:49.200 simple repeatable workflows
00:13:53.120 And then the second component that
00:13:55.519 we talked about was configure.
00:13:58.720 And when we say configure, it is about
00:14:01.279 every brand has their own unique voice.
00:14:04.720 So figuring out how it speaks,
00:14:07.760 understands the conversation and very
00:14:09.920 importantly
00:14:11.760 at what point of time should the AI
00:14:14.160 agent transfer that call to a human with
00:14:19.120 full context. That is the third part.
00:14:21.600 Escalate it and fourth iterate it on a
00:14:24.880 daily basis once you have deployed it
00:14:27.519 and you know improve the AI agent.
00:14:30.240 To give a little more understanding what
00:14:32.000 I mean to say about this is
00:14:34.800 let me give you a use case or a case
00:14:37.680 study. Let me talk about that that we
00:14:40.079 did with a customer. So they were a
00:14:42.399 benefits administration firm and
00:14:47.519 they had people calling in a large
00:14:49.680 contact center. They had people calling
00:14:51.199 in trying to understand what were the
00:14:53.760 benefits they had with their customers,
00:14:56.480 sorry with their employers, what
00:14:58.959 insurance did they have, what sort of
00:15:00.639 premiums did they need to pay, how do
00:15:03.120 they navigate the portal and any other
00:15:06.480 thing around it. So on a Friday they
00:15:10.560 reached out to us that we need a few I
00:15:14.240 think close to 80 to 100 people. We
00:15:16.959 said that we cannot get you and they
00:15:18.480 wanted to deploy it in 2 days. So we
00:15:21.120 said that we could not do it. We could
00:15:22.800 not get those many people in 2 days and
00:15:25.600 then we said that you could but use the
00:15:27.839 platform and they were always hesitant
00:15:31.040 to do so but once they deployed this now
00:15:35.600 they had no choice right. So they said
00:15:37.519 that okay let's go ahead and deploy it.
00:15:39.279 So on a Friday they came to us they
00:15:41.839 uploaded all the
00:15:44.560 training material they used for their
00:15:47.839 human agents to be trained. Monday the
00:15:51.120 AI agents were live.
00:15:54.160 The AI agents started taking calls. 45%
00:15:57.440 of the calls on day one they were able
00:15:59.680 to manage
00:16:02.079 and then throughout the week from
00:16:05.759 live calls they kept
00:16:08.880 fine-tuning those AI agents just after 5
00:16:12.720 days on such a complex use case they
00:16:15.600 were automating close to 80% of the
00:16:17.920 calls their AHT obviously had dropped
00:16:21.279 their CSATs were higher just within that
00:16:23.519 one week because they had these active
00:16:25.360 surveys that they were taking over
00:16:27.120 the calls and in general maybe they did
00:16:29.839 not use to earlier but now they were
00:16:32.160 because they wanted to understand if
00:16:33.680 their CSATs or their customer
00:16:35.440 satisfaction was taking a hit live but
00:16:37.759 what they observed was no that
00:16:40.320 their customer satisfaction grew over a
00:16:43.440 period of time over that one week
00:16:46.720 now that the AI agents were configured
00:16:50.320 and they had a lot of confidence right
00:16:53.199 is when you know
00:16:55.600 the leadership was convinced that there
00:16:57.759 is value in this and more importantly
00:17:00.000 the human agents they also saw the value
00:17:02.079 that they now did not have to take care
00:17:04.640 of repetitive simple query
00:17:08.079 based calls but they could actually
00:17:10.160 focus on the calls that really matter
00:17:12.480 where they could actually create value
00:17:14.880 and have more satisfying jobs. So next
00:17:18.319 what they did was they started expanding
00:17:20.480 their
00:17:22.079 to different workflows.
00:17:24.720 Now over a period of time they saw that
00:17:27.199 within that workflow 84% of calls were
00:17:30.480 managed by the AI agents and because of
00:17:33.760 the location that they were based in
00:17:35.600 they saw 78% reduction in their cost as
00:17:39.280 well. So now they shifted to more
00:17:42.400 complex workflows. And in one of the
00:17:46.320 workflows I remember they had 183
00:17:49.919 documents which were basically guides,
00:17:52.640 product info, SOPs, catalogs which
00:17:55.679 would give these AI agents which used to
00:17:58.400 give their human agents the information
00:18:00.960 that were required to resolve those
00:18:02.960 queries and calls right and
00:18:07.360 what happened after that was they kept
00:18:10.160 you know going to the next next use
00:18:12.320 case. So this is what elevate is. Elevate
00:18:16.080 is around expanding that and then they
00:18:19.520 started also re-engineering the
00:18:21.280 processes. Now that they knew
00:18:24.160 what the AI agents could
00:18:27.919 reliably handle, the next step was how
00:18:31.200 do they rework the entire operations
00:18:33.760 around it? Because now they like for
00:18:36.559 example they understood that within this
00:18:38.400 queue 95% of the calls are automated. So
00:18:41.600 they moved those people to another
00:18:43.280 queue. They needed a few people to
00:18:45.760 orchestrate the automation. So they
00:18:47.520 started retraining them. They knew that
00:18:51.200 they need to divide a few queues into
00:18:53.360 sub-queues to be able to automate at a
00:18:56.559 higher percentage and that's how they
00:18:58.240 got to that 84%. So that's what they
00:19:00.480 started and the beautiful thing about
00:19:02.480 this ACE loop is that it is a constant
00:19:05.919 framework. You go from one workflow to
00:19:08.640 other workflow, one queue to other queue
00:19:11.280 assess
00:19:13.440 simple workflows. assess you know what
00:19:16.400 can be automated what are the metrics
00:19:19.039 that show you that results are delivered
00:19:21.679 CSAT is increasing the AHT is decreasing
00:19:24.880 those kind of things and then obviously
00:19:26.799 you know configure those AI agents for
00:19:29.520 those particular use cases and queues and
00:19:31.919 then obviously you know keep expanding
00:19:33.679 them so that's they kept repeating it
00:19:36.640 repeating it at this point of time
00:19:38.400 earlier they had adopted this just from
00:19:42.480 a thought process that
00:19:45.919 they need to just from thought
00:19:48.480 process that they need to you know take
00:19:50.400 care of a burning firefighting situation
00:19:53.520 but today they have deployed it across
00:19:55.679 everywhere and their agents and
00:19:57.679 leadership couldn't be happier.
00:20:01.200 So to sum it up, what we have observed
00:20:03.520 in this 17 years making mistakes while
00:20:06.000 deploying and creating this AI agents is
00:20:09.360 that the contact centers that win won't
00:20:12.880 be the most automated.
00:20:15.360 They will be the most that are adaptive
00:20:17.360 because every contact center, every use
00:20:20.080 case, every workflow is very unique and
00:20:23.440 that we have to keep in mind and
00:20:26.320 accordingly start adapting it because if
00:20:29.520 not today, we have to do it tomorrow. I
00:20:32.000 remember when we started off and IVRs
00:20:34.559 were the new thing, we ourselves, me
00:20:37.200 myself as a CEO of the company, I pushed
00:20:39.520 back for some point of time and then you
00:20:41.840 know we had to lose a lot of customers
00:20:43.760 because eventually everyone like today
00:20:45.760 IVR is not no more like a new thing,
00:20:48.720 right? Same way what we have seen
00:20:51.840 with our customers is that AI agents is
00:20:55.600 not a new thing to most of them right
00:20:58.000 now. they're just automating as much as
00:21:00.320 they can.
00:21:02.880 The other thing about is that within
00:21:06.320 these 30 minutes obviously there is this
00:21:08.320 very basic level or you know at a
00:21:11.280 superficial level I can share regarding
00:21:13.120 the blueprint right and as a contact
00:21:16.320 center fellow contact center leader I
00:21:18.720 would want you guys also to spend more
00:21:20.880 time with your families not you know
00:21:22.799 firefighting staying on calls so even if
00:21:25.840 you work with someone else or you know
00:21:29.039 you already have or are still not
00:21:31.600 thinking of deploying AI agents. But
00:21:36.960 what we will do for you is if you stop
00:21:39.120 by at our stand M6 M66, we can, you
00:21:42.960 know, create a specialized, tailored,
00:21:46.240 detailed document that will resonate
00:21:49.919 with your workflows and you can
00:21:52.960 basically, you know, take that back to
00:21:55.280 your team and automate or at least start
00:21:59.280 evaluating what you can, you know,
00:22:01.280 automate when it comes to your voice
00:22:04.080 queue.
00:22:05.120 And
00:22:07.760 that will just be you know give you a
00:22:09.919 more resilient faster and more cost
00:22:12.240 effective and more happier agents
00:22:15.600 operations.
00:22:17.120 Thank you so much.
Ali Merchant introduces the ACE loop, a practical blueprint for deploying AI agents in contact centers without long timelines or operational chaos. He explains that the framework is built from 17 years of real contact center experience, not theory. He shares how AM Infoweb faced slow, expensive tech deployments in the past and built an internal product team to deliver faster, more usable tools. When the AI era arrived, they built CallBotics while also benchmarking external platforms, and found CallBotics delivered higher resolution rates and CSAT because it was designed by contact center leaders around real operational nuances.
The ACE loop has three parts: Assess, Configure, and Elevate. Assess by starting with simple, repeatable, low-judgment workflows where value is quickly provable. Configure focuses on tuning AI agents on real calls, defining escalation rules, and iterating daily. Elevate is expanding automation across workflows, integrating into the tech stack, and then re-engineering operations based on what AI can reliably handle. The loop is continuous and repeatable, helping contact centers adapt workflow by workflow.
00:00:06.240 Thank you for being here today.
00:00:08.800 I am Ali Merchant, the CEO of CallBotics.
00:00:13.759 And
00:00:15.839 today I'm going to share with you guys a
00:00:18.560 practical and simple blueprint called
00:00:22.720 the ACE loop, which will help you deploy
00:00:00:26.160 AI agents
00:00:28.400 without the chaos, without the overwhelm
00:00:32.320 and the long timelines that leaders
00:00:35.360 generally in contact center experience.
00:00:39.840 But before we get into that, let me
00:00:42.640 share a little bit about myself about
00:00:45.600 CallBotics. So you guys know that
00:00:50.960 this blueprint is not just a theory, a
00:00:55.360 philosophy or something that was created
00:00:58.239 out of a whiteboard, but instead it was
00:01:02.079 built by experiencing
00:01:06.320 the realities of contact center for 17
00:01:10.080 years firsthand.
00:01:13.680 So
00:01:15.280 17 years ago I founded a company called
00:01:18.960 AM Infoweb.
00:01:21.040 It basically helps outsourcing companies
00:01:24.640 and enterprises with with in-house
00:01:27.600 contact centers set up their operations
00:01:32.400 and scale it.
00:01:35.840 And whatever you guys do on a regular
00:01:38.640 basis, me and my team, we have done it
00:01:42.240 for these 17 years now. Whether that be
00:01:45.840 opening up offices, setting up floors,
00:01:49.520 hiring, training, managing large teams,
00:01:52.159 the KPI pressures that come with them.
00:01:55.600 And just like you guys, I have also had
00:01:58.320 those long nights where I've been on
00:02:00.799 calls, not
00:02:03.680 um been not been on calls, you know, on
00:02:06.960 the floor, not being able to reach to a
00:02:10.000 family function or my kids events or a
00:02:13.040 date night. And I have been through all
00:02:16.160 of that. So when I say that what what
00:02:20.319 works and what doesn't work in a contact
00:02:22.239 center, it's because I have lived it
00:02:26.480 firsthand with my team just like you
00:02:29.520 guys have.
00:02:34.560 So and
00:02:38.080 while we did all of this right, one
00:02:42.640 thing
00:02:44.239 uh that we that we observed was that you
00:02:47.519 know all these years that there are
00:02:51.519 multiple challenges but the biggest
00:02:53.920 challenge that we faced as a contact
00:02:56.560 center leader was you know year on year
00:02:59.680 since 2008 I saw that technology was
00:03:02.959 changing and the frustration or the
00:03:07.280 challenges that came with deploying it.
00:03:09.599 Understanding what works, what doesn't
00:03:11.360 work, how to kind of really make or
00:03:14.159 create value out of that for my
00:03:15.840 organization was the biggest challenge
00:03:20.000 and in one particular project everything
00:03:24.159 kind of you know changed for us. What
00:03:27.519 happened was one of our
00:03:30.959 customers they needed a CRM and this was
00:03:35.840 back in 2015.
00:03:38.080 So we went to the best reputed vendor
00:03:41.360 that was out there. They came in gave a
00:03:44.720 great presentation. The verbiage, the
00:03:47.440 terminology was very impressive. And
00:03:49.680 then the proposal arrived
00:03:52.560 and what the proposal said was that it
00:03:55.680 will take 6 months and £95,000
00:03:59.599 just for the blueprint. Not even the
00:04:01.840 product. This is 10 years back, right?
00:04:04.720 I'm sure you guys have also experienced
00:04:06.799 this. And for the final product, what
00:04:09.599 they said was 18 months to build it and
00:04:12.640 half a million pounds.
00:04:15.439 Sure, the budget was a little heavy,
00:04:19.279 which was fine with us, our client. But
00:04:22.000 the bigger bigger challenge was the
00:04:23.840 timeline.
00:04:25.840 And we knew that instantly we knew that
00:04:29.360 this is not going to work for us or our
00:04:31.520 customers rather
00:04:33.840 because our customers they needed
00:04:35.840 agility and every future change meant
00:04:40.400 that we had to go back in the same
00:04:42.720 expensive but more importantly very very
00:04:45.360 very long timelines. Right? So that's
00:04:48.800 when we understood that
00:04:51.759 we need to build a team that is meant
00:04:56.160 for contact center leaders not a tech
00:04:59.840 team which is built by contact center
00:05:02.080 leaders so it has real impact but also
00:05:04.800 understands their challenges. Hence we
00:05:07.039 set up a team internally. Long story
00:05:09.600 short we built that same CRM just for
00:05:12.800 £15,000. It's been a decade, 10 years
00:05:15.840 now and still that CRM works and it
00:05:18.800 works really really well
00:05:22.240 and once we built that internal
00:05:24.080 capability
00:05:25.600 everything changed for us
00:05:28.160 and for the past 10 years
00:05:31.440 that tech team has been building these
00:05:34.160 multiple products like CRM, dialers,
00:05:36.880 workflow tools for our customers
00:05:40.400 and they provided not only the
00:05:43.199 reliability but the most important thing
00:05:45.600 which is applicable in a contact center
00:05:48.560 environment is the speed to deploy it.
00:05:52.160 And then obviously came the era of AI.
00:05:55.680 And while we did multiple projects, we
00:05:58.479 set up a dedicated good team just to
00:06:02.400 build
00:06:04.319 voice automation and conversational AI.
00:06:06.880 We started
00:06:08.960 to source the experts globally to build
00:06:12.639 that platform but we had a duty towards
00:06:15.759 our clients right we had to provide the
00:06:18.800 best that was out there not what just we
00:06:21.120 made. So constantly while building that
00:06:24.800 platform we were also
00:06:28.080 we also sourced technology from outside
00:06:31.840 and
00:06:34.080 we basically you know we we were
00:06:36.400 basically benchmarking this technology
00:06:38.960 how it is working how things are
00:06:41.360 happening and what we saw during those
00:06:44.000 tests was something what surprised us as
00:06:46.960 well that the platform which we called
00:06:50.160 CallBotics that we had built internally
00:06:52.080 was performing on use cases 20 30% times
00:06:55.520 better meaning it was resolving
00:06:58.880 you you know calls 20 30 times more and
00:07:02.400 the CSATs were also higher on the queues
00:07:04.800 where we had CallBotics and since the
00:07:09.039 goal was always been you know that we
00:07:12.720 wanted to build products that supported
00:07:14.720 contact center leaders we've understood
00:07:16.479 that it is time that we launch this as a
00:07:19.199 separate or a different company because
00:07:22.160 the other company that we had was more
00:07:23.919 like a consulting operations company and
00:07:26.479 this is like a pure SaaS company.
00:07:29.680 What I mean to say by or convey by all
00:07:33.919 of this is that when we say that
00:07:39.120 our resolution rates are higher, it's
00:07:42.319 because it's not the technology or the
00:07:45.520 platform is not just built by tech
00:07:49.759 leadership but more of the leadership
00:07:53.440 that we have within the organization are
00:07:55.599 contact center leaders and hence the
00:07:57.759 technology or the platform form is built
00:08:00.240 on small small nuances that you and me
00:08:04.240 we have seen for so many years which
00:08:06.639 makes the resolution rates so high
00:08:10.160 and that's why I'm here today to share
00:08:13.759 with you this blueprint because while
00:08:16.560 building that tech we built that
00:08:18.400 technology and we deployed these
00:08:20.400 multiple external platforms voice
00:08:22.720 automation as well we kind of made any
00:08:26.479 number of mistakes and finally we
00:08:28.720 figured out what was the best
00:08:30.960 way to deploy voice automation and I'm
00:08:34.159 sure you know a lot of you will resonate
00:08:36.958 with this at least I did for many many
00:08:39.200 years that every time a new technology
00:08:42.320 came in or a new platform came in there
00:08:45.120 was always this pressure on me to
00:08:47.040 implement it and also at the end of the
00:08:50.000 day I had to prove the value to the
00:08:51.760 leaders to the rest of the team to my
00:08:54.080 colleagues everyone
00:08:57.360 so enters
00:09:00.160 the ACE loop. This is the blueprint that
00:09:03.279 we created which has three major
00:09:07.519 parts to it. The first is assess, second
00:09:10.880 is configure and third is elevate. This
00:09:13.760 is the framework that we created or the
00:09:15.600 blueprint that we created. Assess
00:09:19.920 is basically what we talk about here in
00:09:22.959 assess is about
00:09:25.519 you assess the simplest of the workflows
00:09:29.200 first and then you know you put them up
00:09:31.279 for voice automation not the most
00:09:33.680 complex ones. The second part of it is
00:09:37.760 configure, right? Meaning generally when
00:09:41.040 we talk about configure, we think about
00:09:44.959 integrations and you know other
00:09:47.120 technology related heavy lifts. But what
00:09:49.519 we mean to say about configuration
00:09:51.200 configure is you have to configure your
00:09:53.360 AI agents and iterate them in real time.
00:09:58.240 The third thing is elevate which is
00:10:01.839 basically now that you have assessed
00:10:06.000 which workflow to automate and you have
00:10:09.519 configured your AI agents it's time to
00:10:14.160 expand those to other workflows
00:10:17.680 integrate it with your current workflow
00:10:19.920 system or current workflows and tech stack
00:10:24.079 and most importantly now that we know
00:10:27.040 what the AI agents are able to do
00:10:30.320 reliably is when we re-engineer the
00:10:33.760 processes around that
00:10:37.519 and
00:10:40.240 let's first talk a little about assess.
00:10:44.160 So with assess
00:10:47.360 as I mentioned we need to first
00:10:49.839 understand what are the easiest
00:10:52.640 workflows
00:10:54.640 which are very very much defined and on
00:10:58.240 top of that where value can be proved
00:11:02.079 immediately.
00:11:03.600 So
00:11:05.200 the thing that we have observed is a lot
00:11:07.040 of time AI doesn't fail but the
00:11:10.000 processes around it are so complex
00:11:12.959 that
00:11:14.880 the AI in the first go we take
00:11:19.040 up the most most complex use cases. So
00:11:21.279 this is what we figured with all the
00:11:23.360 mistakes that we made that we should
00:11:26.399 actually take the high volume if
00:11:28.399 possible but if not low judgment and
00:11:31.920 repeatable processes the easiest ones
00:11:35.200 where there will be the least amount of
00:11:37.360 friction. And if that is not possible
00:11:41.200 the second best option would be those
00:11:44.000 backlog or surge calls that our current
00:11:47.279 teams are not able to handle. And then
00:11:50.240 if even that is not available go for the
00:11:54.320 after hours or weekend queues where you
00:11:57.519 are understaffed or don't have staff
00:11:59.600 at all and then you go for you know side
00:12:04.079 queues if even that is not available
00:12:08.720 and what we have observed is that
00:12:11.440 in any contact center this
00:12:14.720 kind of workflows are easily available.
00:12:17.839 some of the examples if I have to share
00:12:20.320 and probably already identified some
00:12:22.560 within your workflows.
00:12:24.800 On the inbound side, we saw that these
00:12:26.959 simple lookups when people call and they
00:12:29.200 just need some information or balance
00:12:31.680 check or payment checks. On the outbound
00:12:35.279 side, confirmations, callbacks, you
00:12:38.639 know, voicemail handling. These are the
00:12:40.720 queues where we saw that not only did
00:12:43.920 the internal teams not mind it. But
00:12:47.360 with that the customers also were not
00:12:50.480 looking for too much
00:12:52.800 judgment or emotions. They were just
00:12:55.120 looking for updates. And just think of
00:12:58.639 it this way.
00:13:00.720 We a lot of times think that maybe this
00:13:03.600 will not be you know they will not be
00:13:05.440 happy with it. But what we have seen is
00:13:07.519 just that feeling for 10 years. You and
00:13:10.160 me whenever we have called a queue
00:13:12.720 either we expect an IVR a robotic IVR or
00:13:16.560 we are expecting
00:13:18.560 long wait at least 5 to 10 minutes.
00:13:23.279 With this what happens is suddenly a
00:13:25.839 customer calls and boom here you know
00:13:28.560 they are already talking to someone
00:13:30.560 immediately and because the agent has
00:13:34.079 all the information integrated within
00:13:36.320 itself it doesn't have to even read
00:13:38.399 through those documents they just have
00:13:40.399 the answers so your AHT also drops
00:13:43.120 tremendously
00:13:45.120 so that is what assess is all about
00:13:49.200 simple repeatable workflows
00:13:53.120 And then the second component that
00:13:55.519 we talked about was configure.
00:13:58.720 And when we say configure, it is about
00:14:01.279 every brand has their own unique voice.
00:14:04.720 So figuring out how it speaks,
00:14:07.760 understands the conversation and very
00:14:09.920 importantly
00:14:11.760 at what point of time should the AI
00:14:14.160 agent transfer that call to a human with
00:14:19.120 full context. That is the third part.
00:14:21.600 Escalate it and fourth iterate it on a
00:14:24.880 daily basis once you have deployed it
00:14:27.519 and you know improve the AI agent.
00:14:30.240 To give a little more understanding what
00:14:32.000 I mean to say about this is
00:14:34.800 let me give you a use case or a case
00:14:37.680 study. Let me talk about that that we
00:14:40.079 did with a customer. So they were a
00:14:42.399 benefits administration firm and
00:14:47.519 they had people calling in a large
00:14:49.680 contact center. They had people calling
00:14:51.199 in trying to understand what were the
00:14:53.760 benefits they had with their customers,
00:14:56.480 sorry with their employers, what
00:14:58.959 insurance did they have, what sort of
00:15:00.639 premiums did they need to pay, how do
00:15:03.120 they navigate the portal and any other
00:15:06.480 thing around it. So on a Friday they
00:15:10.560 reached out to us that we need a few I
00:15:14.240 think close to 80 to 100 people. We
00:15:16.959 said that we cannot get you and they
00:15:18.480 wanted to deploy it in 2 days. So we
00:15:21.120 said that we could not do it. We could
00:15:22.800 not get those many people in 2 days and
00:15:25.600 then we said that you could but use the
00:15:27.839 platform and they were always hesitant
00:15:31.040 to do so but once they deployed this now
00:15:35.600 they had no choice right. So they said
00:15:37.519 that okay let's go ahead and deploy it.
00:15:39.279 So on a Friday they came to us they
00:15:41.839 uploaded all the
00:15:44.560 training material they used for their
00:15:47.839 human agents to be trained. Monday the
00:15:51.120 AI agents were live.
00:15:54.160 The AI agents started taking calls. 45%
00:15:57.440 of the calls on day one they were able
00:15:59.680 to manage
00:16:02.079 and then throughout the week from
00:16:05.759 live calls they kept
00:16:08.880 fine-tuning those AI agents just after 5
00:16:12.720 days on such a complex use case they
00:16:15.600 were automating close to 80% of the
00:16:17.920 calls their AHT obviously had dropped
00:16:21.279 their CSATs were higher just within that
00:16:23.519 one week because they had these active
00:16:25.360 surveys that they were taking over
00:16:27.120 the calls and in general maybe they did
00:16:29.839 not use to earlier but now they were
00:16:32.160 because they wanted to understand if
00:16:33.680 their CSATs or their customer
00:16:35.440 satisfaction was taking a hit live but
00:16:37.759 what they observed was no that
00:16:40.320 their customer satisfaction grew over a
00:16:43.440 period of time over that one week
00:16:46.720 now that the AI agents were configured
00:16:50.320 and they had a lot of confidence right
00:16:53.199 is when you know
00:16:55.600 the leadership was convinced that there
00:16:57.759 is value in this and more importantly
00:17:00.000 the human agents they also saw the value
00:17:02.079 that they now did not have to take care
00:17:04.640 of repetitive simple query
00:17:08.079 based calls but they could actually
00:17:10.160 focus on the calls that really matter
00:17:12.480 where they could actually create value
00:17:14.880 and have more satisfying jobs. So next
00:17:18.319 what they did was they started expanding
00:17:20.480 their
00:17:22.079 to different workflows.
00:17:24.720 Now over a period of time they saw that
00:17:27.199 within that workflow 84% of calls were
00:17:30.480 managed by the AI agents and because of
00:17:33.760 the location that they were based in
00:17:35.600 they saw 78% reduction in their cost as
00:17:39.280 well. So now they shifted to more
00:17:42.400 complex workflows. And in one of the
00:17:46.320 workflows I remember they had 183
00:17:49.919 documents which were basically guides,
00:17:52.640 product info, SOPs, catalogs which
00:17:55.679 would give these AI agents which used to
00:17:58.400 give their human agents the information
00:18:00.960 that were required to resolve those
00:18:02.960 queries and calls right and
00:18:07.360 what happened after that was they kept
00:18:10.160 you know going to the next next use
00:18:12.320 case. So this is what elevate is. Elevate
00:18:16.080 is around expanding that and then they
00:18:19.520 started also re-engineering the
00:18:21.280 processes. Now that they knew
00:18:24.160 what the AI agents could
00:18:27.919 reliably handle, the next step was how
00:18:31.200 do they rework the entire operations
00:18:33.760 around it? Because now they like for
00:18:36.559 example they understood that within this
00:18:38.400 queue 95% of the calls are automated. So
00:18:41.600 they moved those people to another
00:18:43.280 queue. They needed a few people to
00:18:45.760 orchestrate the automation. So they
00:18:47.520 started retraining them. They knew that
00:18:51.200 they need to divide a few queues into
00:18:53.360 sub-queues to be able to automate at a
00:18:56.559 higher percentage and that's how they
00:18:58.240 got to that 84%. So that's what they
00:19:00.480 started and the beautiful thing about
00:19:02.480 this ACE loop is that it is a constant
00:19:05.919 framework. You go from one workflow to
00:19:08.640 other workflow, one queue to other queue
00:19:11.280 assess
00:19:13.440 simple workflows. assess you know what
00:19:16.400 can be automated what are the metrics
00:19:19.039 that show you that results are delivered
00:19:21.679 CSAT is increasing the AHT is decreasing
00:19:24.880 those kind of things and then obviously
00:19:26.799 you know configure those AI agents for
00:19:29.520 those particular use cases and queues and
00:19:31.919 then obviously you know keep expanding
00:19:33.679 them so that's they kept repeating it
00:19:36.640 repeating it at this point of time
00:19:38.400 earlier they had adopted this just from
00:19:42.480 a thought process that
00:19:45.919 they need to just from thought
00:19:48.480 process that they need to you know take
00:19:50.400 care of a burning firefighting situation
00:19:53.520 but today they have deployed it across
00:19:55.679 everywhere and their agents and
00:19:57.679 leadership couldn't be happier.
00:20:01.200 So to sum it up, what we have observed
00:20:03.520 in this 17 years making mistakes while
00:20:06.000 deploying and creating this AI agents is
00:20:09.360 that the contact centers that win won't
00:20:12.880 be the most automated.
00:20:15.360 They will be the most that are adaptive
00:20:17.360 because every contact center, every use
00:20:20.080 case, every workflow is very unique and
00:20:23.440 that we have to keep in mind and
00:20:26.320 accordingly start adapting it because if
00:20:29.520 not today, we have to do it tomorrow. I
00:20:32.000 remember when we started off and IVRs
00:20:34.559 were the new thing, we ourselves, me
00:20:37.200 myself as a CEO of the company, I pushed
00:20:39.520 back for some point of time and then you
00:20:41.840 know we had to lose a lot of customers
00:20:43.760 because eventually everyone like today
00:20:45.760 IVR is not no more like a new thing,
00:20:48.720 right? Same way what we have seen
00:20:51.840 with our customers is that AI agents is
00:20:55.600 not a new thing to most of them right
00:20:58.000 now. they're just automating as much as
00:21:00.320 they can.
00:21:02.880 The other thing about is that within
00:21:06.320 these 30 minutes obviously there is this
00:21:08.320 very basic level or you know at a
00:21:11.280 superficial level I can share regarding
00:21:13.120 the blueprint right and as a contact
00:21:16.320 center fellow contact center leader I
00:21:18.720 would want you guys also to spend more
00:21:20.880 time with your families not you know
00:21:22.799 firefighting staying on calls so even if
00:21:25.840 you work with someone else or you know
00:21:29.039 you already have or are still not
00:21:31.600 thinking of deploying AI agents. But
00:21:36.960 what we will do for you is if you stop
00:21:39.120 by at our stand M6 M66, we can, you
00:21:42.960 know, create a specialized, tailored,
00:21:46.240 detailed document that will resonate
00:21:49.919 with your workflows and you can
00:21:52.960 basically, you know, take that back to
00:21:55.280 your team and automate or at least start
00:21:59.280 evaluating what you can, you know,
00:22:01.280 automate when it comes to your voice
00:22:04.080 queue.
00:22:05.120 And
00:22:07.760 that will just be you know give you a
00:22:09.919 more resilient faster and more cost
00:22:12.240 effective and more happier agents
00:22:15.600 operations.
00:22:17.120 Thank you so much.
CallBotics is the world’s first human-like AI voice platform for enterprises. Our AI voice agents automate calls at scale, enabling fast, natural, and reliable conversations that reduce costs, increase efficiency, and deploy in 48 hours.
For Further Queries Contact Us At:
