Operator by Intercom — The bot for better customer experiences

@chrismessina Damn, Chris. Ouch. This really hurts. 😞 In my opinion, "shitting on bots" is an unfair description of our position. Our whole thesis is that businesses will benefit from being more personal and human with their customers. We're trying to fight transactional, fake, spammy crap. We've been skeptical that bots as they have been implemented thus far would not actually exacerbate these problems. That post you linked lays out part of that thesis, but also acknowledges that we do think bots have value: "Bots will augment conversations between humans … will help answer simple questions in win-win scenarios". This pitch for Operator is an extension of that thesis. Yes, it effectively says "our bot is better" (at some things, specifically the customer experience), but that's marketing. We need to make the case for why people should use our stuff over others'. And we feel like our pitch is actually honest. We're not trying to pretend our bot is anything it's not. We're actually specifically trying to position it as a limited bot. Sorry we pissed you off so much.

@chrismessina the article you listed has a somewhat strong tone in the headings, but accounts for it in the actual details: for example, saying that bots are good at repetitive tasks right now and will get more powerful over time. I wouldn't necessarily call that "shitting on bots for ages." And what's wrong with the word better? They didn't say "Hey our bot is better than all other bots," they just said "a bot for better customer experiences" -- very different. I think your tone comes off as dramatically more arrogant than theirs. But I might not have the context you have, or perhaps there are more articles worth referencing you didn't link.

@chrismessina wow Chris! There is no 'corporate' here 🙂 Just a bunch of people trying really hard to make something people find useful. Our launch today is the direct line execution of that post you linked that I wrote last year. I'd love to understand where you think the disconnect is. The thesis of the post: Humans are better than bots at some things, bots are better than humans at others. Let's understand that and work with it. Our launch today imo is that same idea in a product, rather than merely in writing. ps for others Chris and I used to work together 🙂

@chrismessina fwiw, I totally agree with Intercom's stance on the type of chat bots we're starting to see. They're just telephone menus dressed up in poor natural language processing. Operator doesn't really fit in with the modern crop of bots—I probably wouldn't call it a bot at all. It seems more like a third-party assisting the user than a first-class participant. As it should be

@eoghanmccabe @padday normally I'm more measured in my tone and response, but I like Intercom (generally) too much to feel let down in this way. I'll own my inarticulation of why Operator ticked me off so much, so let me try again... First, for everyone else, I recorded a podcast for Intercom a while back while I was at Uber in which one of my theses was capably promoted: The conversational commerce hype is real. I have what I would consider a good relationship with the folks at Intercom — like @eoghanmccabe @padday. A good relationship is one where you can bring up things that you disagree with, respectfully, and talk things out (among other more positive things). Second, when I talk about "shitting on bots", I'm referring to this kind of stuff: "Much of the excitement around bots focuses on one of two things: [distribution and removing expensive humans from customer communication]." "So if bots can’t read between the lines and make us feel better, what are they actually good for? Turns out that with today’s and tomorrow’s technology, they are good for simple, low-level, repetitive questions and tasks. That’s it." — @padday -- "Every year in design circles, there’s a buzzword. This year, the buzzword is without question bots." "...When we experimented with building a bot, we gave [it] a name. [...] People hated this bot — found it off-putting and annoying. It was interrupting them, getting in the way of what they wanted (to talk to a real person), even though its interactions were very lightweight." —@EMCGUANE -- "Text-based messaging works best with a human on both sides of the conversation" —@HUGHDURKIN -- "Despite plenty of excitement it’s still unclear how conversational UIs can be made to work in a practical sense." —@THOUGHTWAX And then in today's announcement: "Bots are still a long way from competing with humans. When it comes to context, sentiment, subtle knowledge and big picture thinking, humans will continue to triumph for a long time to come. At Intercom, we are completely ok with that. We are betting that bots will augment, rather than replace humans in the long-term." —@MARK_V_RYAN I can get on board with that sentiment, and indeed, it aligns with much of what I've been saying for a while. It just feels like a bait and switch... like there's been this reluctance to admit or accept that bots are critical to Intercom's future, or to scaling Intercom's customers's needs, and that anything but human customer service reps on the other end of an Intercom chat thread is a fail (not to be *too* cynical, but Intercom charges on a per-seat basis, so the more human reps a company has, the more money they make). I mean, I know that Intercom takes a long time with their product development strategy, so it's not like this product just came about spontaneously. As they were penning the articles that contained the above quotes, they were working on this. Perhaps it's less their perspective that I disagree with than their disengagement with the broader bot and conversational product ecosystem. And now that they've launched their own bot platform, I'm just like left agast — yes, I agree with so much of their sentiments about the product experience of automated conversational experiences — but couldn't they have been more supportive and conciliatory along the way to encourage the dozens or hundreds of companies that have started in this space? Couldn't they have taken a more active role in defining this space? Maybe my expectations were set too high, or were too self- or industry-centric. I want to embrace Intercom Operator, like I've done with nearly all of Intercom's other products... I'm just lamenting how aloof they've been so far, and how it feels like their perspective was always Intercom-centric, instead of looking broadly at this shift, and guiding the industry forward, rather than sandbagging their plans until this moment. I'm not really that upset, I think I'm just reacting to the form over the substance. Operator is a good thing for my mission, FWIW (humanizing the experience of using technology). I just wish that the path to arrive at this moment had been different.

@mattnhodges will the skills be built only by intercom or will some kind of SDK/framework made available for teams to build their own skills?

@ugo_alves I'd be lying if I said we haven't thought about that ;) Today is step one. We'll get there. Stay tuned 😀

@mattnhodges It looks like the bulk of the features already existed in Intercom. Expectation setting, contact information collection, intelligent educate articles suggestion, customer satisfaction, and github integration pieces were already parts of intercom, just not branded as "Operator", and not coming from a "bot" in the conversation experience. However, I'm really excited and looking forward to the upcoming "Integrations" and "Skills management" features 🙌

@theoblochet intelligent article suggestions in the Messenger and the ability to close conversations on your behalf are brand-new :)

@theoblochet check out the quick video here for a quick overview https://app.intercom.io/a/change...

Drift allows reverse IP lookups to customize the initial message with company name — do you guys support this too, or have plans to do so?

@alexdbauer I've experienced that myself. This really is just day one for Operator. We're already looking to understand ways to help you make your messages be more personal when it’s appropriate. Thanks for the suggestion :)

@stevenlu you're right, Steven. Some of Operator's skills were existing capabilities of our Messenger (e.g. set expectations, collect contact info) while others are entirely new (e.g. conversation ratings, article suggestions, close conversations). Operator ties all these things together with a bunch of logic about how and when it should invoke those skills. There's a lot going on under the hood that you can read more about here: https://docs.intercom.com/interc... While we would have loved to have nailed every problem on day one, we can't do everything. I'm sure you know what that's like :) Operator is only 5 hours old. It's still a baby, so please stay tuned. We'll get there.

@mattnhodges understandable, maybe it was just me thinking that some of those features should have already existed? Don't get me wrong, still glad that they are there. Just making sure that you guys have something in the works for the visitor part cause that's the biggest pain I have right now.

@stevenlu good to know, mate. I am pretty sure you're not the only one :)

@cancom10 brilliant question, mate. I'm not the best person to answer. Keep your 👀 out for a reply from our product team–I won't do them justice :) They are over in Dublin, so it might be tomorrow before they get back to you.

@mattnhodges @cancom10 Thanks for the question, Can! I'm Kevin, I helped build some of the logic behind how Operator behaves. This is something we're actively thinking about right now. Currently, we have measures of success for each individual skill. For example, with Article Suggestions, we look at everything from whether a person actually interacted with suggested articles (weak signal) all the way to parsing their subsequent reply (potentially strong signal). Some of these signals give us a virtuous feedback loop that we can use not only to measure success, but train the underlying recommender system we've built that actually makes the article suggestions. On a more holistic level, we are getting a ton of great feedback from our customers already and are using that as a barometer for helpfulness of Operator. We hope that all these signals tied together will give us a good sense of where to go next, and how to make Operator better for our customers and indeed our customers' customers! Hope that answered your question, let me know if not! 🤓

@zachtratar nah. No one wants a lawsuit.

@zachtratar exactly, pretty sure these guys knew about Operator, and if they didn't wondering how they will respond to this.

@chrismessina of course no one wants a lawsuit, but if companies don't prosecute their name it becomes fair use and they lose it for perpetuity. If a lawyer was making a recommendation and Operator was wanting their name long-term, a lawsuit would likely be recommended.

@awelfle Thanks Andy! And thanks for being so supportive in all the language-related discussions we've had about this along the road. Content strategy community FTW :)

@florijhehdb intelligent article suggestions in the Messenger and the ability to close conversations on your behalf are brand-new :) As is all the logic that runs in the background to determine when Operate invokes a skill. Check out the video here for a quick overview: https://app.intercom.io/a/change...