On and off today I’ve been pushing forward my learning with various SaaS chatbot creator tools.
The tools in question
I’m definitely further along the road with Landbot, which I rebuilt the miggle website in back in March, purely because I’ve had more experience with it.
Alongside that, Ayce in our team is working with an open source solution called botpress.io, and I’m looking forward to reviewing his progress on that later this week.
Ironic point #1
One interesting thing about the three SaaS products I’ve been looking at is that they all use Intercom for support, and make good use of throwing back automated responses based on triggers in user input before a human gets involved. Given that this path seems to be one of the most obvious use cases in investing in messaging style products to triage issues, saving reps time, I find it ironic that each of these providers has to rely on what is arguably a competitor to deal with the sort of questions that many people might be asking, as they aim to build a process that does what Intercom does, rather than being able to use their own platform. I’ve not taken a lot at Intercom yet, but it’s a no brainer that I do. (I should maybe look at drift too.)
Ironic point #2
So for now, I’ve been splitting my time amongst the three aforementioned. So what’s been determining how much time I spend on each? Well, simply put, I’ve kept cracking a way at one platform until I get stuck to the point where the online help documentation doesn’t help me, such that I need to ask a question of a human. Not a bot. And because us humans can’t answer 24/7/365 with zero lead time, while I wait, I move onto the next one until I get stuck. And then so on.
Where have I made the most progress?
Well, today I’ve made the most progress on Chatfuel, although right now I’m waiting on responses or additional info to move both Chatfuel and Clustaar on. I’ve no problem with that, because the inevitable time it takes to get a considered response is not causing me any issues.
Tomorrow, as a business, we will make the most progress with Landbot, because a new-ish feature, the ability to link to another bot, is for me a bit of a game changer in terms of how we can re-purpose blocks of content. What’s more, I’m at the stage in my knowledge as regards Landbot where I can easily delegate that work to Jess to do, allowing me to focus on the other two products (or review botpress, or look at Intercom). That said, while this Landbot feature is great, what I’m struggling to see in any of these products is how I can build multiple bots from a library of intents, stories, triggers, actions and routines I build. My guess is that I’ll find botpress more useful there, and, as is often the case, we’ll be back to that old chestnut of trading off the power and flexibility of open source versus the ease of use of SaaS. (Of course, in general, open-source is better for Business Continuity than SaaS, but that assumes a solid, well distributed and community supported bit of open source — and I don’t know if botpress is that yet.)
And what do I need to move chatfuel and clustaar forward?
With both (and to be fair clustaar have given me some guidance here) I need to be able to make more progress with decision logic. i.e, Do you want this? Yes/No, or, Which of these options are useful?
This is so much easier with Landbot…
…but then, what the last week has underscored for me is that Landbot is not really chatbot creation technology like what the others are. It’s a decision tree tool, in a pseudo-chat front-end. But the extent to which it can triage a funnel of incoming requests effectively, allowing an employee to step in at the right juncture, for now, will provide an appropriate solution for what most businesses might want from chat, at an early stage, and lowish cost, if they want to experiment with what they might do here.
What I don’t get yet
Why or when Facebook/WhatsApp will launch it’s own platform for people to build these sort of bots (or, more likely, acquire someone). There’s a huge technical integration side of things between these SaaS products and Facebook/WhatsApp that I don’t know enough about yet. There’s also the whole issue of trust. I’d probably rather have a product I can use on top of Google Home, WhatsApp, Facebook, Alexa and natively on my own website than have something else in Mark’s walled garden. I like Clustaar for this. I think it’s why I’ll like botpress too. (Here’s an interesting view on whether Facebook can be trusted — by Landbot’s CEO!)
We’re only human after all
Last Friday, Clustaar, at the end of a long week, gave me a priceless belly laugh when I dropped in early in one of their open webinars. I must stress, it was laughing with them, not at them, and as such it made my day. It showed me we’re all human. Reminded me, that whatever we often do workwise, we hit the same challenges and frustrations — and being able to recognise that in each other, and have some empathy, is what often drives us to want to work more closely together and support each other. Which is important. Until the machines take over. (And of course, when they really take over, then we’ll need to remember how to stick together anyway….)