Most of the Agent Work Happens After Go-Live

Lots of people seem to expect that agents and automations will behave like any other software implementation. You buy the tool, you implement, you do the training, and you're off to the races, right?

In my experience implementing these things, the vast majority of the work is after the first go live.

There's so much to tweak, tune, and evaluate to get to good results. I've been humbled many times now as something that seemed straightforward surprised me in the opposite direction. Some of my favorite moments of the last year or two have come after complete false starts that eventually turned into something really great.

I think this has been one of the biggest barriers to true adoption - people see others raving about what they can do but they try something themselves and get mediocre results. Even after some training. It can take real time, effort, and reps to get to the true "eureka!" moments.

Right now when the interfaces are still not close to good enough, it can take a lot longer than it feels like it should. You might need to summon the patience of a kid trying over and over again at a video game to have your own breakthrough.

Even a month or so ago, I was really impressed with something I built for a content marketing use case. It was like having a software pipeline but for content! Then they tried it and it completely stomped over the important parts of the creative process and produced lukewarm garbage. Back to the drawing board.

But working more closely with a true expert, I was able to set up a system that helped them take the grunt work out of making their content accurate, well sourced, and on brand without steamrolling all the ways their expert human touch made it more engaging and worth seeking out.

This stuff is not as easy as people make it seem - unless you're ok with putting lukewarm garbage out. It's also approachable by anyone if you're persistent enough. Keep at it and you'll see results too.

Join the conversation on LinkedIn