#announcements

We've been very careful as we evaluate this, because we need to be confident whatever tools we leverage:

1. Have sufficient security measures in place, and
2. Won't use sensitive data to train their models I don't think people realize how much sensitive stuff can get disclosed in a call and retracted later.

At the moment, we've largely leveraged AI for tasks outside of creation: outlining, process docs, etc. -- and in some projects, we've used it for custom question set generation at scale, provided with some very granular context.

We're exploring how we might leverage AI to support storytelling and repurposing, but again, caution comes first because LLMs have a propensity to hallucinate, invent things, and emphasize the wrong elements of a story -- or to key in on weird aspects of context.

So... we're experimenting, developing our policies, and when we feel confident we've found what works to both accelerate our process AND improve the output/experience for clients, we'll roll it out.

It's interesting to see how software-driven solutions like UserEvidence or Laudable are leveraging AI at the moment (e.g. ingesting from Gong/customer feedback and spinning it into a story); there's a lot of potential there, but without first-hand experience it's hard to say how accurate, human, etc. those things are. From what I've seen, they're a solid shortcut to high-level stories; it's exciting to imagine\ where it might go, and there's some very cool stuff you can do in terms of isolating and pulling clips that I think is hugely valuable.

I've also tried tools that try to go customer-facing and have them chat/speak to the bot to produce the story, and found the overall experience jarring/not great for a part of the process that is meant to be very human, but... early days.

One thing I can say with relative confidence is that we're a ways away from doing anything without strict human oversight. Ai just isn't nearly as good as keying into the strategic/human elements of these stories as our team is, so there's no reality I see near-term where we just turn a bot loose and let it roll.

Hey Emily! Currently unemployed due to a recent RIF, but before that we were not using AI for writing customer stories. All story writing is still written by our content marketers and briefs and interviews done by me.

My Marketing Manager at the time had begun using AI to get started on her blogs, but would edit them heavily.

Hi Emily! We have not yet integrated AI into our customer story creation. However, we are as a company focusing on AI in product development. Within our advocacy team, however, we don't leverage AI - perhaps Chat GPT every once in a while to get some inspiration on different ways sentences can be constructed for customer outreach.

Looking forward to learning here how others use or anticipate the usage of AI!

Ditto what @Camille Shortridge said on our end. 🙂

going to echo what others have said in that I have not used it at all yet when it comes to helping develop customer stories. I do see a future where eventually AI will have the capability to produce usable, high-fidelity assets merely by listening to a customer interview. That being said, I'm skeptical as to whether the sort of security guardrails needed to feel comfortable giving that access will ever exist with AI.

I feel like I'm also not alone here in really deeply believing in the value of the genuine human element in storytelling, and don't know if I'll ever feel good about giving that over to AI

The most that our team is using it right now is to organize our call notes and improve our call log transcripts 😅 (as in, Microsoft Teams has gotten better). All of our content is always written, edited, and reviewed by humans. I think there is a lot of interesting potential, but I think that potential lies in making the work easier (with more intelligent recordings, transcripts, etc.), not passing the work to AI entirely. As said above, the nuance of storytelling is so innately human its hard for me to imagine a future without a human in the writer's chair.

Similar to what @Alexie Glover is saying, at Zoom, we use our own AI assistant for meeting summaries of introductory calls and interviews for customer stories, specifically the "Meeting Summary with Zoom IQ" seen here https://explore.zoom.us/en/ai-assistant/ - I also do what @Camille Shortridge suggested and use ChatGPT to improve my outreach emails, but I won't put any customer data (like an interview transcript) into ChatGPT.

We recently started using a tool called SummarAIze - it's pretty unreal. It allows us to take our interviews, webinars, podcasts etc. with customers and transition them into written studies and blogs super quickly. It also generates social copy, email copy, gives you time codes for key moments to pull out and use in your blog post, on social, however you would want a snippet.

It's really a game changers and allows you to use a webinar or recorded interviews as the foundation content and then the written side of the content is generated with maybe the best accuracy I've seen so far when using AI.

And it's CHEAP, $15 per 1hr of recording. You upload the MP4 and it generates:
• Transcript
• Twitter Thread reco
• LinkedIn Post
• optimized YouTube title and Desc
• Blog post (could also be used as a case study)
• Optimized title and highlights
• Email content
• Bios if part of your recording included an intro from your speaker
You do still need to edit a bit but it saves some serious time and allows you to rethink you process for creating content. Agreed with @Joel Klettke, we certainly would never let the tool create content and publish it unedited or reviewed by necessary parties. Maybe one day?

Echoing everyone else, GPT can be useful to give you more ideas quicker.

@Rob Ayre How do you ensure your recorded interviews and content don't become part of the overall ecosystem? The security issue is a big factor for us.

@Laurie Timms not to speak for Rob, but so far the only (laborious) way we've found to explore this is to review the privacy policies of the various APIs involved, e.g. this is what fuels SummarAIze

So far we've been testing it with only non-sensitive content. Anything more we would do a need dive into the privacy policy

Got it.

I’m ready to pay AI a salary $$$

I've used it to turn a transcript into a case study outline with some iterative prompts, with people and company names swapped out for privacy. (I'll say something like "the interviewee is Fake Nameski at ACME corp".)

@Angela Ferrante your new gong integration might weigh in here with using AI for customer stories 🙂

Yes, 100%! Doesn’t take the place of human writing, but using AI we get great “shitty first drafts” using sales + CS call recordings on various customer story asset types using call recording data - ranging from an outline of a case study to short-form pocket stories (my fave). This avoids starting from scratch with a customer interview bc a LOT of the data we ask in case study interviews actually already lives hiding in call recordings.

Angela and Ryan ran a test for me with this amazing tool before I was released back into the wild from Allbound and I was super pumped about it. Highly recommend chatting more with her about it if this peaks your interest.

Sounds really exciting!

We started to use AI to extract transcripts, to draft a story overview and to pull quotes from conversations. Our writer still writes all of our customer content and we don’t anticipate this to change.

@Valeria Gomez would love to hear more! Will send you a dm.

We're expanding our use of generative AI (a mix of LLMs including GTP4 but also others) more in the UserEvidence product. It obviously does not replace the authentic voice of the customer - but here are a couple things that it is useful for:
1. summarizing/analyzing large data sets of customer feedback to create more of a narrative/analysis/report that highlights the key themes
2. turning static feedback surveys into more of a "conversation"
There's a couple more I can't share yet but they are pretty wicked 🙂

So yes, it will speed up story capture, creation, derivation, packaging, delivery, design - but ultimately the quality of the story will still come down to the interestingness of the story in the voice of the customer.

Would love to have a workshop on this topic with members of the community.

@Joel Klettke I'm with you! Lots of potential, but it's early days.

@Rachel Ward I'm hearing similar things. AI to get started on blog posts, but not yet for customer stories.

@Camille Shortridge @Laurie Timms There's no shortage of curiosity, that's for sure! 🙂

@Jake Sotir Agree about the security piece. And I also feel so strongly that we need humans to tell human stories - but somehow I feel like I'm not really allowed to say this because this is my livelihood.

@Alexie Glover Yes to having AI do more menial tasks. And yes to having people engaging with people to tell meaningful stories.

Like everyone else, we’re using it mostly for summarization and grammar/tone smoothing.

In the future, though, I anticipate using it to identify, select, and infuse existing customer evidence into other marketing assets like what demand/content or product marketing teams might create.

@Rob Ayre Very interesting. Are you open to sharing an output from SummarAIze?

@ That's interesting. And has the outline been accurate and helpful? Has it saved you time?

@Evan Huck I can't wait to hear more about what you've got up your sleeve.

@Jeanne Talbot I'd be interested in a workshop too. Anyone feel like leading it?

@Emily Amos @Jeanne Talbot I'm happy to host it later in august or early sept (we're doing a big launch in mid august)! It was a big part of our fundraising narrative - so we've learned a ton about it from experts in the space as we weave it into the fabric of UE.

@Mary Green (Owner CMAweekly) could we piggy back off of one of your Luma calls later in the month or in early Sept? I can also just pick a date and use our own Webinar infrastructure if that's easier

Absolutely!!

cool, thanks - aug 25th during your 11 AM PT/2 ET slot would work for me if that works for you!