Captions is an AI-powered creative studio that enables millions of creators to produce studio-quality videos with just a few taps or clicks.

Scaling support for millions of users

Delivering a great customer experience is core to Captions' business, and providing 24/7 customer support across multiple channels (in-app and email) is an essential part of this.

As Captions' customer base is growing fast, keeping up with the support inquiries by hiring new support team members is nearly impossible.

As an AI-first company, Eli Winderbaum, Head of CX at Captions, started looking into AI solutions that could help him and his team handle the ever-growing number of support tickets.

Customized but easy-to-implement

When exploring potential AI solutions, Eli and his team had 4 requirements:

1. Little or no engineering lift: The solution had to require little or no engineering lift, as the team had limited engineering resources to allocate to the project.

2. Multilingual: The solution had to understand and reply in multiple languages to support customers who speak more than 50 languages.

3. Quick to adapt: Captions continuously adds new features to its iOS, Android, and desktop apps, so the solution had to adapt quickly to changes in our help center and agent responses.

4. Knows when not to answer: The solution had to be reliable, which included knowing when to answer and when to escalate the ticket to a teammate.

Compromises in current solutions

After reviewing all dominant AI solutions on the market, Eli and his team saw two potential compromises:

1. Add Intercom Fin to their already exciting Intercom setup. This required no engineering lift but would compromise customization and response quality, not to mention that it could only answer customers in English.

2. Buy and migrate to a separate chatbot solution. This would allow more customization and better responses but would require an engineering lift in the form of new SDKs and integrations.

This was until we shared how we can highly customize and implement a Parahelp agent without any engineering lift.

"The kicker for me is that it had zero engineering lift while still allowing high customization as it works inside our existing support solution," says Eli.

Results from implementing Parahelp

From we started working together, it took 7 days to fully customize and implement the Parahelp in Captions' support setup.

7 days after implementation, the Parahelp agent was already resolving 46% of all Captions support tickets end-to-end.

Since implementing Parahelp, Captions has also seen significant improvements in CSAT, while time to first reply, time to resolution, and cost per resolution have all decreased.

Eli says, "Parahelp has transformed our CX org overnight. Besides resolving tickets 24/7 in all languages, I'm amazed at how indistinguishable the Parahelp agent is from a human agent. So far, I can count the number of customers who asked if it’s a bot. This also includes many L2 tickets, which is truly incredible.”