Zelta AI

We've seen a surge in demand for Zelta from companies deploying customer support AI with products like Intercom's Fin, and it's for an interesting reason... Customer support plays two crucial roles for software companies: • Handling and resolving customer issues • Providing insight to product teams about broken, confusing, or missing functionality Fin excels at resolving customer issues, but product teams can't hold monthly meetings with Fin or drop by its desk to understand where customers are struggling. This creates a problem. Without support teams acting as the bridge for customer struggles, product teams are left in the dark or forced to sift through closed tickets themselves. Zelta AI addresses this by ingesting conversations Fin has with customers and quantifies customer struggles for product teams With the power of AI, product teams effectively gain an army of analysts capable of instantly running complex queries and analyses across support data. For example, our customer have been running queries like: • "Since we launched the new survey builder, has there been a reduction in support issues related to surveys? If so, by how much.." • "What were the most frequent issues in the support channel last month? Provide a breakdown by product area, feature, and issue, including support conversations as backup for each." These are questions that would have been a burden on support teams to answer in the past and impossible to answer if there is no support team Interested to hear about others experience switching to Fin / customer support AI. Let me know in the comments 👇

AI summaries of customer feedback are boring and useless The problem with AI summaries is… that they summarize. Averaging everything you’re summarizing into a boring grey goo For example, feeding Spotify app store reviews to GPT gives you something like: Spotify app reviews reveal a mix of praise and criticism. Positive Feedback: • Music Discovery: Users appreciate Spotify's music discovery features and recommendation algorithms • User Interface: The interface is praised for its ease of use, making it simple to find songs and create playlists Negative Feedback: • Ads and Free Tier Limitations: Users on the free tier report frequent ads which can be frustrating  • Subscription Issues: Some users experience difficulties managing subscriptions • Content Recommendations: Some users feel the algorithm suggests unwanted genres This is boring/ useless… The purpose of a summary is to hit highlights, not to homogenize everything. A better version might look like: 𝐔𝐬𝐞𝐫𝐬 𝐚𝐫𝐞 𝐡𝐚𝐯𝐢𝐧𝐠 𝐢𝐬𝐬𝐮𝐞𝐬 𝐰𝐢𝐭𝐡 𝐬𝐡𝐮𝐟𝐟𝐥𝐞 𝐦𝐨𝐝𝐞 𝐚𝐧𝐝 𝐬𝐨𝐧𝐠 𝐪𝐮𝐞𝐮𝐞 (𝐫𝐞𝐩𝐨𝐫𝐭𝐞𝐝 𝐛𝐲 𝟒𝟎𝟎 𝐮𝐬𝐞𝐫𝐬 𝟒% 𝐨𝐟 𝐭𝐨𝐭𝐚𝐥) "𝑟𝑒𝑐𝑒𝑛𝑡𝑙𝑦 𝑖𝑛 𝑠ℎ𝑢𝑓𝑓𝑙𝑒 𝑚𝑦 𝑠𝑜𝑛𝑔𝑠 𝑤𝑖𝑙𝑙 𝑓𝑟𝑒𝑒𝑧𝑒 𝑎𝑓𝑡𝑒𝑟 𝑎 𝑓𝑒𝑤 𝑠𝑘𝑖𝑝𝑠"  "𝐼 𝑚𝑎𝑘𝑒 𝑝𝑙𝑎𝑦𝑙𝑖𝑠𝑡𝑠, ℎ𝑖𝑡 𝑡ℎ𝑒 𝑝𝑙𝑎𝑦 𝑏𝑢𝑡𝑡𝑜𝑛, 𝑎𝑛𝑑 𝑡ℎ𝑒 𝑠𝑎𝑚𝑒 4 𝑠𝑜𝑛𝑔𝑠 𝑝𝑙𝑎𝑦 𝑜𝑣𝑒𝑟 𝑎𝑛𝑑 𝑜𝑣𝑒𝑟 𝑖𝑓 𝐼 𝑑𝑜𝑛’𝑡 𝑟𝑒𝑠ℎ𝑢𝑓𝑓𝑙𝑒 𝑚𝑦𝑠𝑒𝑙𝑓" "𝑇ℎ𝑒 𝑞𝑢𝑒𝑢𝑒 𝑖𝑠 𝑏𝑟𝑜𝑘𝑒𝑛 𝑓𝑜𝑟 𝑚𝑒. 𝐴𝑑𝑑𝑒𝑑 𝑠𝑜𝑛𝑔𝑠 𝑎𝑟𝑒 𝑟𝑒𝑚𝑜𝑣𝑒𝑑" 𝐈𝐬𝐬𝐮𝐞𝐬 𝐰𝐢𝐭𝐡 𝐜𝐚𝐫 𝐦𝐨𝐝𝐞 𝐦𝐚𝐲 𝐛𝐞 𝐜𝐚𝐮𝐬𝐢𝐧𝐠 𝐬𝐚𝐟𝐞𝐭𝐲 𝐢𝐬𝐬𝐮𝐞𝐬 (𝐫𝐞𝐩𝐨𝐫𝐭𝐞𝐝 𝐛𝐲 𝟖𝟎 𝐮𝐬𝐞𝐫𝐬 𝟎.𝟖% 𝐨𝐟 𝐭𝐨𝐭𝐚𝐥) “𝑁𝑜𝑤 𝑒𝑣𝑒𝑟𝑦 𝑡𝑖𝑚𝑒 𝐼 𝑛𝑒𝑒𝑑 𝑡𝑜 𝑓𝑖𝑥 𝑚𝑦 𝑒𝑞𝑢𝑎𝑙𝑖𝑧𝑒𝑟 𝑐𝑢𝑧 𝐼’𝑚 𝑜𝑛 𝑡ℎ𝑒 ℎ𝑖𝑔ℎ𝑤𝑎𝑦 𝑎𝑛𝑑 𝑤𝑎𝑛𝑛𝑎 𝑡𝑢𝑟𝑛 𝑢𝑝 𝑡ℎ𝑒 𝑣𝑜𝑙𝑢𝑚𝑒 𝐼 ℎ𝑎𝑣𝑒 𝑡𝑜 𝑑𝑖𝑠𝑎𝑏𝑙𝑒 𝑐𝑎𝑟 𝑚𝑜𝑑𝑒 𝑓𝑖𝑟𝑠𝑡? 𝐻𝑜𝑤’𝑠 𝑡ℎ𝑎𝑡 𝑠𝑎𝑓𝑒𝑟" “𝐼 𝑤𝑖𝑠ℎ 𝑡ℎ𝑒 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑠 𝑤𝑜𝑢𝑙𝑑 𝑠𝑡𝑎𝑦 𝑖𝑛 𝑡ℎ𝑒 𝑠𝑎𝑚𝑒 𝑜𝑟𝑖𝑒𝑛𝑡𝑎𝑡𝑖𝑜𝑛 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑟𝑒𝑔𝑢𝑙𝑎𝑟 𝑎𝑛𝑑 𝑐𝑎𝑟 𝑚𝑜𝑑𝑒” 𝐓𝐡𝐞𝐫𝐞 𝐢𝐬 𝐚 𝐬𝐩𝐢𝐤𝐞 𝐢𝐧 𝐢𝐬𝐬𝐮𝐞𝐬 𝐰𝐢𝐭𝐡 𝐨𝐟𝐟𝐥𝐢𝐧𝐞 𝐦𝐨𝐝𝐞 (𝟒𝟎𝟎 𝐮𝐬𝐞𝐫𝐬, 𝟒% 𝐨𝐟 𝐭𝐨𝐭𝐚𝐥 𝐮𝐩 𝟏𝟎𝐱 𝐌𝐨𝐌) "𝐷𝑜𝑤𝑛𝑙𝑜𝑎𝑑𝑒𝑑 𝑠𝑜𝑛𝑔𝑠 𝑎𝑟𝑒 𝑢𝑛𝑎𝑣𝑎𝑖𝑙𝑎𝑏𝑙𝑒 𝑤ℎ𝑒𝑛 𝑜𝑓𝑓𝑙𝑖𝑛𝑒, 𝑒𝑣𝑒𝑛 𝑡ℎ𝑜𝑢𝑔ℎ 𝑡ℎ𝑒𝑦 𝑠ℎ𝑜𝑤 𝑎𝑠 𝑑𝑜𝑤𝑛𝑙𝑜𝑎𝑑𝑒𝑑."  "𝐴𝑝𝑝 𝑘𝑒𝑒𝑝𝑠 𝑑𝑒𝑓𝑎𝑢𝑙𝑡𝑖𝑛𝑔 𝑡𝑜 𝑜𝑓𝑓𝑙𝑖𝑛𝑒 𝑚𝑜𝑑𝑒 𝑒𝑣𝑒𝑛 𝑤ℎ𝑒𝑛 𝐼 ℎ𝑎𝑣𝑒 𝑖𝑛𝑡𝑒𝑟𝑛𝑒𝑡 𝑐𝑜𝑛𝑛𝑒𝑐𝑡𝑖𝑜𝑛." This is a sampling, not a summary of the most interesting specific comments made by users, including specific facts and statistics. Moreover it is actionable and gives a starting point as to where we might dig further. This is not something AI is good at out of the box. It requires the ability to quantify what is being reported by users, decide on the most important points and present an output that is suitable to the use case Others see the same?

How to increase user retention? Add “pain” to your onboarding flow… let me tell you why The most common problem our customers think they have is bad onboarding... as they see users sign up and then never truly activate Flow goes like this: • User signs up  • Clicks some buttons  • Never comes back The typical response to this: “we need to improve onboarding”: • Build guided onboarding process • Add user to drip email campaign 🙃 Here's the rub: the problem usually isn't onboarding… If users have a strong need they will endure lots of pain to get the job done and if they hit problems along the way they will let you know about it So what is the problem then? • Attracting the wrong users (those who don't really have the need your product solves) • Not communicating value to the right users In order for us to diagnose why users are dropping off, we need to first segment to those who truly have the need we are solving. The solution is to add some "pain" to your onboarding process... Users who are willing to endure some “pain” to use your product are communicating to you that they really have the need. Examples of positive onboarding “pain”: • Sign up questionnaire  • Upfront payment • Book a demo • Set up an integration to get started This friction serves too purposes, users who proceed through these steps: 1. 𝐑𝐞𝐚𝐥𝐥𝐲 𝐡𝐚𝐯𝐞 𝐭𝐡𝐞 𝐧𝐞𝐞𝐝: giving you a clean sample to judge your retention metrics 2. 𝐇𝐚𝐯𝐞 𝐬𝐨𝐦𝐞 𝐬𝐤𝐢𝐧 𝐢𝐧 𝐭𝐡𝐞 𝐠𝐚𝐦𝐞: making them more likely to stick around For any onboarding friction lovers out there, what have you found works? Let me know in the comments

Delighted to share this research paper, Zelta has collaborated on with Martian, G2, Copy.ai, 6sense, Supernormal and other AI leaders For anyone hoping to get into the weeds on automated prompt optimization and techniques to get more from LLMs, this is a must read

At Martian, we are fortunate to work with many of the world's most advanced users of AI. We see the problems they face on the leading edge of AI and collaborate closely with them to overcome these challenges. In this first of a three-part series, we share a view into the future of prompt engineering we refer to as Automated Prompt Optimization (APO). In this article we summarize the challenges faced by leading AI companies including Mercor, G2, Copy.ai, Autobound, 6sense, Zelta AI, EDITED, Supernormal, and others. We identify key issues like model variability, drift, and “secret prompt handshakes”. We reveal innovative techniques used to address these challenges, including LLM observers, prompt co-pilots, and human-in-the-loop feedback systems to refine prompts. We invite the broader AI community to collaborate with us on research in this area. If you are interested in participating, please reach out to us! https://lnkd.in/eyvfEe2X #AI #ArtificialIntelligence #PromptEngineering #APO #MachineLearning #AIResearch #Collaboration #Innovation #ModelVariability #ModelDrift #LLM #FutureOfAI #AICommunity #HumanInTheLoop #AIChallenges #AIsolutions

At Martian, we are fortunate to work with many of the world's most advanced users of AI. We see the problems they face on the leading edge of AI and collaborate closely with them to overcome these challenges. In this first of a three-part series, we share a view into the future of prompt engineering we refer to as Automated Prompt Optimization (APO). In this article we summarize the challenges faced by leading AI companies including Mercor, G2, Copy.ai, Autobound, 6sense, Zelta AI, EDITED, Supernormal, and others. We identify key issues like model variability, drift, and “secret prompt handshakes”. We reveal innovative techniques used to address these challenges, including LLM observers, prompt co-pilots, and human-in-the-loop feedback systems to refine prompts. We invite the broader AI community to collaborate with us on research in this area. If you are interested in participating, please reach out to us! https://lnkd.in/eyvfEe2X #AI #ArtificialIntelligence #PromptEngineering #APO #MachineLearning #AIResearch #Collaboration #Innovation #ModelVariability #ModelDrift #LLM #FutureOfAI #AICommunity #HumanInTheLoop #AIChallenges #AIsolutions

$10 billion business idea below 👇 Software companies track everything users do in their products with analytics tools like Mixpanel, Amplitude, Heap, Pendo etc But “what people are doing” is just one side of the coin, we also care about “what people are telling us” .. the qualitative insight We have the data; recorded sales & success calls, interviews, surveys, emails, community forums and social media… oceans of it… growing every day Yet, most orgs utilize ~1% of it Everyday customers/prospects communicate: • Their needs and goals to Sales • Why they use/ or don't use your product to Customer Success • Bugs, questions and feature requests to Customer Support • Their pain points to User Research • What they love or hate about your product on review sites • Asks for help on community forums So much of what we do stems from this customer input, yet most orgs rely on periodic team meetings to share anecdotes of what they’ve heard A company should build an AI for this, one that listens to every Gong call, Zendesk Ticket, survey response, G2 review etc…maps it all to CRM and product analytics data and uses AI make sense of it all The AI would: • Stack rank customer problems and their associated ARR • Triage product issues to appropriate person automatically • Automatically “close the loop” with customers/ sales when enhancements are shipped • Draft segment specific marketing content • Tell us who our ICP is Call it something like "Customer Intelligence"… this would be a game changer for the industry… Anyone building this? 🤔

Thanks Melissa Perri! Product Ops teams are quickly becoming a key leader and thought partner in the AI strategies of our customers. Such an exciting time!

I am really excited about these new products that use AI to help gather insights. Here's some that I think are awesome: - Dovetail: user research platform, uses AI to surface insights -Zelta AI: distills insights from sales calls - Intercom: just said they are doubling down on AI both for support and generating insights - unitQ: generating insights after asking for user feedback What else out there that you're using to make better strategic decisions? I am very intrigued about how this can help Product Operations. #productmanagement #productoperations

Gaining buy-in for building 0 to 1 products inside large companies relies on effectively communicating the product vision.. Lenny Rachitsky recent podcast with Mihika Kapoor (product leader at Figma) is packed full of insight on building new products inside large companies and her process for building compelling product vision documents 𝗚𝗮𝘁𝗵𝗲𝗿 𝗺𝗲𝗮𝗻𝗶𝗻𝗴𝗳𝘂𝗹 𝘂𝘀𝗲𝗿 𝗶𝗻𝗽𝘂𝘁: "The first is that you cannot go into a vacuum and come out with a compelling vision that does not exist. You have to be fundamentally inseparable from your users, and also, fundamentally inseparable from your team. And so, I think that there is sort of this important cross-pollination of functions that is really important in crafting a compelling vision" 𝗦𝘆𝗻𝗰 𝘄𝗶𝘁𝗵 𝘆𝗼𝘂𝗿 𝗦𝗮𝗹𝗲𝘀 𝗢𝗿𝗴: "And then if you're on the larger side, I think that having a really tight relationship with your sales team is really important. And basically, just being on sales calls because you want to be in a situation where the customer pain points on sales calls are cross pollinating into the product roadmap. And so I think really leaning into that, building that relationship between these traditionally more siloed orgs and hopping on those calls is something that I'd really recommend." 𝗔𝗱𝗱 𝘂𝘀𝗲𝗿 𝘁𝗲𝘀𝘁𝗶𝗺𝗼𝗻𝗶𝗮𝗹𝘀 𝗳𝗿𝗼𝗺 𝘁𝗵𝗲𝘀𝗲 𝘀𝗼𝘂𝗿𝗰𝗲𝘀 𝘁𝗼 𝘆𝗼𝘂𝗿 𝘃𝗶𝘀𝗶𝗼𝗻 𝗱𝗼𝗰: "So, when I put together a vision with my team at Figma, it's all about not just your traditional, "Okay, here are pain points. And then, here are solutions. And then, here is the timeline and costing." But rather how can you bring all of those things together and how can a vision pitch effectively Because at the end of the day, simply describing a product idea in words is not as compelling as seeing a testimonial from a user on top of a prototype or a mock, and really feeling the pain points." Link to the episode in comments for those interested, love to hear reactions from anyone else who listened in 👇