Tahsim Ahmed’s Post

Thank you to UNPARSED for awarding Tico the 'Top Bot (Conversational AI Experiences)', announced in 🇬🇧 London, UK earlier this month. When building 'Tico' alongside Nicolas – we knew that we had build the best AI chatbot to solve the problems we faced ourselves with manual support, and pave the way to help our teams and customers build their own. Results? ✅ 93% support ticket resolution rate ✅ First reply time: 308min → 1min ✅ CSAT increased to 93.8% ✅ Slashed cost/ticket from 31 USD to 16 USD, to net savings. Most teams spend too long or create a complex scope for their AI Agents. At Voiceflow, our approach to Conversational AI Agents focus on extensibility and scaleability – taking you from idea & design to Production in 10 days. One-tool, end-to-end – all with one PM & one Dev. How it works: 1. Obsess over your intended customer experience (both granularly and broadly) for all 10 phases. It’s the only thing you should do. 2. Define scope and user flow first. Then design it, prototype it, and push to beta-prod immediately. 3. Fix any hallucinations, chunks and KB approaches that arise for initial release. It's scary fast, and straightforward, focus all-in the tooling - the rest falls into place. Comment if you’re interested, and I'll send the 10 key phases and a supplemental deck (covering Tico's award-winning components and features) to help you build your first automated AI Agent, or improve your existing AI #chatbot. --- Thank you to the judges and organizing team for this honor! Kane Tom Taras #ConversationalAI #unparsed2024

This is the MOST strategic hour you can spend on deploying AI in the Enterprise. Denys Linkov (Head of ML) and I looked at the deployment and measuring ROI on AI Agents, through the CX lens. Regardless of whether you oversee Support or any other function – this recording should be essential for any leader or PM planning an AI Agent deployment or automation. P.S. There’s a hot downloadable (Blueprint spreadsheet) available accompanying the video too. Comment if you want me to send it over.

Han and Zeno's recently shared perspectives that changed how I’ve been thinking about Customer Support. It’s a concept from the lens of simple engineering. The key idea in the support analogy is that the ideal scenario is for customers to be able to self-serve and resolve their issues without needing to escalate to a human support agent. The "try block" represents the resources available for customers to help themselves, like documentation, FAQs, and AI-powered chatbots or virtual agents. To me, the "catch block" is escalating to a human agent when self-service fails. Seeing this in real-world parallels, Customer Support is like filing your taxes. The try block includes tax prep software, online guides, and chatbots that answer common questions. If you get stuck, you move to the catch block - hiring a CPA or visiting an IRS/CRA office. I imagine companies like Intuit think about taxpayer-friendly systems investing heavily in the try block for their customers, using AI to tailor guides to your specific situation, flag potential errors, and suggest deductions. By making the software as helpful as possible, a true Copilot – they reduce the need for costly human intervention. Optimizing the 'try' workflows is how award-winning AI Agents like Tico are able to achieve a constant 96% AI automation rate, and it can only be done by being able to configure your AI Agent, and it starts with #ConversationalAI. If you’re navigating high support volumes with a lean team, reach out and I'll show how you can help blend & configure AI automation with quality for your customers.

Still buzzing IRL from last week's full house, where Denys and I dissected deploying AI Agents, namely Conversational AI chatbots in Customer Support. For those that missed it, I am thrilled to invite targeted leaders this week, to dive deeper into their AI Blueprint in our upcoming invite-only session. This is an opportunity to look inside the deployment of your first automated AI Agent or dive deeper into the Blueprint of your existing AI #chatbot. Comment if you'd like an invite. Aimed at leaders planning to 10x their Gen AI/AI Agent efforts this year. To encompass the ethos of what was discussed, it's exactly as Support legend, Sarah Betts calls out in her most recent post... – aiming for industry averages keeps you average. Instead, strive to stand out like a beacon. Our AI Agent was built with the unique ability to have data integrations, from customer interactions, product feedback loops and underlying workflows – to thoughtfully impact factors that driving customer happiness and success. Thanks to Mike, Michael, and Burcin for organizing & an absolute honor sharing the stage and panel with heavy hitters: Neil Pengelly, National Tech Partner from EY and Renée D. from Thoughtworks Canada.

49% of AI Leaders are struggling this year estimating and demonstrating value for AI projects... That's why we're so back – Denys and I are back on-the-road, with (2) Enterprise AI Agent-focused talks kicking off the summer. Here are the ones coming up: [05.28] TORONTO! ProductTank Toronto's: Artificial Intelligence in Product Management. Join us tomorrow IRL, for a talk discussing AI in enhancing customer interactions, and the metrics behind it. Additionally, I'll also be sharing my perspectives on a panel discussing PMing AI Projects. [06.06] AI Strategy to Blueprint, feat. Tico - focused online event discussing key metrics for AI customer support, how to develop and track those metrics, and real dashboards we use to report on progress. From the two sessions, we'll cover: - Overall approach on integrating AI agents into products and processes/workflows. - Progression of metrics, the basic, intermediate and advance ones - How we build technical systems to better capture metrics - Breaking down levels of support and how support moves revenue metrics - Real examples from a live support agent linked to Zendesk I'll throw registration details in the comments. #TechTO #AIAgents #ConversationalAI #ProductManagement | Voiceflow Mike Burcin Neil Sam

I have the privilege to be on the operating-edge of deploying production AI agents into Customer Support, achieving substantial results. But last week, I noticed a prominent vendor STILL using out-of-box Zendesk with auto-replies... I compiled examples of tickets where Tico, an award-winning Conversational AI Agent resolving 6000+ tickets/monthly (without ever reaching a human) – deployed directly on Zendesk. The journey to 96% deflection starts with approaching support as a journey – starting with breaking the mold of traditional methods, like auto-replies. From there, Customer Care leaders can maneuver Virtual Assistants in outcomes to drive – from overall resolution to efficient escalations, even revenue-generating. Reach out if you want an exclusive invite-only access to a deep-dive session into metrics around AI Agents, deployed with rich data sources and customer-centric design.

Earlier this week, OpenAI released their flagship model GPT-4o, and instead of just talking about it – we deployed it to production. Here are our results with Tico, Voiceflow's Customer Support AI Agent automatically resolving 96% of support volume.