Przemyslaw Modrzewski

Voice AI II: Voice-in-motion v.s. Voice-at-rest One way to think about Voice AI use cases in contact centers is to distinguish between two large categories of voice ai use cases: voice-in-motion and voice-at-rest. They are different. We have implemented and deployed both across the Cresta platform. 𝗩𝗼𝗶𝗰𝗲 𝗶𝗻 𝗺𝗼𝘁𝗶𝗼𝗻: 𝘄𝗵𝗲𝗻 𝘃𝗼𝗶𝗰𝗲 𝗶𝘀 𝘀𝘁𝗿𝗲𝗮𝗺𝗶𝗻𝗴  • Integration: telephony platform dependent, can fork the audio in various points in the audio stream.    • Requirement: low latency, high speed, low-cost inference, see only conversation to this point, efficient state orchestration  • Use cases: real time coaching,guidance, workflows automation (e.g. automatic form filing), taking notes live, voice to voice language translation, noise cancellation, call center fraud detection, etc 𝗩𝗼𝗶𝗰𝗲 𝗮𝘁 𝗿𝗲𝘀𝘁: 𝗮𝗳𝘁𝗲𝗿 𝘃𝗼𝗶𝗰𝗲 𝗰𝗼𝗻𝘃𝗲𝗿𝘀𝗮𝘁𝗶𝗼𝗻𝘀 𝗮𝗿𝗲 𝘀𝘁𝗼𝗿𝗲𝗱 𝗼𝗻 𝗽𝗲𝗿𝘀𝗶𝘀𝘁𝗲𝗻𝘁 𝘀𝘁𝗼𝗿𝗮𝗴𝗲.  • Integration: none if already has streaming voice integration. Otherwise, it can be a generic file dump.   • Requirement: aggregating massive historical conversations, see the conversation in its entirety, faster filtering based on structured meta data , sampling is essential.    • Use cases: Insights on customer, voice of customer, Insights on workforce, QA, discover unknown unknown winning sales behaviors, automation opportunities. 𝗦𝘆𝗻𝗲𝗿𝗴𝘆 There is a ton of synergy from a pure data processing perspective, for voice-in-motion and voice-at-rest, and at Cresta, we believe to harness the full power of Voice AI, you need one platform to do both, a few reasons:   • Shared compute, no waste: expensive computation like ASR, classification, sentiment can be shared between real time inference and post-call batch processing. Tags generated from AI inference at real time can be used for post-call BI, and filtering for deep research agents.    •  Single plane of glass, no data silos, unified UX.    •  Insights → Action. Insights that can only be derived from many many conversations should be immediately applied to change the outcome of real time conversations of the future. In Cresta, you can do that with a few clicks to take insights to action via Cresta Opera - the command center.

71

9 Comments

DV 360 vs Trade Desk Features (A Detailed Comparison Article) Choosing between DV360 vs The Trade Desk is one of the biggest decisions for marketers exploring programmatic advertising. Both demand side platforms (DSPs) offer advanced tools, but they differ in identity solutions, inventory access, optimization, brand safety, and reporting. DV360, part of the Google Marketing Platform, stands out with exclusive YouTube access, Ads Data Hub measurement, and Custom Bidding. In contrast, The Trade Desk excels in open-internet CTV reach, UID2 identity, and supply path optimization through OpenPath. This detailed comparison highlights their unique strengths, pros and cons, and practical use cases to help advertisers choose the right DSP—or combine both—for maximum impact.

74

12 Comments

You can rank #1 on Google and still lose to a competitor with weaker products. Why? They show in AI answers, and you don’t. Billions of searches now run through LLMs and Google’s AI daily. After testing AISEO tactics across client sites, here’s what actually works: 👇 1. The Entity Recognition Game AI thinks in entities, not keywords. When someone asks “Who’s the best SEO agency?” it scans for entities. Become one: - Complete every business listing (Google, Bing, Apple Maps, directories) - Get cited in Wikipedia - Claim your Knowledge Panel - Keep NAP (Name, Address, Phone) consistent across 50+ directories 2. The Question-Answer Architecture AI pulls direct answers, not web pages. Your content should work like a Q&A database. Format: H2: “How long does SEO take?” Direct answer in first line: “SEO takes 3–6 months to show results.” Bullets support below. Pro tip: Drop your query into ChatGPT and reverse-engineer the structure of content it cites. 3. Reputation Signals AI Reads AI crawls what people say about you. Signals that matter: - Reddit comments mentioning you - Quora answers citing your expertise - LinkedIn posts tagging you - Forum discussions where you add value - YouTube comments on relevant videos 4. The Citation Multiplication Strategy AI stacks citations. When Harvard Business Review quotes a study, AI trusts it more than the source. So get quoted by sources AI already trusts: - Pitch yourself to journalists - Comment on trending news with your take - Publish research bigger sites can cite - Guest post on authority sites One citation can trigger many AI mentions. 5. Conversational Content Patterns AI favors natural queries. People don’t ask Google “SEO services Chicago.” They ask “What’s the best SEO company in Chicago?” Rewrite content like: Instead of: “Chicago SEO Services – Professional Search Engine Marketing” Use: “7 Best SEO Companies in Chicago (2025 Guide)” Instead of: “Link building strategies for higher rankings” Use: “How to get high-quality backlinks without penalties” 6. The Review-to-Recommendation Pipeline AI weighs reviews heavily. Not just stars, but the words. Ask happy customers specific questions: - How did our service solve your challenge? - What results did you see, and when? - Would you recommend us to others? These become AI training data. 7. Multi-Platform Content Syndication AI crawls everywhere you publish. Repurpose content across: - Blog - LinkedIn - Medium - Industry newsletters - Podcasts (transcripts get crawled) YouTube video descriptions More touchpoints = higher AI recognition. Traditional SEO is about ranking pages. AI SEO is about ranking your expertise. Stop optimizing for robots. Start positioning yourself as the expert AI should recommend.

137

37 Comments

🚀 Veni, Vidi, Vici: What CEOs Can Learn from Caesar’s 5-Hour Blitz ⚔️ What do a Roman general and a modern CEO have in common? More than you think. 💡 In just 5 hours, Julius Caesar crushed his enemy at Zela and sent a message that echoed through history: “I came, I saw, I conquered.” 🏛️ This article breaks down how Caesar’s bold leadership applies to: 🔧 Engineering firms 🏗️ Construction companies ⚖️ Legal practices 💻 IT providers 💙 Non-profits seeking innovation No more red tape. No more endless meetings. The first to act often wins. 🥇 🌍 These principles work everywhere—from San Antonio to Singapore, London, Dubai, and beyond. 👉 Read the full article and start leading like Caesar: https://lnkd.in/ejWTMbrG #Leadership #BusinessStrategy #Innovation #ITLeadership #VeniVidiVici #SanAntonioBusiness #CEOInsights

149

42 Comments

SEOs will benefit more from reading books like this than reading half the garbage being peddled online by gurus. AI powered search represents the biggest opportunity for businesses to grow in history. Allowing the creation of new brands to reach customers ignored or deemed not worth going after by larger brands. However, SEOs will need to think less about traditional SEO tactics and learn more about building brands online. That's the key to success.

105

36 Comments

Before implementing a unified system, Morgan & Morgan saw huge differences in case outcomes depending on who touched the file. The fix? Data analytics + workflows + collaboration tools to ensure every case gets the same high-quality treatment. Consistency drives growth — here’s how they proved it.

37

AI search is transforming how people and AI agents find and buy. A new field AI Search Optimization (GEO/AEO/ASO) is emerging to boost visibility in generative engines. New LLM-native platforms (2023–2025) are leading with four key capabilities: multi-platform monitoring, competitive benchmarking, sentiment tracking, and actionable insights. Those who can shape how AI models read the web will define the future of search. Source: Jason Saltzman & CB Insights

21

“Will online shops turn into GPT inventories? How would that play for differentiation?” 🤯 The comment on last week’s Financial Times article says it all, we’re up for a massive transformation of what differentiation means in digital commerce. What happens once #agenticcommerce fully disintermediates commerce websites? SEO - gone, site search - gone, product page - gone. And it's not just B2C either, ChatGPT is now prevalent in how B2B buyers make their choices. Where you need to invest to drive commerce is evolving fast, and product data is becoming the last barrier of differentiation. It is just crazy to see how fast that change is coming! FT article: https://lnkd.in/dkGVSJbx

71

1 Comment

Google recently announced the shift from the Shopping Content API to the new Merchant API. But tucked away in the documentation is something most merchants won't have spotted yet: 👉 AI-powered text suggestions in Product Studio. Until now, Product Studio has only been about images. But with this update, the API can generate product titles, descriptions, and even attributes like size or material. Sounds great, right? Yes and no. While AI can speed things up, it's far from perfect. Titles can miss context (like regional sizing), and descriptions often turn into fluffy, generic text. That's why manual review is still essential if you want high-quality results. I believe Google will roll this out into the Merchant Center UI soon, which could be a game-changer for merchants managing large catalogs. For now, it's API-only, but worth keeping an eye on. I’ve broken this all down in my latest FeedArmy video, including examples of what the AI actually generates and where it falls short. https://lnkd.in/gA_-hddY

24

Most people have glossed over the fact that there are 2 very distinct ways to “win” in LLM marketing: citations vs mentions. Semrush's new AI Visibility Index report offers one of the clearest breakdowns I've seen on the 2 strategies, and a look at how brands from Bankrate to Patagonia are doing it right. The skinny: You can "win" in ChatGPT/AI Mode by either: - publishing content that gets directly cited, or - getting mentioned on a 3rd party source that gets referenced The report breaks down the playbooks of who is winning for both mentions and direct citations, and how: For Mentions: - conduct original survey/research, get cited by news outlets - create list of top 5 review sites specific to your industry, create a strategy - form affiliate partnerships with respective players in your space. Get on "Best of" lists. Sometimes (many times) this means creating a commercial relationship. For Citations: - Dial down marketing speak on your pages. Remove things like "industry-leading" on your own pages. (great examples on slide 89 of their report) - Prioritize static/HTML content - put your data in easy to read tables, not fancy javascript Surprisingly, the report shows that brands most frequently mentioned in AI answers aren’t always the ones most frequently cited as sources. Few are doing both well, which means a lot of white space... Full index/report from Semrush here: http://bit.ly/420HjlJ

37

15 Comments

Senior citizens give clicks, Gen Z gives Add-to-cart. A few years ago, we launched a brand campaign on Meta to boost brand consideration. The campaign had an "interest+behavior" targeting and was optimized for clicks. It drove a lot of traffic to the site. However, the traffic had a high bounce rate and fewer retained sessions. The creative was very good and had a great CTR. We found a fascinating insight when analyzing what went wrong with the campaign. In Meta, you get what you optimize for. If you optimize for clicks, the campaign targets so-called clickers, and you get many clicks. The same happens when you optimize for add-to-cart. You get explorers. However, it was interesting to know who those clickers and explorers are. In the campaign mentioned above, - 57% of the spends and most of the clicks were from people over 55, and - 39% of the clicks(overall) were from people who are 65+ I have seen this phenomenon repeatedly. The elderly audience tends to interact with the ads more than the other age groups. But they often bounce off the website when they land there. Hence, in Meta, they form a significant chunk of 'Clickers'. This is true even today. If you are running a campaign for link clicks or a landing page view with an age restriction, check your campaign CTR by age. You will see this phenomenon clearly. Conversely, the younger audience, under 21 years of age, makes up the majority of the explorers. I have seen this in many campaigns, too. If you have an active Meta campaign where you recently changed the optimization from bottom funnel to the mid-funnel event, check the age distribution change. The campaign's age group would have skewed towards the lower side. In parallel, your add-to-cart-to-purchase conversion would have gone down.

137

13 Comments

I went down a rabbit hole 🐇📊 with the new Semrush AI Visibility Index report and it completely shifted how I think about search. We wanted this to be one of the most comprehensive looks at AI-driven visibility out there, so we analyzed 2,500 AI search prompts across industries 🌐 from digital tech to fashion 👗💻. And when you look at data on that scale, you start noticing some really interesting patterns. The biggest one for me: AI doesn’t treat every brand or every industry the same way. A few things that really stood out: 🔸 Being popular isn’t the same as being trusted. In many sectors, the brands AI talks about aren’t the ones it trusts enough to cite. 🔸 Industries behave differently. Fashion is fragmented, consumer electronics is dominated by a few big players, and business services is an all-out battle with AI mentioning almost five brands per query on average. And that’s really the point: there isn’t one universal playbook for AI visibility. It depends on your industry, your competitors, and the sources AI relies on when building its answers. That’s why we created this report so you can dig into the data for your own sector, see who’s winning and why, and figure out what it means for your strategy. If you want to understand how AI is shaping visibility in your space, you can explore all the findings in our free AI Visibility Index report: 👉https://shr.bi/8evrFlEy #AIVisibilityIndex

34

2 Comments