In today’s world, where technology is developing at an incredible pace, OpenAI is implementing controversial “improvements” in ad creatives, changing headlines and links without human oversight. Simultaneously, media giants, including CBS and CNN, are actively entering into partnerships with betting companies, raising questions about the boundaries of ethics and transparency in the run-up to elections. These changes are shaping a new reality for digital arbitragers and marketers, offering both unprecedented opportunities and significant risks.
OpenAI and Ad “Improvements”: Automation or Loss of Control?
Since launching its advertising platform earlier this year, OpenAI has emphasized the importance of safe and relevant advertising. However, recent “improvements” to ad creatives, which can affect everything from the headline to the target link, raise serious concerns.
Essentially, this means the potential exclusion of human control from the creative decision-making process. We, as digital arbitragers, see this as a double-edged sword.
On the one hand, automation can reduce CPM and simplify scaling. On the other hand, without adequate testing and transparent statistics, it can lead to budget waste.
Media and Gambling: A New Era of Partnerships
From Prediction Markets to Direct Sportsbooks
“Media outlets used to justify collaborations with prediction markets like Kalshi and Polymarket by saying they weren’t exactly gambling. But now CBS is openly partnering with DraftKings, showcasing their branding on air during elections,” Status reports.
CNBC became the first news organization to partner with Kalshi, integrating prediction data into its news. Now CBS, CNN, Fox News, and Associated Press also have deals with Kalshi.
Substack and Dow Jones, parent company of The Wall Street Journal, are partnering with Polymarket. These “prediction markets” deny that they are sportsbooks, although they accept bets on sports and partner with sports leagues.
But now CBS is taking the next step, directly partnering with DraftKings, which means open integration of a betting brand into news content.
Sports Sponsorships: The New Reach Economy
Amid growing pressure on marketers to prove advertising ROI, spending on traditional ad placements during NBA games is no longer an obvious choice, as Digiday reports. Instead, sponsorships are becoming an increasingly popular way for brands to create associations with the NBA and its players, going beyond intrusive 30-second spots.
Ledger, a blockchain company, signed a deal with the San Antonio Spurs. According to the company, Spurs fans are 4.8 times more likely to name Ledger as a leading firm in its category and twice as likely to prefer it over competitors. This is real math proving effectiveness.
However, large sports partnerships require careful verification of effectiveness. Adidas, which spent over $1 billion on marketing during the World Cup, is now under pressure from investors demanding proof of ROI.
Frequently Asked Questions
What is the danger of OpenAI’s ad “improvements”?
The main danger lies in the loss of human control over creatives, which can lead to irrelevant or ineffective advertising, as well as potential reputational risks. This requires thorough hypothesis testing and continuous monitoring.
Why do media partnerships with sportsbooks raise questions?
Such partnerships blur the lines between news content and gambling advertising, which can undermine trust in the media and raise ethical questions, especially in the context of election coverage.
What are the alternatives to expensive advertising contracts with celebrities?
Alternatives include strategic sponsorships, banner advertising in short videos (YouTube Shorts, TikTok, Reels), where you can get cheap reach and mass reach for pennies. This allows you to not pay influencers, but get transparent statistics and test hypotheses for 1000 rubles.
Conclusion
The world of advertising and media is rapidly changing. While OpenAI experiments with creative automation, media giants are looking for new revenue streams in partnerships with sportsbooks. We, as arbitragers, see both risks and huge opportunities for those who are willing to adapt.
Forget about bloggers and expensive contracts. Today, CPM cheaper than 50 rubles for millions of reaches is a reality that can be achieved through banner advertising in Shorts. Do you want to buy reach without creating a video and get a native banner in Shorts? Contact us to find out how we can help your business get mass reach for pennies with transparent statistics.
Creating videos with AI has become commonplace, but until now, the process of refining the first draft to perfection took hours. Not anymore. Lumen5 Agent is a revolutionary tool that not only generates videos but also efficiently processes feedback, making the iteration process as fast and intuitive as possible.
The Problem with AI Video: Not Generation, But Iteration
We’ve all seen this “trick”: you insert a link, blink, and the AI produces a video in a minute. But then hours of edits begin: the scene is too long, the title is unreadable on a light background, a beach clip in a video about a quarterly report.
Turning “some video” into “a video the team can send” is where precious time is lost. This is the gap Agent closes, and it’s already available in Lumen5.
How Agent Changes the Game in AI Production
Lumen5 Agent is a conversational way to create and edit videos. You describe what you want in simple language, watch it build the video scene by scene, and then make changes directly in the editor until the result is perfect.
Most AI video tools produce a first draft and “disappear.” Agent, however, stays for the second, third, and fourth drafts. It’s not “here’s your video, good luck,” but “okay, what are we changing?”
“Before, dealing with AI video was like this: you gave a link, and you got whatever came back. Like a vending machine, but for footage. Agent demands more, because that’s how you get something usable.”
Length, style, media: Specify the desired duration, style, and suitable media for your audience, and Agent will build the video according to this brief.
Smart questions: If your request leaves something important open, Agent doesn’t guess, but asks clarifying questions through a new questionnaire. Only those it cannot ascertain from the information already provided, so you don’t fill out a tax return for a 30-second promo.
New Opportunities for Startups and Teams
Now there are six ways to get started on the “Create” page. If your team works with PDFs and slides, this is for you. Agent understands page numbers, including those printed on the pages themselves.
When you ask for a diagram on page 12, you get a diagram on page 12, not on page 11 or “something that looks like a diagram from somewhere in the middle.”
Page snapshots: Agent can take an image of a page from a PDF, PowerPoint, or Word document and insert it directly into your video. This is useful when the document itself is a key element, such as a report cover or an important slide.
Acceleration: According to our product lead’s testing, the document workflow is now approximately 50% faster than before. This is scalable efficiency that saves time and resources.
Process Transparency and Instant Feedback
While Agent is working, the creation window now shows each scene as it’s being built and updated. Goodbye, loading bar and silent prayers. Hello, real-time understanding of what’s happening.
This sounds like a cosmetic improvement, but it’s not. If you see that the second scene is going wrong while the fifth is still being built, you can intervene immediately, instead of discovering the problem at the end and starting over with a heavy sigh.
New Recommendation Engine: Quality, Not Automation
Under the hood, a brand new recommendation engine is making choices. We are proud of its features:
Text and icon synchronization: Text and icons appear the moment they are spoken, so nothing appears too early or lingers after the phrase is uttered.
Contrast detection: Checks text readability against any background, not assuming every photo is conveniently dark.
Smart text wrapping: Headings break at sensible points and resize less awkwardly. No more lonely words left on a separate line.
Listening length: Requests to make a video longer or shorter now hit the target duration much more accurately.
On-demand clip replacement: “Replace this clip with something from the office” – and Agent understands.
These details collectively distinguish a video that looks automated from a video that looks made by a caring human. This is key to AI efficiency versus real UGC.
Agent for Teams: Democratizing Video Production
A new Agent tab has appeared in the editor. After creating the first draft, you can continue to make requests in plain language without leaving the editor. Your video is adjusted on demand, not by “hide-and-seek” with the right panel.
All manual controls remain in place, so experienced users can continue to work in their usual way. The point is not that you stop editing, but that the fastest way to a specific change is simply to describe it.
Brand Consistency with “Recipes”
For a team, this is a big shift. Instead of one person who knows the editor by heart (and gets all the “quick requests” in the office), everyone can release a video.
The Create page now starts with “Recipes,” each with a preview of what it produces. These are great for inspiration if you’re new. But their real superpower is consistency. When four people on the same team choose “Report Summary,” you get four videos that are clearly from the same brand, not four creative interpretations, plus a brand manager reaching for the paracetamol.
The Future of AI Video: Focus on Iterations
If you’ve been following AI video, you’ve noticed the conversation has changed this year. The interesting question stopped being “can a model create a clip?” and became “can it follow directions?”. Editing has always taken up most of the production time. Faster footage generation doesn’t fix that.
On the contrary, it can make things worse, because now you have ten versions to choose from instead of one, and no extra hours in the calendar. This aligns with what we see in marketing teams every day. The bottleneck isn’t creating the video, but the endless revisions after it’s made, and how many people on the team can actually manage that process.
So we built Agent around the iteration cycle, not just the first generation. Creating a video is easy now. Creating your video, the one that’s actually ready to send, that’s what we wanted to simplify.
Lumen5 Agent: Your New AI Production on Subscription
Agent is already in Lumen5, waiting on the Create page. If you’ve only used us by pasting a link and taking what came out, this is worth ten minutes of your team’s time. Give it a real brief, with length, style, and audience. Then tell it what to fix. Then change your mind and tell it again. It won’t take it personally.
And we genuinely want to hear where it falls short for your team. We’re still fine-tuning the recommendations, and the fastest way to improve them is to hear what you asked for and didn’t get. Tell us. We’re ready.
Want more ideas on how to turn existing content into more effective video? Subscribe to the Supercharge Marketing podcast, where marketing leaders share what really works for their teams.
Frequently Asked Questions
What is Lumen5 Agent?
Lumen5 Agent is a conversational tool for creating and editing AI videos within Lumen5. You describe the desired video, watch it build scene by scene, and then request changes in plain language until it’s ready.
How is Agent different from other AI video generators?
Most AI video tools focus on the first draft. Agent is built around the iteration cycle, so after creating the first version, you can continue to request edits (e.g., shorten a scene or replace a clip) directly in the editor.
What can I use to start a video with Agent?
You can start with a blog post, script, document, recording, screen recording, or an existing video similar to what you want.
Can Agent work with PDFs, PowerPoints, and Word files?
Yes. Agent understands page numbers, including those printed on the pages, and can turn a page from a PDF, PowerPoint, or Word file into an image in your video.
Do I still have manual editing controls?
Yes. All manual controls in the Lumen5 editor remain in place. Agent simply provides a faster way to make specific changes by describing them.
Does Agent change Lumen5 pricing or usage limits?
No. Your pricing and usage limits, as well as Blueprints and Script Composer, work exactly as before. This allows you to scale your AI content production without hidden costs.
How does Agent help teams stay on brand?
“Recipes” on the Create page, such as “Report Summary” or “Brand Story,” give everyone on your team the same starting point, so videos made by different people still look like they belong to the same brand. This is important for AI video for advertising and AI UGC for crypto projects, where consistency is critical.
Lumen5 helps marketing teams turn existing content into polished, on-brand video, faster and at a fraction of traditional production costs. Learn more or request a demo.
Despite multi-billion dollar lawsuits and litigation, Meta demonstrates impressive revenue growth, especially in the third quarter. How does the company manage to stay afloat and even thrive amidst such financial challenges? The answer lies in strategic investments in artificial intelligence, which is becoming a key driver of Meta’s advertising revenue growth, even when new products like Muse are not yet profitable. Let’s delve into the details of this phenomenal success and learn how Meta uses AI to optimize its advertising platforms.
Meta: 26% revenue growth despite Muse’s unprofitability and AI’s influence
Meta has had a challenging year, marked by multi-billion dollar lawsuits and litigation. One of them, related to the harmful effects of social networks on children, led to an $18 billion settlement. Another lawsuit, concerning data privacy violations, resulted in an order to pay the state of New Mexico $25 billion. This decision was made hours before Meta’s third-quarter report, and the company has already stated its intention to appeal it.
Despite these legal battles, the third quarter was strong for Meta. Revenue from the family of apps grew by 26% year-over-year, reaching $64.2 billion. Of this amount, $63.1 billion came from advertising revenue. The number of ad impressions also increased by 10%.
Muse: Popularity without Profit
Weeks after Muse’s launch in early September, Meta introduced Muse for small businesses. This is a “new set of skills and connectors” within the Muse app that helps companies find customers, analyze ad effectiveness, optimize it, and develop campaigns, as Meta CFO Susan Li told investors.
From a consumer perspective, the launch of Muse was “one of the most successful launches in the company’s history,” according to Meta CEO Mark Zuckerberg. This claim may be debatable given the privacy concerns associated with early versions of the model. However, as with other Meta platforms, privacy concerns have not hindered widespread adoption.
Zuckerberg noted that all these agents are “very computationally intensive.” Meta expects cost efficiency to improve 10x over the next year, which “will allow it to become profitable once revenue lines scale,” according to Zuckerberg. In other words: Muse is popular for now, but not making money.
If Meta’s ad revenue growth cannot be attributed to Muse, then what is the reason? The answer is artificial intelligence, which has long been central to Meta’s overall ad platform strategy.
Earlier this year, Meta launched Generative Recommender — an LLM-based tool that analyzes ad content and user preferences to determine the most relevant ads for a specific person. Target audience segments have seen an almost 2% increase in conversions since the tool’s launch in Facebook feeds, Meta CFO Susan Li said.
Meta also expanded GEM, its generative ad recommendation model, to include more user touchpoints, Li said. This means Meta can now unify recommendation data from various platforms in one place. This capability has led to:
A 1.7% increase in conversions for external campaigns.
A 9.6% increase in in-app conversions last quarter.
“AI investments have been benefiting Meta’s advertising business for many years. As LLMs become the foundation of more and more of the AI infrastructure, Meta will gain more insights into people’s preferences and interests, both in organic content and recommendations,” — Susan Li, Meta CFO.
The goal is not only to fuel Meta Advantage+’s ad optimization engine but also Meta’s creative personalization features.
Netflix: Revenue Growth and Advertising Challenges
Netflix’s revenue grew 12% year-over-year to $12.9 billion in the third quarter, meeting forecasts. However, the advertising business is slightly behind expectations, which led Netflix to adjust its forecast for the remainder of the year.
Frequently Asked Questions
What was Meta’s total revenue growth in the third quarter?
How much did Meta’s total revenue grow in Q3?
Meta’s total revenue grew by 26% year-over-year in the third quarter, reaching $64.2 billion.
How much ad revenue did Meta generate in Q3?
Meta’s ad revenue was $63.1 billion in the third quarter.
Is Muse profitable for Meta?
No, Muse is not currently profitable. Mark Zuckerberg noted that Muse agents are “very computationally intensive,” but Meta expects a 10x improvement in cost efficiency next year.
Which AI tools contributed to Meta’s ad revenue growth?
Growth was driven by the Generative Recommender, which increased conversions by 2%, and the expansion of GEM, which led to a 1.7% increase in conversions for external campaigns and a 9.6% increase for in-app campaigns.
How much did Netflix’s revenue grow in Q3?
Netflix’s revenue grew by 12% year-over-year, reaching $12.9 billion.
Conclusion
So, Meta is demonstrating strong revenue growth, primarily due to investments in AI, which significantly improve advertising performance. Products like Muse are currently unprofitable but have potential. While Netflix faces challenges in its advertising business, Meta continues to prove that favorable CPM and cheap hypothesis testing with AI optimization are a reality. Want to know how much money each ruble spent will bring? Learn how AI is changing the unit economics of advertising campaigns.
In modern marketing, efficiency is key to success. How many browser tabs do you have open right now? Probably quite a few. Transferring texts from one tool to another, searching for a brand book, coordinating with the team – all this takes up precious time and nerves. Imagine being able to shorten this path, getting a ready-made, brand-compliant video without leaving your familiar AI assistant or by fully automating the process. With Lumen5, this has become a reality thanks to new integrations with Claude, ChatGPT, Zapier, and Make, offering full-cycle video production.
Now you can order everything in one place: from video editing to posting, replacing an entire SMM team. This is not only budget savings but also a significant relief for your nerves. All created videos automatically appear in your Lumen5 workspace, where you can always make final edits before publishing. Artificial intelligence does the rough work, and you maintain control.
Comprehensive Short Video Production: Forget the Routine
Claude excels at a task that usually consumes hours: transforming lengthy reports, articles, or scattered notes into coherent, engaging text. Previously, after creating the perfect script, you had to transfer it to a video editor and start over. Now, thanks to the Lumen5 connector, this step is skipped.
Script Preservation: Lumen5 uses your script word-for-word for voiceover, ensuring full compliance with the approved text.
Context: Claude can summarize voluminous documents (reports, articles, notes) into a short script and create a video within a single dialogue.
Flexibility: The ability to work in Claude Code allows developers to create videos (e.g., for product updates) without leaving the terminal.
Customization: You can specify a particular template, AI voice, brand book, or leave the choice to Lumen5.
“If Claude is your tool for perfecting a long draft, then the integration with ChatGPT is the express lane.”
Lumen5 + ChatGPT: Video on Demand
Add the Lumen5 app to ChatGPT, describe your needs, and Lumen5 will write the script and create the video. One sentence is enough to start, without the need to pre-write a script or open a new tab.
Quick Start: A single-line prompt is enough to create a promo, a social media clip, or a review.
Using Existing Scripts: If you already have a script, simply paste it, and Lumen5 will use it for voiceover.
Voice Selection: Listen to various AI voice options by accent, gender, and tone to find the perfect one for your video.
Status Tracking: Ask ChatGPT about the video status or a list of recently requested videos, getting links when they are ready.
Branding: Integration supports brand books, ensuring that even the quickest requests comply with your corporate identity.
The Lumen5 app for Zapier allows you to automatically create videos when new content (posts, products, updates) appears, without the need for coding. This is an ideal solution for those looking for social media content outsourcing.
Triggers: Start video creation from your CMS, online store, CRM, forms, or spreadsheets.
Unified Style: Set up your workspace and style once, and all created videos will match your brand.
Easy Setup: Log in to Lumen5 via Zapier, grant access, and Zapier will take care of the rest.
Flexibility: Lumen5 can write a script for you if you provide post text or a brief description.
AI vs. Real Clipping Comparison: In this case, AI offers automation that can be significantly cheaper than manual clipping, especially for mass content production.
Lumen5 + Make: for advanced automations
The upcoming Lumen5 app for Make is designed for routing, filtering, and batch processing content into ready-made videos within existing scenarios. These are the quality criteria for mass posting, taken to the absolute.
Batch Processing: Create videos for each table row or feed item, with individual scripts, templates, or voices.
Content Type Branching: Send each item to a separate template, voice, or brand book. For example, a product launch will get a promo, and a “how-to” article will get a tutorial video.
Reliability: Using idempotency keys prevents the creation of duplicate videos when the scenario is run again.
Availability: Currently, you can use Lumen5 in Make via the HTTP module and the Lumen5 Video Creation API using an API key.
Is it worth unique content or new? With Make, you can automate the creation of unique videos for each product or article, which is significantly more effective than creating one general video.
Whichever integration you choose, the principle remains the same: your videos go into the Lumen5 workspace, use your brand books and templates, and open in the Lumen5 editor for final refinement.
Frequently Asked Questions
Can I create Lumen5 videos from Claude or ChatGPT?
Yes. Using the Lumen5 connector in Claude (including Claude Code) or the Lumen5 app in ChatGPT, you can request a video in a conversation and get a link to the finished video in minutes.
Do I need to write a script first?
No. Provide a brief description, and Lumen5 will write the script. If you already have a script, Lumen5 will use it word for word for the voiceover.
Will my videos be on brand?
Yes. You can request a specific brand book, template, AI voice, or blueprint from your workspace. In Zapier, you set up your workspace and style once when creating a Zap.
Can I edit videos created through integration?
Yes. Each video opens in the Lumen5 editor, allowing you to review and adjust it before publishing.
Who can create videos through integration with Claude?
Anyone who has permission to create videos in your Lumen5 workspace. Users with view or review permissions can view content but not create videos, just like in the Lumen5 app.
Can I automate video creation without code?
Yes. The Lumen5 app for Zapier allows you to trigger video creation from your CMS, store, CRM, forms, or spreadsheets without writing code or managing API keys.
Conclusion: How to choose a video content contractor
These integrations were created because we constantly heard the same thing: the difficulty is not in the ideas, but in their implementation, in transferring content from one tool to another. Lumen5 has come to where your ideas already live. If you are looking for a full cycle of short video production that will save you time and resources, and replace a team with a service, these solutions are for you.
Quality criteria when choosing a contractor include not only the final product, but also the efficiency of the process. Lumen5 offers exactly that: reduction of routine, automation, and preservation of creative control. Learn more or book a demo to see how teams turn articles, scripts, and ideas into polished, branded videos.
For those working with long documents: start with Claude.
For quick ideas: start with ChatGPT.
For automation when new content appears: start with Zapier.
For high-volume or multi-step processes: keep an eye on Make (or start today with the API).
In today’s world, where content is king, mass video posting becomes a key tool for audience reach. Imagine being able to publish dozens, or even hundreds, of videos a day across 10+ accounts while you sleep. This is not fantasy, but a reality thanks to automation and mass posting services. We will look at how these tools allow you to publish videos to 5 platforms simultaneously, using a single control panel, and how to choose the optimal solution for your tasks.
Why is mass video posting necessary?
Automatic publication of short videos significantly increases audience reach and saves resources. Instead of manually uploading each video to each platform, you set up a scheduled auto-post, and the content seamlessly spreads across all selected platforms.
This is especially relevant for projects requiring high publication frequency, such as news channels or bloggers working with trending content.
Saving time and resources
No manual labor: Forget about monotonous copy-pasting.
Scheduled reach: Videos are released exactly on time, even if you are busy or resting.
Unified panel: Manage all accounts from one place, be it TikTok, YouTube, or Instagram.
How to choose a mass posting service?
Choosing the right mass posting service is an investment in efficiency. It is important to consider functionality, cost, and integration capabilities. Key aspects to pay attention to:
Key selection criteria
Supported platforms: Make sure the service works with the platforms you need (TikTok, YouTube Shorts, Instagram Reels, Facebook Reels, Pinterest Idea Pins, etc.).
Auto-posting functionality: Flexible scheduling of publications, including bulk upload and content rotation.
Mass upload with proxies and anti-detection: For posting to 10 accounts or more, this is critical to bypass platform restrictions.
API integration: The presence of an API allows you to create your own automated action chains.
Monthly cost: Compare tariff plans. For example, the cost of 500 publications can vary greatly.
“While you sleep, videos are already being released – this is the principle of effective mass video posting,” experts in content automation note.
Implementation example: mass posting via API
For advanced users and large projects, mass posting via API opens up unlimited possibilities. This allows not only publishing content but also automating statistics collection, analytics, and even content generation.
For example, using the API, you can set up a system that will:
Track metrics and optimize the strategy of short video distribution.
Thus, impressive results can be achieved, for example, publishing 100 videos a day without spending hours of manual labor.
How to start with mass posting?
First, define your needs. How many videos do you plan to publish? To which platforms? What is your budget? After that:
Research services: Explore market offerings.
Set up accounts: Connect all necessary social networks to the chosen platform.
Create a schedule: Upload content and set up a schedule for 10+ accounts.
Monitor results: Track publication effectiveness and adjust your strategy.
Many services offer a mass posting package for a month, which allows you to test the functionality without significant costs.
Frequently Asked Questions
What is mass video posting?
Mass video posting is the process of automated publication of a large number of videos on various social media platforms simultaneously, often using specialized services and tools.
What platforms are supported for video cross-posting?
Most modern mass posting services support key platforms such as TikTok, YouTube (including YouTube Shorts), Instagram (Reels), Facebook (Reels), and Pinterest (Idea Pins).
How to choose the best service for automatic publication of short videos?
When choosing a service, pay attention to: supported platforms, scheduling flexibility, the presence of proxy and anti-detection functions, the ability to work via API, and pricing policy, including the cost of 500 publications or a mass posting package for a month.
Can I publish 100 videos a day using mass posting?
Yes, using a properly configured mass posting service and automation via API, it is quite realistic to publish 100 videos a day and even more, significantly increasing your short video distribution.
Conclusion
Mass video posting is a powerful tool for anyone striving for maximum audience reach and workflow optimization. Using automatic publication of short videos through a unified panel significantly saves time and resources, allowing you to focus on creating quality content. Start optimizing your strategy today so that your content works for you while you sleep.
The moment has finally arrived when the creator economy has received official recognition. On October 15, the U.S. Small Business Administration (SBA) issued a directive that classifies qualified content creators under the same industry code as motion picture and video production studios. This is a historic step, confirming what marketers have long observed: content creators are full-fledged media businesses. Ahead of budget formulation for 2027, internal categorization of creator partnerships will determine the amount of allocated funds, as well as methods for measuring effectiveness and contract terms, especially in the context of content repurposing.
New SBA Classification: What Does It Mean?
The SBA has assigned qualified content creators NAICS industry code 5121, pertaining to “Motion Picture and Video Industries,” which grants them access to SBA loans and contracts. The scale of the change is immense: the SBA counts 27 million businesses led by content creators in the U.S. alone. This is not just a formality; previously, creators had to choose unsuitable categories to obtain funding.
“Previously, a creator who shoots, edits, distributes content, and sells merchandise had to choose from industry categories that did not describe their work. Now there is a clear category for them to select on their application.”
This recognition places creators on par with major film studios and streaming services, which is an accurate reflection of their activities. Large creators operate as full-fledged production companies with staff, sets, and regular shows. Smaller creators are often “one-person bands,” performing all roles, from scriptwriting and filming to editing and distribution.
Why Is This Important for Brands and Marketers?
For a long time, brands viewed partnerships with creators as experiments or a tool to increase awareness, funding them on an ad-hoc basis. Now that content creators are recognized as media businesses, the same standards should apply to them as to other media partners: sales, contribution to the sales funnel, and incremental growth.
Budgets
Budgets for creators should be planned on par with budgets for paid advertising. According to the IAB, spending on advertising with creators reached $37 billion in 2025 and is growing four times faster than overall media spending.
Partnership Terms
Federal recognition, improved access to funding, and growing demand from brands give creators more leverage in negotiations. Influencers like Alix Earle are already reducing the number of partners, opting for more favorable terms. Long-term contracts or ambassador programs now are a chance to lock in current rates before they increase.
Contracts
If creators are businesses, contracts should be drafted accordingly. Working with the legal and procurement departments to update payment terms, scope of work, and content usage rights becomes critical.
How to Adapt Your Strategy for 2027: Checklist
Before finalizing your plans for 2027, three key checks need to be performed:
Budget: Plan creator expenses alongside paid advertising expenses. Evaluate them using the same sales and funnel metrics.
Partners: Identify key creators for long-term collaboration and offer them multi-quarter terms before their rates increase.
Contracts: Update contracts to include clear payment terms, scope of work, usage rights, and “whitelisting.” Ensure everything complies with the requirements of the legal department, as the status of creators as taxpayers or employees does not change.
Content creators have long functioned as media businesses, and now this is officially enshrined. Brands that adapt their plans accordingly will be better prepared for negotiations in 2027.
Frequently Asked Questions
What is content repurposing in simple terms?
Content repurposing is the process of transforming existing content into new formats or for new platforms. For example, turning a long webinar video into a series of short clips for TikTok, Reels, or YouTube Shorts. This allows you to scale content, reach new audiences, and extend its lifecycle without creating new material from scratch.
How does SBA recognition affect bypassing ad bans?
SBA recognition does not directly affect bypassing ad bans. However, when creators are perceived as full-fledged media businesses, it enhances their professionalism and work standards. This can indirectly contribute to the creation of higher-quality content that complies with advertising rules, which, in turn, reduces the likelihood of blocks. For video uniqueness and bypassing anti-plagiarism in advertising campaigns, technical solutions are still required, such as changing the video DNA, background, or voice timbre, to repackage creatives and create 100 unique ones from one.
Should you order a new video or unique an old one?
The choice between creating a new video and uniquing an old one depends on budget, goals, and resources. Ordering a new video from scratch can be costly and time-consuming. Repurposing and uniquing allow for significant savings by using existing material. For example, one webinar can be turned into 50 clips, adapted for different platforms and target audiences. This is an effective way to generate variations for A/B tests and scale content with minimal costs, especially when it comes to a uniquing package for mass launch of advertising campaigns.
What is “video DNA” and how to change it for uniquing?
“Video DNA” is a metaphorical term for the unique characteristics of a video file that can be used by algorithms to identify it. This includes metadata, visual and audio patterns, time codes, and other parameters. To change video DNA for uniquing and bypassing anti-plagiarism, various methods are used: re-encoding, changing resolution, adding or removing frames, changing playback speed, overlaying effects, changing the audio track, voice timbre, or background music. The goal is to deceive algorithms by creating a “clone” that will be perceived as new and unique content.
In a world where video content dominates all platforms, <a href="/ru/blog/algoritm-YouTube-shorts-2026-kak-narezat-strimy-dlya-virusnogo-uspekha/” title=”YouTube Shorts Algorithm 2026: How to cut streams for viral success”>podcast cutting and their video formats are becoming increasingly popular. However, despite the growing interest in video podcasts, traditional audio still plays a key role. The question arises: will audio be overshadowed while podcasts actively explore video formats and CTV?
Video Podcasts: Goldmine or Distraction?
Streaming services and even cable networks are actively licensing and producing their own podcast content. Their goal is to monetize through advertising and attract new subscribers. In this context, traditional audio finds itself in a familiar but uncomfortable position.
Maria Tullin, SVP of Audio Strategy and Podcasts at Horizon Media, argues: this will not happen. “I would look at video as a gateway,” says Tullin, who oversees all of Horizon’s audio projects with a focus on their effectiveness. She emphasizes that Horizon tailors audio investments to each client, but the main focus is on streaming and podcasting as the two largest channels.
Measuring Effectiveness: New Approaches and Old Problems
Maria Tullin notes that methods for measuring advertising effectiveness in audio have evolved significantly. If previously it was only promo codes and vanity-URLs, now the focus has shifted to pixel attribution. The main problem is YouTube. “We recommend clients participate in YouTube simulcast, but I cannot measure it as accurately,” explains Tullin.
Video Podcasts: Who is responsible for them?
For Horizon Media, video podcasts remain in Tullin’s team’s area of responsibility, especially when it comes to partnering with shows. “We handle almost everything related to the show,” she says. “Although sometimes due to overlap with traditional broadcasting and video, a podcast might end up on another team’s desk. We are not too fastidious about task distribution, but that doesn’t mean we stop doing it just because it’s video.”
YouTube: Brand or Performance?
Working with YouTube requires a special approach. Recommendations depend on the client’s goals. If the client is accustomed to pixel attribution, YouTube can be allocated a separate budget as a branding tool. For other brands focused on direct consumer contact and with social budgets, simulcasts may be more acceptable, as they are accustomed to a different type of data than pixels.
Podcast Budgets: Where Does the Money Come From?
Typically, funds for podcasts are allocated from the audio budget. However, there are instances when there is no audio budget at all. In such cases, clients may use an influencer budget to explore podcasts. In these situations, Horizon Media gets involved and, in the process, may suggest considering streaming as well.
Market Evolution: Advantages and Risks
“It’s a double-edged sword. On one hand, it’s great that people are showing interest in podcasts. On the other hand, I’m less thrilled that while many shows with strong audiences and a large video presence are great, they are not the only ones on the market,” shares Maria Thullin.
There are many talented creators with both video-first podcasts and audio-only formats that deserve attention. Thullin aims to interest clients in opportunities but warns against over-investing solely in big names. This leads to oversaturation, decreased advertising effectiveness, and a poorer user experience.
Advice for Aspiring Media Professionals
Maria Thullin emphasizes the importance of setting boundaries at work but calls for balance. It’s important not just to do your job, but also to strive to understand what’s happening at the level above. “Doing your job in isolation won’t get you further,” Thullin asserts. “Ask questions about what your manager is doing, offer help if you’re interested. That’s what leads to advancement because you show that you care about the next level, not just wanting a promotion for the sake of it.”
Frequently Asked Questions
What is a simulcast?
Simulcast is the simultaneous broadcast of media content (e.g., a podcast or stream) across multiple platforms, such as YouTube and Twitch or Libsyn.
How is podcast advertising effectiveness measured?
Traditionally, promo codes and vanity URLs were used. Now, pixel attribution is actively employed. For YouTube, this is more complex, and it is often viewed as a branding tool.
What budgets are allocated for podcasts?
Typically, these are audio budgets. In some cases, influencer budgets may be used, especially if the company does not have a separate audio budget.
Why shouldn’t you invest only in large podcasts?
Over-investing in popular shows can lead to advertising oversaturation, reduced effectiveness, and a poorer user experience, ultimately harming the entire industry.
What skills are needed for career advancement in media?
It’s important not only to do your job well but also to show initiative, be interested in the tasks of your superiors, ask questions, and offer help. This demonstrates a desire for development and an understanding of the bigger picture.
In conclusion, while video podcasts are gaining momentum, audio formats remain a fundamental part of the media landscape. The value of audio and its ability to attract an audience should not be forgotten. For those who want to delve deeper into the world of podcasts and their monetization, it is important to consider both video and audio, and to constantly adapt to changing methods of measuring effectiveness. Continue to explore, ask questions, and develop to stay at the forefront of the media industry.
LinkedIn, a platform known for its professional networking, has introduced a new and very intriguing way to monetize for experts – Advice Sessions. This is not just another update, but a real opportunity for specialists and influencers on the platform to earn from their knowledge through paid consultations. Forget superficial content; this is about real value and transparent monetization, opening new horizons for professionals.
LinkedIn and Monetization: A New Era for Experts
On October 15, LinkedIn introduced a number of significant updates. Among them, Advice Sessions stands out – a tool that allows professional influencers to schedule and conduct paid consultations directly within the app. This is not just a feature; it’s an entirely new niche for the digital arbitrageur who understands that CPM cheaper than 50 rubles for quality leads is no longer a fantasy, but a reality.
How do Advice Sessions work?
Direct consultations: Experts can schedule and conduct paid sessions.
Unique monetization: Unlike standard advertising integrations, this is a direct sale of expertise.
Transparent statistics: Every arbitrageur knows that without transparent statistics, there’s no point in starting. Here, it will be clear as day.
“LinkedIn has always been a platform for professional development, and now it offers a direct path to monetizing that expertise. This is not about paying an influencer for a post, but about buying a quality consultation from a verified specialist,” noted one market analyst.
Alternatives to Influencers: New Advertising Formats
In addition to Advice Sessions, LinkedIn continues to develop other earning methods for content creators. This shows that the platform is moving towards revenue diversification and offering solutions for different types of users.
BrandLink and Creator Marketplace
Previously launched programs, such as BrandLink, provide revenue-sharing opportunities by connecting brands with suitable creators. Creator Marketplace scales these relationships. However, these tools have much in common with what other platforms offer.
In-Stream advertising: The ability to place ad inserts in video clips.
Programmatic video banners: Allows you to buy reach without creating a video, using a native banner.
Cheap reach: Ideal for promoting mobile applications or advertising crypto projects.
Frequently Asked Questions
What are Advice Sessions on LinkedIn?
Advice Sessions is a new LinkedIn feature that allows professional experts and influencers to conduct paid consultations directly through the platform, monetizing their knowledge and experience.
How does LinkedIn help monetize content?
LinkedIn offers several tools for monetization, including BrandLink for brand partnerships, Creator Marketplace for scaling these relationships, and the new Advice Sessions format for direct paid consultations. New video advertising formats are also being introduced.
Can banner ads be placed on LinkedIn without shooting a video?
While Advice Sessions focus on direct consultations, the announced new video advertising formats may open up opportunities for programmatic video banners, which will allow you to buy reach without creating a video, by placing a native banner in Shorts or other video formats, similar to advertising in YouTube Shorts with low CPM.
Conclusion: The Future of Monetization on LinkedIn
LinkedIn continues to evolve, offering new, more sophisticated ways for its users to monetize. From Advice Sessions to new video advertising formats – the platform is becoming not just a place for networking, but a full-fledged tool for earning. For the digital arbitrageur, this is a signal: it’s time to test hypotheses for 1000 rubles and master these new channels. Don’t miss the chance to order banner advertising or become an expert who actually earns, not just posts “I’m happy to announce…”. This is a real opportunity to get mass reach for pennies and reach a new level of profit.
Google released the Nano Banana 2.1 update on October 6, 2026, promising significant improvements in image generation. The new neural network is designed to better follow prompts, generate more accurate text, ensure character consistency, and work with wide formats. We checked how these promises align with reality by comparing both versions on five identical tasks. The short answer: for new projects, it’s worth choosing Nano Banana 2.1 right away, as it more often accurately fulfills minor requirements and less frequently adds unnecessary elements. However, the update hasn’t solved all problems, especially when transferring appearance between scenes, where the old version sometimes preserved features just as well. This is an important point for those planning to scale AI video and image generation.
Technical Features and Version Costs
Nano Banana 2 operates under the technical name gemini-3.1-flash-image, while Nano Banana 2.1 is based on Gemini 3.6 Flash and is named gemini-nano-banana-2.1. Google claims improvements in realism, instruction understanding, text rendering, and multi-step editing. The new model can accept up to 14 references: up to four characters and up to ten objects.
According to Google’s official pricing, a 2K image in Nano Banana 2.1 costs approximately half as much as in Nano Banana 2 ($0.0504 versus $0.101), making it attractive for AI content production. At the same time, the 0.5K mode has disappeared from 2.1, and Nano Banana 2 is already marked as an outdated model.
Testing Methodology: 5 Prompts for Comparison
For an objective comparison, we relied on the open Fuser test from October 6, 2026. In it, both models received identical English-language prompts and source data, and images were generated in 2K with one attempt without restarts. This allows us to evaluate the effectiveness of AI under controlled conditions. If you decide to repeat the test, use the same settings and do not select the best results to avoid distortions.
Prompt 1: Photorealistic Scene with Precise Requirements
“Photorealistic documentary photograph of an elderly fisherman mending a bright orange net on an old wooden pier at dawn. In the background, a foggy harbor. The fisherman’s hands are in sharp focus, the background is softly blurred. Natural light, 50mm lens effect, muted color palette. Do not add text or inscriptions to the scene.”
Nano Banana 2.1 kept the cool misty background, showed the fisherman in full, and left the frame free of extraneous text. Nano Banana 2 made the scene warmer, more saturated, and added a fictional inscription on the boat. Here, 2.1 won due to more precise adherence to the constraints, which is critical for AI video for advertising.
Prompt 2: Poster with exact text
“Create a minimalist concert poster on light paper. The top half is occupied by a large dark blue wave. Place four lines of text: “NIGHT TIDE” “Live at Harbour Hall” “Friday 14 November · Doors 8 PM” “harbourhall.com/nighttide” Do not add any other words. Make the first line a large narrow sans-serif, the second a lighter typeface.”
In this test, the difference between the versions was minimal. Both models correctly wrote the specified lines, including the date and address. This shows that in some tasks where Nano Banana 2 already performs well, an update may not provide a noticeable leap in quality, especially in text rendering.
Prompt 3: Product photography with space for ad text
“Studio photograph of a matte ceramic sage-green coffee funnel on a light oak board. Soft window light from the left, a light shadow should fall to the right. Clean, almost white background. Leave the right third of the frame free for ad text. Realistic product photography. No additional objects or inscriptions.”
Nano Banana 2.1 fulfilled all conditions: light from the left, shadow to the right, and a free area for text. Nano Banana 2 also left space, but divided the image with an empty light area, and the shadow direction was less pronounced. For creating AI commercials and banners, this is a significant advantage, saving designer time.
Prompt 4: Transferring a product to another location without alteration
“Place this same bottle on a wet, moss-covered stone. The scene is outdoors in the morning, with soft sunlight and a blurred forest stream in the background. Change only the environment and lighting. Do not change the bottle’s shape, amber glass, black dropper cap, or the inscription “LUMA 02”.”
Both versions successfully preserved the major features of the bottle (shape, glass color, text). However, both violated a minor condition by changing the smooth detail of the cap to ribbed. 2.1 chose a wider shot and kept the label clean, whereas Nano Banana 2 zoomed in on the object and added water droplets to the label. This demonstrates a limitation of image-to-image models: the phrase “change nothing” does not guarantee pixel-perfect identity of small details.
Prompt 5: Transferring a character to a new scene
«Use the woman from the reference portrait. Show her full-length riding a bicycle on a wet city street at night. Neon signs all around, she’s wearing a yellow raincoat with the hood down. Keep the same face, red curly hair, freckles, and round glasses. Photorealism.»
There is no clear winner here. Nano Banana 2 preserved the curls and glasses well but changed her facial expression to a wide smile. Nano Banana 2.1 framed the full-length shot more accurately but weakened the resemblance in details: her hair became less curly, the frames thinner, and freckles almost disappeared. The choice depends on priorities: either precise preservation of individual facial features or a natural composition of the new scene. This is especially relevant for animated AI characters and AI avatars.
Conclusions after five prompts: is it worth switching?
According to the test results, Nano Banana 2.1 looks like a careful refinement of the previous version, not a fundamentally new model. Its advantage is evident in tasks with many specific requirements, which aligns with Google’s official description of better prompt adherence and text rendering. For AI UGC for crypto projects, where detail accuracy and prompt following are critical, this is a significant plus.
Nano Banana 2.1 won two tasks and did not clearly lose to the old version in any.
When transferring appearance, the results remained ambiguous.
If you are starting a new project, it is more logical to use 2.1 immediately, especially given the reduced cost and Nano Banana 2‘s status as deprecated.
If you have a working pipeline on Nano Banana 2, test your typical prompts on both models to ensure the desired generation style is maintained.
Frequently Asked Questions
Is Nano Banana 2.1 really better than Nano Banana 2?
According to official specifications, 2.1 is the recommended version with improved instruction following, quality, text, and character stability. In our test, it showed better results in tasks with multiple constraints.
Why does the same prompt give different results?
Image generation in neural networks is not deterministic by default. Even within the same model, repeated runs can change the composition, face, lighting, or small objects. For a correct comparison, a series of runs with identical settings is required.
Why does the model ignore part of the prompt?
Often the problem arises when there are too many equally important instructions in one request. For critical constraints, it is better to use separate instructions, for example: «change only the background», «the text must be written verbatim», «do not add logos», «keep the face unchanged».
Should I always include high Thinking in Nano Banana 2.1?
No. 2.1 has minimal, medium, and high levels. A higher level can be useful for a complex layout or a large number of constraints, but it usually increases computation volume and execution time. For simple drafts, it is logical to start with minimal or the standard medium.
What Nano Banana 2 has that 2.1 doesn’t?
The most noticeable technical difference is the 0.5K generation mode. In 2.1, the minimum resolution starts at 1K. However, the new version is cheaper at matching resolutions and adds extremely wide and vertical formats.
How to improve results when working with a reference?
For the best results, use high-quality references. Make sure the reference images are clear and contain all necessary details. Experiment with the number of references (up to 14 in 2.1) to find the optimal balance.
How to access Nano Banana in Russia?
Nano Banana is available on special platforms that work without VPN and foreign cards. You can upload photos, change backgrounds and clothes, collect multiple references, and create images from text descriptions. Services usually work with ruble payments and provide a common balance for different models. Use promo code HABRSPESHUAI to increase your balance by 115%. This is a great way for crypto projects to order AI creatives or scale the generation of 100 videos per day.
Overall, Nano Banana 2.1 is a significant step forward, especially for tasks requiring high accuracy and detail. Its reduced cost and improved capabilities make it an attractive tool for subscription-based neuro-production and AI content production. Don’t miss the opportunity to scale the impossible – now it’s possible with new Google neural networks.
Hello, fellow arbitragers! My name is Timur, and I am responsible for the marketplace at Cloud.ru. We provide a platform for AI services that integrate into clients’ cloud infrastructure. My task is to ensure pricing transparency. Currently, many evaluate AI by token cost, but this is fundamentally wrong. The client pays not for tokens, but for a specific result. In this article, we will analyze how to correctly calculate the unit economics of an <a href="/ru/blog/massovyy-posting-video-4-strategii-povtoreniya-zadach-ii-agentami/" title="Mass video posting: 4 strategies for repeating tasks by AI agents">AI agent, based on Cloud.ru data, and determine the real cost of a useful response to understand what CPM is profitable.
“A vendor on the marketplace must set a price for their agent, the client must understand it, and the platform must make it possible to compare prices on the storefront. As long as all three parties measure cost in tokens, none of this works.”
Evaluating the Cost of AI Services: Focus on Unit Economics
Unit of Measurement: Useful Response, Not Tokens
A token is a bad metric. The client buys not tokens, but a solution to their problem. A “closed dialogue” is also deceptive. Intercom, for example, distinguishes two types of solutions: confirmed (the client said “thank you”) and assumed (the client simply left).
Both are paid for, but there’s a nuance: if the client returns with the same question, the solution is annulled. An example from Lorikeet (195 cases, 38 vendors): out of 4200 dialogues closed by AI, 900 were closed with a link to a knowledge base article. Without them, the share of resolved issues drops from 70% to 55%.
The difference between “closed” and “resolved” is huge. Therefore, our unit of measurement is a useful response. This is a dialogue closed by AI without human intervention, where the client did not return with the same question within 72 hours and did not leave a negative rating. Everything that did not pass these filters goes into the numerator of costs, but not into the denominator of useful responses. Unsuccessful attempts are paid for by successful ones. This is a key point.
Model Case: Online Store Support
For calculations, let’s take a typical scenario: online store support. This is a mass case with a large volume of public data. The store receives 40,000 inquiries per month. 90% of them (36,000 dialogues) are routed to the AI agent.
Dialogues closed by the agent without an operator, let’s call them auto-closures – this is the metric that vendors show. Agent structure: RAG (knowledge base search + response generation) plus the “order status” tool. One dialogue accounts for three LLM calls:
Understand the request and choose a route.
Respond based on the found fragments.
Refer to the tool or compile a summary for the operator.
On average, this is 12,000 input tokens (system prompt, fragments, history) and 800 output tokens. There are two knowledge base queries per dialogue. The filter model checks the client’s reply and the agent’s response (1000 tokens for each of the three calls). Another 15% of tokens are allocated for re-generations.
Infrastructure and Costs
We use Cloud.ru services:
Knowledge Base: Evolution Managed RAG.
Agent: Evolution AI Agents.
Logs: Evolution Managed PostgreSQL.
Cost of a support operator in Moscow: 125,000 RUB/month (85,300 RUB salary according to hh.ru for 2025 + contributions and overheads). With 40 tickets per day, one ticket costs 149 RUB. Operator minute – 15.5 RUB (125,000 RUB / 80% of 10,080 minutes). For chat in retail (60 requests/day), the ticket cost is 99 RUB.
Assumptions for Calculations
Key assumptions:
Share of auto-closures: median for public cases – 70%. Ozon: 27% (2020) -> 62% (2022). We take the middle – 50%.
Dialogues that did not pass the filter: client returned within 72 hours or gave a negative rating. Assumption: 6% (in the Lorikeet example, 21% of closed dialogues were links).
Cost of such a dialogue: 150 RUB compensation + 3 minutes of operator time (200 RUB).
Operator time for escalation: 1.5 minutes beyond a regular ticket.
Manual check: 3% of all dialogues for 3 minutes each.
Total infrastructure: 13,420 RUB/month. This is the cheapest line in the calculation.
Calculation of the Cost of a Useful Response
Cost of one dialogue
First, we calculate the cost of one dialogue using different models:
Qwen3-30B-A3B: 0.43 RUB
GigaChat-3-Pro: 1.32 RUB
Claude Haiku 4.5: 3.76 RUB
Claude Sonnet 4.6: 10.99 RUB
A 25-fold difference! If we stopped here, model selection would be critical.
Cost of a useful response
Out of 36,000 dialogues, with 50% auto-closures, 18,000 were closed. Of these, 6% were errors. This leaves 16,920 useful responses. Each useful response incurs not only token costs but also:
Tokens from failed attempts.
Operator minutes on escalations.
Cost of errors.
Verification.
Knowledge base.
Implementation.
The 25-fold difference in tokens translates to a 1.4-fold difference in useful response cost: from 51 RUB to 73 RUB. The expensive model (Claude Sonnet 4.6), which is 25 times more expensive in terms of tokens, turns out to be only 40% more expensive in terms of result cost. Why? Because all other costs are the same for all models.
Price structure: GigaChat-3-Pro (53 RUB)
For GigaChat-3-Pro (53 RUB per useful response):
Escalations and erroneous responses: 71% of the cost. These costs are usually not accounted for as “AI expenses” but are embedded in support salaries or compensation.
Tokens: 2.8 RUB (5%). This is the only line item everyone compares.
Cloud infrastructure: 0.80 RUB.
For Claude Sonnet 4.6, tokens already account for 32% of the price. Here, it makes sense to save on tokens. How?
Context reduction: reduce 12,000 input tokens to 6,000. The useful response on Sonnet becomes cheaper from 73 to 65 RUB.
Caching: reading from cache is 10 times cheaper. If half of the context is a system prompt, the price drops to 66 RUB.
On Qwen, these techniques save pennies.
Key levers for savings
1. Auto-closure rate
Transitioning from a pilot (35-50% auto-closures) to a production environment (65%) saves 15-19 RUB. Ten percentage points of auto-closures yield as much as switching to a cheaper model (on Sonnet, an increase from 50% to 60% saves 14 RUB, replacing Sonnet with Haiku saves 15 RUB). But the former works on any model, the latter – only once.
2. Error rate
Few measure the percentage of closed dialogues that failed the filter, but it is critical.
Percentage of failed filters
Cost of useful response
Expenses for them (RUB/month)
6%
X
Y
3%
X – 8 RUB
Y – Z
Reducing errors from 6% to 3% saves 8 RUB on each useful response. Switching from GigaChat-3-Pro to Qwen3-30B-A3B saves only 2 RUB. 15% errors can cost more than half of an operator’s salary if every dialogue where the client went silent is considered resolved.
3. Escalation quality
I allocated 1.5 minutes of operator time for escalation (reading summary, continuing). This works with a “warm” handover: the agent provides the operator with a concise history, order number, and list of actions. With a “cold” handover (the operator reads the entire dialogue where the bot answered incorrectly three times), 1.5 minutes turn into 3 or 6.
At 3 minutes, a useful response on GigaChat-3-Pro increases in price from 53 to 77 rubles, at 6 minutes – up to 127 rubles. This is already more expensive than a regional operator. If the agent identifies the client and pre-fills the ticket, the operator saves a minute or two, and this line goes into the red.
“Transmission quality turns out to be a separate budget item with the widest spread in the entire model, and it should be measured: the average processing time for escalated tickets should be calculated separately from regular ones.”
Fighting for accuracy = burning tokens? If, to reduce errors from 6% to 3%, tokens per dialogue increase by 1.6 times, on GigaChat-3-Pro, a useful response still becomes cheaper from 53 to 46 rubles. On Sonnet, extra tokens are more expensive than the accuracy gained; it’s better to improve the knowledge base there.
Conclusions: what to consider a profitable CPM
Practical conclusion: in the working support circuit, the most expensive thing is not tokens, but the denominator. Everything that increases the proportion of useful responses (knowledge base quality, correct routing, test sets) pays off faster than any bargaining for the price of a million tokens.
Expensive models (e.g., Sonnet) pay off if they bring 10 points of auto-closures and 2 points of accuracy compared to cheaper ones. If the difference on your test set is less, you are overpaying. If it’s more, you are unnecessarily saving on tokens.
Comparison with a human: two caveats
1. Ticket price: at 149 rubles per ticket, even a pilot with 35% auto-closures on Sonnet (110 rubles per useful response) looks profitable. Parity with an operator is at 20-27% auto-closures. But if an operator closes 60 tickets a day (99 rubles/ticket), the same pilot on Sonnet (110 rubles) is already more expensive. A regional/outsourced operator (64 rubles/ticket) changes the picture: GigaChat-3-Pro (42 rubles) is cheaper, Sonnet (62 rubles) – parity. The question of “will it pay off” depends on your ticket cost and the proportion of auto-closures.
2. Volume: fixed costs (knowledge base, implementation, infrastructure) – 163,000 rubles/month. For 36,000 dialogues. For 6,000 dialogues, a useful response on GigaChat-3-Pro is 101 rubles, for 3,000 – 159 rubles (more expensive than an operator). Below 3,500 dialogues per month, the agent does not pay off. If fixed costs are higher (a full-time position for the knowledge base – 200,000 rubles/month, implementation – 1.5 million rubles), the break-even point shifts to 7,000 dialogues per month. The share of tokens in this case drops to 4%: the more realistically you calculate other costs, the less significant the price per million tokens appears.
What is not accounted for in the model (in both directions)
Against the agent: ticket system or chat platform license.
In favor of the agent: staff turnover (up to 50% per year), recruitment and training costs, slow work of newcomers, scaling for sales, payment for night/weekend shifts. These things are difficult to calculate without specific data, but businesses are often willing to pay for them, even if daily unit economics show parity.
Frequently Asked Questions
What is a “helpful answer” in the context of an AI agent?
A helpful answer is a dialogue closed by AI without human intervention, where the customer did not return with the same question within 72 hours and did not leave a negative rating. This is a key metric for evaluating the real effectiveness and cost of AI services.
