TL;DR
· Anthropic launched Slack’s @Claude beta in June, directly entering the AI employee application layer market.
· Viktor claims it bought Claude Tag search terms, losing 5 customers that day but gaining 407 new registrations.
· The giant’s release validated demand but also shifted enterprise customer competition toward permissions, compliance, and context control.

On June 23, Anthropic launched the native AI agent for Slack, Claude Tag, directly colliding with the startup Viktor, which focuses on "AI employees." However, internal data later disclosed by Viktor’s founder showed that this seemingly dangerous competitor release actually became one of the strongest growth moments for Viktor since its launch.

The reason this conflict attracted attention is that both sides are vying for the same enterprise entry point: putting AI into collaboration tools like Slack and Microsoft Teams, not just to answer questions, but to read team context, receive tasks, call tools, and execute work asynchronously—like a permanent "AI colleague" in the company workspace. If foundational model companies like Anthropic and OpenAI start building application-layer products themselves, the startups that rely on their model ecosystems will face a key question: are they being squeezed or validated?

After Being “Copied” by a Giant, First Buy Their Search Terms

Viktor founder Fryd was not surprised by Anthropic’s entry. In his response article, he wrote that from day one of the company, almost everyone asked the same question: What if Anthropic launches a competing product?

Viktor’s answer was not to block it, but to leverage the attention it brings for customer acquisition.

On the day of Claude Tag’s release, Viktor said it launched Google Ads brand campaigns, bidding on keywords related to "Claude Tag," directing users searching for terms like Claude Tag, multi-agent, and AI employee to Viktor’s own landing page. The founder’s statement was direct: Anthropic educates the market, and Viktor captures demand.

This is a typical launch-day intercept strategy. For a small company, the challenge is often not just building the product, but making the market aware the category exists. Previously, few people actively searched for concepts like "multi-user collaborative AI employee." When a large frontier model company releases a similar product, search demand is concentrated and created.

Viktor claims that day did not see the customer exodus many had imagined. The company proactively encouraged some customers to try Claude Tag and then decide whether to switch. According to the founder, Viktor lost 5 customers on launch day while gaining 407 new registrations. The company did not disclose the conversion rate of these registrations into paying users.

This is the most interesting contrast in the conflict: a giant launching a competing product is indeed a threat, but the attention it brings can also serve as a free demand generator for small companies.

Viktor Says It’s Not the Same Product as Claude Tag

Viktor repeatedly emphasizes that the two are not identical products.

Anthropic’s official page shows that Claude Tag is currently available for Claude Enterprise and Team users, allowing them to tag Claude in Slack, enabling Claude to read thread context, execute long or scheduled tasks, and manage permissions via Agent Identity. TechCrunch and TechRepublic also reported that the product is still in beta, focusing on integrating Claude into Slack workflows.

Viktor’s comparison article describes Claude Tag as a Slack-only beta, tied to Claude Opus 4.8, with approximately 14 connectors, no free tier, and output primarily based on recent conversations in Slack channels. Since this comparison comes from a direct competitor, the functional differences should be seen more as Viktor’s product positioning statements rather than neutral evaluations.

Viktor positions itself as a more complete "AI employee": supporting both Slack and Microsoft Teams, connecting over 3,000 tools, and allowing users to choose different models. According to official information and Accel, Viktor operates within enterprise collaboration tools, aiming not just to reply in threads but to deliver presentations, dashboards, spreadsheets, or ad campaigns, and execute tasks via a browser.

This narrative addresses a survival question: when a model company builds a similar application, can the application-layer company retain users through narrower positioning, faster iteration, and more complete workflows?

The founder uses the example of Ford and General Motors to explain his judgment. A single standardized product may not eat the entire market; different customers require different forms. In the AI employee scenario, agents in Slack, cross-tool employees, developer-oriented components, and button-like products for business teams may correspond to different buyers.

Viktor’s key message is that Anthropic’s entry is not the endgame, but rather pushes a previously vague category into the market’s spotlight.

The Real Disagreement Lies in Enterprise Context

Beyond product features, Viktor focuses competition on two layers: model choice and enterprise context.

The first layer is the model. Viktor argues that Anthropic’s own application is naturally tied to the Anthropic model. Once OpenAI, Google, or other vendors release stronger models, independent applications can switch their underlying models, while the model company’s own products struggle to escape such lock-in.

This is not a new issue. Cursor, Perplexity, Granola, and other AI application companies face a similar situation: they use the capabilities of large model companies while competing with them at the application layer. A common defense from application companies is that they are not just "model shells," but integrate multiple models, product experience, workflows, and customer needs into an end-to-end product.

The second layer is enterprise context. Viktor believes that in the future, what is truly hard to migrate is not the model itself, but the operational memory that enterprises deposit in the AI system, including customer commitments, exception flows, historical attempts, team preferences, tool permissions, and cross-project knowledge.

If this context is locked into a particular model vendor’s agent layer, companies may not be renting intelligence but rather buying back their own operational memory from the supplier. Viktor therefore describes itself as "renting the best intelligence, but owning your own context."

This line of reasoning carries an obvious defensive tone. Foundational model companies not only have models, but also brands, capital, enterprise client relationships, and distribution channels. By placing Claude Tag within the Claude product ecosystem, Anthropic might siphon self-service customer acquisition from application companies, but in enterprise sales, the broader Claude brand could also bring a trust advantage.

A Giant Validates the Track, but Also Squeezes Enterprise Orders

Viktor does not frame this as an unqualified victory.

In his response, the founder admits that the self-service market can leverage advertising and traffic strategies to ride the wave of growth, but the enterprise market is more complex, and enterprises are where most global knowledge work happens. For enterprise customers, purchasing an AI employee is not just about trying a tool; it involves security, permissions, compliance, integration, data retention, and long-term vendor selection. In these areas, large model companies have natural advantages.

He also says bluntly that Anthropic is Viktor’s biggest competitor and the greatest threat to its mission. The launch-day new registrations are not enough to change that assessment.

This makes the conflict a microcosm of AI application-layer startups: when foundational model companies not only sell the "engine" but also start building the "whole car," downstream companies are forced to prove they are not a replaceable interface. They need faster customer acquisition, deeper integration into workflows, better multi-model integration, and the ability to convince customers that their context and business memory will not be locked into a single model ecosystem.

Anthropic’s launch of Claude Tag did bring traffic and category validation to Viktor. But the same move also pushed the company into more direct competition. The giant helps educate the market, but at the cost of also competing for the market. For startups like Viktor, the short-term issue is not whether they get killed, but whether they can convert launch-day hype into long-term retention and real enterprise orders.

Click to learn about the job openings at BlockBeats

Welcome to join the official BlockBeats community:

Telegram Subscription Group: https://t.me/theblockbeats

Telegram Discussion Group: https://t.me/BlockBeats_App

Twitter Official Account: https://twitter.com/BlockBeatsAsia