Blog// AI agents

Instinct vs. Muse and Dots: Comparative Analysis of Next-Gen AI Frameworks

Instinct’s text-based AI agent competes with Muse and Dots by turning conversations into actions. Here is what small businesses should know about its uses and risks.

09-10-2026 · 9 min read
Instinct vs. Muse and Dots: Comparative Analysis of Next-Gen AI Frameworks

The Verge reports that in August, startup Instinct launched its AI agent without traditional marketing, a website, or open sign-ups: access was invitation-only. Despite this minimalist launch, the agent quickly became one of the most talked-about products in AI. Its appeal lies in a simple text interface and an ability to handle routine tasks—making an appointment at the DMV, sending reminder emails, or signing a child up for swimming lessons. Before long, major players introduced alternatives: Muse from Meta* and Dots from OpenAI. The Verge notes that even under this pressure, Instinct held its own in the publication’s tests, and its founder, Noah Shinn, remains confident in the strategy.

For the owner of a small shop, service business, or manufacturer in Russia and the CIS, this matters because AI agents are moving beyond experiments and beginning to offer real time savings on routine work. Understanding how these tools work, what their limitations are, and how they might make money in the future can help you decide whether to experiment now or wait for more mature solutions.

How Instinct Works and How It Differs from Competitors

Instinct presents itself as a personal assistant that communicates exclusively through text channels such as iMessage, WhatsApp, or ordinary email. The Verge emphasizes that the agent has no separate app and no bear avatar or “cute blobs” like the ones Muse and Dots use to attract attention. Users receive concise replies, sometimes with emojis, and can immediately see the result of an action—whether that is a calendar booking or an email sent.

The agent connects to external services through a “secure vault.” Users enter a login and password once, after which the agent can connect to Google Workspace, Slack, Notion, and other platforms. The Verge notes that without this step the agent cannot perform actions that require authentication, and that entering credentials remains the only point of friction in an otherwise seamless text-based interaction.

According to Shinn, the absence of a conventional interface is a key advantage. AI platforms are usually excellent at answering questions, but Instinct focuses on turning a conversation into a concrete action, imitating the experience of messaging a real assistant.

What The Verge’s Tests Found

In its own tests, a Verge journalist asked Instinct to handle several everyday tasks: find swimming lessons for a child, book an appointment with an eye doctor, and arrange a return at Ikea. The agent completed most of them and showed that it could interpret vague requests—for example, identifying the right item from a description as “vertical, white, with branch-like elements” and finding its correct name, Tjusig.

The agent’s proactive help was particularly noteworthy. While trying to book a doctor’s appointment, Instinct independently found a confirmation email from 2024 and pointed out a possible calendar conflict, even though the user had not explicitly asked it to check. The Verge describes this kind of foresight as a real step toward making AI assistants genuinely useful.

Significant limitations remain, however. To access some websites, the agent provides a link to a secure vault where users must enter their credentials manually. The AI can also work only with information that is publicly accessible. If important details are hidden in a pop-up or require extra clicks, it may miss them. When arranging swimming lessons, for example, Instinct failed to see a prerequisite course listed in a pop-up, leaving the user to catch the omission.

Business Model and Future Monetization Risks

At the time of The Verge’s report, Instinct was free for users who had received an invitation or joined the waitlist. Shinn said he intended to keep the service accessible for as long as possible, while acknowledging that the company’s current goal was growth. Avi Greengart of Techsponential notes that a venture-backed startup like Instinct typically first seeks to build a user base and then looks for ways to monetize it—through direct payments or by selling to a larger company.

The contrast with competitors is clear: Dots from OpenAI is available only to paid subscribers, while Meta* offers Muse for free. Gregor Airland of IDC believes monetization is inevitable in the long run, particularly given these agents’ focus on transactions. The Verge says Instinct has already experimented with showing users “recommended products,” which could be a prelude to an advertising model. Airland sums up the risk: all roads in the consumer market eventually lead to advertising. Users who do not pay directly may therefore pay indirectly, through sponsored recommendations or the sharing of data with partners.

For small businesses, this means that free access today may not last. If an agent starts showing sponsored suggestions, an owner should consider whether those recommendations are relevant to the business and whether they might distract or irritate customers. Changes to access terms also matter: a move to subscriptions could put the tool out of reach for anyone unwilling to pay.

What This Means for Small Businesses in Russia and the CIS

Where This Kind of Agent Could Be Used

For a retail shop or service business, an AI agent could take on routine customer communication: confirming appointments, sending reminders about upcoming visits, and answering simple questions about product availability or opening hours. Because Instinct works through familiar messaging services, customers do not need to install another app; they can simply send a message in a chat they already use.

For a small manufacturer, an agent could help coordinate internal work: synchronizing calendars, automatically sending reports to Slack or Notion, or preparing standard emails to suppliers. The secure-vault integration makes it possible to connect a corporate Google Workspace account without handing passwords directly to the agent.

Risks and Potential Pitfalls

The first risk is dependence on the quality of external data. If information about a service, schedule, or product is buried on a complicated web page with pop-up hints, the agent may not see it. This could lead to booking or ordering errors. Owners should verify the agent’s critical conclusions, especially when financial transactions or legally significant documents are involved.

The second risk is future monetization through advertising or a change in access terms. If an agent starts inserting recommended offers, users may receive irrelevant ads, undermining their trust in the service. If the product becomes paid, businesses will need to revisit their subscription budgets and compare the cost with the time they expect to save.

The third risk is data security. The Verge does not disclose the details of the encryption protocols, but entering a login and password into a “secure vault” requires trust in the agent provider. A business that handles customers’ personal data should ask Instinct exactly how credentials are stored and protected and, if necessary, limit access to non-confidential services.

Practical Steps to Take This Week

  1. Request an invitation. Visit the Instinct website if one becomes available, or join the waitlist through existing channels, so you can evaluate the interface without a financial commitment.
  2. Choose two or three use cases. For example, automatically confirm appointments through WhatsApp and export tasks to Notion each week. Be specific about which actions should happen without human involvement.
  3. Set up a test connection. Once you have access, add one work account—Google Workspace or Slack—to the secure vault and check how the agent creates a calendar event or sends a template email. Record the setup time and compare it with doing the task manually.
  4. Assess accuracy and proactivity. Give the agent a task with an incomplete description—for example, find the nearest parcel pickup point with particular opening hours—and see whether it offers useful information you did not request. Note any details it misses because they are hidden on a website.
  5. Schedule a review of the terms. After a week of use, check Instinct’s updated FAQ or blog for changes to its access policy and any mention of future fees or advertising. If monetization appears imminent, decide whether the business is willing to pay or should look for an alternative.

Conclusion

The Verge’s reporting shows that Instinct has managed to hold on to a niche despite the entry of major competitors, thanks to a simple text interface and a focus on turning conversation into action. For small businesses in Russia and the CIS, this approach promises real time savings on routine tasks, but it calls for careful attention to data quality, possible changes to access terms, and security. Start with a small pilot, assess the results, and only then decide whether to integrate the agent more deeply or look for other solutions.

Frequently Asked Questions

What is the Instinct AI agent?

Instinct is a startup that released its AI agent in August without traditional marketing. Access to the product was via an invitation system and without an official website.

Why did Instinct abandon traditional marketing?

The decision may have been a strategy to create exclusivity and buzz. This approach allows gathering a target audience and obtaining quality feedback at early stages.

What does this launch mean for small businesses?

The launch of Instinct demonstrates alternative ways to bring a product to market. Small businesses can consider non-traditional promotion methods, focusing on niche communities and viral effects.

  • Meta Platforms Inc. (including social networks Facebook and Instagram*) is recognized as an extremist organization and its activities are prohibited in the Russian Federation.

Based on: The Verge

Analysis & review by
Alexander K.

Web development, digital advertising and marketing expert since 2012. Founder of AIFORSALE.

How we can help
Online storesMarketingAI agentsContact us
← All articles