Instinct and Meta’s Muse Bring Outbound Calling to AI Assistants

Instinct and Meta’s Muse are adding outbound calling, pushing AI agents beyond screens into reservations, support calls, and other high-touch tasks.

AI News

Instinct and Meta’s Muse are adding the ability to place phone calls, bringing a previously differentiating capability into the increasingly crowded market for consumer AI assistants. Instinct is rolling out Instinct Concierge in early access, while Meta is expanding outbound calls to U.S. businesses for selected Muse users.

The change matters because phone calls remain a barrier for many everyday tasks, including restaurant bookings, cancellation requests, and customer-service disputes. By handling those conversations on a user’s behalf, both products are attempting to move from answering prompts to completing tasks across channels.

Instinct turns calls into a concierge feature

Instinct founder Noah Shinn announced the new calling capability on X, according to TechCrunch. The company says Instinct Concierge is designed for “high-touch” cases that are difficult to complete through websites or apps.

Examples cited for the feature include booking restaurants that do not accept online reservations, asking a dentist’s office to add a user to its cancellation list, and resolving problems with a cable bill. The assistant is intended to conduct the call rather than merely provide a phone number, draft a script, or tell the user how to navigate an automated menu.

Instinct Concierge is currently rolling out to some users through early access. Shinn said the company plans to expand availability over time, but the source does not provide a general release date or details about limits on call duration, supported businesses, or human handoff.

The calling launch follows other efforts by Instinct to give its assistant more control over users’ digital lives. TechCrunch reports that the company recently introduced email addresses that can be used for account sign-up and management. It has also enabled assistants to contact other users’ assistants for certain tasks when those users are added to a trusted network.

Meta’s Muse follows with U.S. business calls

Meta’s Muse has also gained the ability to place calls, although the initial rollout is narrower. Meta said the feature is being expanded to users who ask Muse to make a call, with the company prioritizing people who had already requested the capability.

That approach makes feature requests part of the beta rollout process. According to TechCrunch, Meta said new Muse features will initially be available to users who ask about them. The company’s post described outbound calling to U.S. businesses as a beta expansion and asked participating users to provide feedback.

The source does not specify which categories of businesses Muse can call, whether calls can be transferred to a user, or how the assistant identifies itself. Those details will be important for evaluating the practical difference between a demonstration and a dependable service.

For now, the parallel launches reduce one of the clearest product distinctions among text-based assistants. Competitors including Wajo, Town, and Ollie are also seeking users in the same market, and calling had been promoted as an advantage over Instinct. With Instinct and Muse both adding the capability, outbound phone work is becoming closer to a baseline expectation for this group of products.

What the evidence says about traction

The product announcements are supported by company statements and reporting from TechCrunch, rather than by independent testing of call quality or task-completion rates. Claims about what Instinct Concierge and Muse can accomplish should therefore be treated as product descriptions, not proof that every call will succeed.

The market context is also partly based on reported and vendor-linked signals. TechCrunch says Instinct attracted significant investor interest and raised $350 million at a reported $2.5 billion valuation last month. The publication also reported that the company was in talks to raise $1 billion at a potential $10 billion valuation. Those discussions and valuation figures are not presented as completed financing in the available evidence.

For Muse, analytics firm Sensor Tower reportedly found more than 730,000 U.S. downloads in the first few days after launch, compared with 707,000 downloads for the Meta AI app during its first five days. TechCrunch also reported that Muse reached the No. 2 position in top charts at one point. Download figures are an adoption signal, not evidence of sustained usage, successful task completion, or user satisfaction.

The comparison is further complicated by the timing of the products. TechCrunch describes Meta’s Muse as having launched the previous week, while Instinct’s broader product and financing activity has been developing over a longer period. Neither the source nor the companies provide comparable figures for active users, completed calls, failure rates, or retention.

Why phone calls raise the product stakes

For builders, outbound calling introduces a different reliability problem from web navigation. An assistant must determine whom to call, explain the user’s request accurately, handle unexpected questions, recognize when a business requires authorization, and avoid making commitments the user did not approve.

A restaurant reservation or cable-bill dispute can also involve personal information, payment details, cancellation terms, or sensitive account data. Product teams will need clear controls for permission, confirmation, call recording and transcripts, disclosure that an AI is speaking, and escalation to the user. The available reporting does not establish how Instinct or Muse handles each of these issues.

The feature could nevertheless make AI agents more useful for people who cannot easily spend time on hold or navigate businesses with poor online support. For enterprise AI buyers, the developments are a reminder that agent workflows may need to include voice and legacy phone systems, not just APIs and browser interfaces.

The competitive effect may be just as important. If calling becomes common across consumer assistants, differentiation will shift toward execution quality: whether an agent reaches the right department, follows instructions, avoids errors, and completes tasks without requiring repeated user intervention. The cost of phone infrastructure and human fallback support could also become a meaningful factor in the economics of these services.

What to watch next

The next signals will be broader access and independent evidence of performance. Instinct’s expansion beyond early access will show whether Concierge is a limited premium workflow or a core product capability. For Muse, the key question is whether the request-based rollout becomes available to the wider beta population and which U.S. businesses it can reliably reach.

Builders and buyers should watch for published success rates, cancellation and booking accuracy, disclosure policies, user approval controls, and support for calls that require authentication. Pricing will also matter: frequent calls could create material operating costs, particularly when assistants remain on hold or need multiple attempts.

It will also be worth tracking whether these products add user-supervised calling, transcripts, call summaries, and the ability to pause or take over a conversation. Those controls could determine whether phone agents are trusted for low-risk errands only or for more consequential customer-service and account-management tasks.

Creati.ai perspective

Instinct and Meta are not merely adding another interface. They are testing whether an AI assistant can act as a delegated representative in a messy environment that was built for humans. That is a meaningful step beyond generating text, but the value will depend less on the novelty of placing a call than on reliable authorization, transparency, and recovery when conversations go off script.

The immediate news is that outbound calls are moving from a competitive talking point toward a shared capability. The stronger product will be the one that can demonstrate safe, repeatable completion of real tasks—not simply the one that can dial a number.

Ads