Verified · Sep 18, 2026
Independently verifiedInstinct and Meta's Muse add phone calling the same day — a Concierge early access here, a request-gated beta there
3 sourcesTechCrunch reported on Thursday (Ivan Mehta, 6:46 AM PDT): 'Text-based AI assistants are a hot market, with Instinct, Wajo, Town, Ollie, and Meta's Muse agent, which launched last week, vying for users' attention and tasks.' Instinct 'added a new calling feature that lets your agent place calls and act as a concierge.' Its founder Noah Shinn announced it on X (post dated September 16, 2026, Wednesday, in the embed): 'Your Instinct can now handle phone calls. Introducing Instinct Concierge – a white glove service meant to handle high-touch cases, such as making phone calls, high-end service booking, and more.' The concrete use cases — booking 'restaurants that don't take online reservations', getting 'on your dentist's cancellation list', sorting out 'problems with your cable bill' — are the report's paraphrase sentence. The pace: 'rolling out in early access to some users, with plans to expand to all users in time, he said.' On the Muse side: 'It also just gained the ability to place calls to U.S. businesses' and will 'first roll out to users who ask Muse to make a call' — the report explains the mechanic: 'This essentially serves as a feature request mechanism', with new features going 'first to people who ask Muse about them.' Ryan Fox's post (September 16): 'we just expanded the @Muse beta for outbound calls to US businesses, prioritizing folks who'd asked their Muse to let us know they wanted it first :)'. The outlet's competitive read: rivals had been touting calling as an advantage over Instinct, but 'with Instinct's Concierge feature rollout, they're all now in parity with each other on this front.' Background: Instinct last week rolled out email addresses for account sign-up and management, and now allows assistants to contact other people's assistants through 'a trusted network'; it 'raised $350 million at a $2.5 billion valuation last month' and is 'reportedly in talks to raise $1 billion at a $10 billion valuation.' Per Sensor Tower, Muse 'garnered over 730,000 downloads in the U.S.' against the Meta AI app's '707,000 times post-launch in the first five days', and 'even reached the No. 2 position in the top charts at one point.' Instinct's own site (opened this run) says the product is 'trained to use a phone and a computer. You can text or call it.' and is 'currently available to a private access group as we're scaling up compute', with waitlist or member-invite access. Muse's identity comes from Meta's own September 8 launch post: 'Muse is a personal AI agent. It doesn't just answer questions, it actually does the work.' — a post that contains no calling announcement at all.
Why now
Both announcements landed on Wednesday, September 16, and TechCrunch's joint read followed Thursday morning at 6:46 AM PDT — this is Friday-morning material. Three timing layers: first, 'AI agents make phone calls' stopped being a single-company gimmick and became a same-day, two-company race that creators can now structure as a comparison; second, Meta's gating mechanic is a story of its own — new features go 'first to people who ask Muse about them', turning feature-requesting into an actual acquisition strategy worth explaining to audiences now; third, the Sensor Tower numbers (Muse over 730,000 US downloads, outgrowing the Meta AI app's first five days) give the category a citable heat anchor. And because both products remain gated — early access here, request-gated there — the scarce explanation right now is 'who's in and how you get in', not 'what it can do'. That's the early-window slot.
Why it is worth publishing
Two companies, one capability, same day — the same-topic-race structure is a natural explainer frame, and both sides have quotable primary posts (as embedded by TechCrunch). The precision play is three-layer attribution: the calling ability and the use-case list come from TechCrunch's report and paraphrase; the two X posts are quotable only up to their embedded ellipses; and Muse's identity and pricing come from Meta's September 8 launch post, which contains no calling at all — pinning 'makes calls' onto that post is the single easiest mistake on this story. Every numeric qualifier survives: 'reportedly' welds to the $10B-valuation talks, and Sensor Tower's 'over', 'first five days', and 'at one point' each stay welded to their numbers. Instinct's own 'private access group' plus waitlist line is the honest answer to the comment-section question 'can I use it right now?'
Evidence basis
Three sources, all opened this run: TechCrunch (Ivan Mehta, Sept 17, 2026, 6:46 AM PDT; read in full, raw HTML fetched, every quoted span grepped verbatim), instinct.com (About and availability sections grepped verbatim; the page carries no dates and no Concierge announcement), and Meta's newsroom Muse launch post (page-stated September 8, 2026; read in full; grep-verified to contain no 'phone', 'call', or 'waitlist' anywhere). The two X posts (Noah Shinn and Ryan Fox, both dated September 16, 2026, Wednesday, in the embeds) are unopened origins, quoted only as embedded in TechCrunch's raw HTML, and both truncate mid-sentence ('get on…', 'helped make…'). Weekday-date pairs calendar-verified: Wednesday = September 16, 2026; Thursday = September 17, 2026; today = Friday, September 18, 2026. Numbers come in exactly two clusters: funding ($350M at $2.5B stated; $1B at $10B 'reportedly') and Sensor Tower (over 730,000 / 707,000 in the first five days / No. 2 at one point).
“AI agents are actually making phone calls now — Instinct shipped Concierge and Meta's Muse expanded its outbound-call beta the same day, and both are still gated.”
Angle
Make it a two-line comparison: 'whose agent can actually call.' Beat one: Instinct — Concierge's white-glove positioning in Shinn's own post words, the three use cases labeled 'per TechCrunch's paraphrase', the pace 'early access to some users.' Beat two: Muse — the request-gated rollout explained as a mechanic ('first to people who ask Muse about them'), the angle most takes will miss. Beat three: the base facts — Instinct's own site says 'private access group' plus waitlist, and Meta's September 8 post defines what Muse is (that post has no calling in it). Beat four: the numbers — funding and downloads with every qualifier attached. Close on TechCrunch's parity judgment as the open question: is the calling race about capability or distribution?
Format
Short talking-head video
Demo idea
A 'who calls first' gating flowchart: left lane Instinct — Shinn announces (September 16) → early access to some users → waitlist on the company site; right lane Muse — Fox announces (September 16) → 'first to people who ask' → Meta's feature-request mechanism. Each node quotes one verbatim line (up to the ellipsis), with a Sensor Tower download strip at the bottom carrying its three qualifiers.
Platform notes
Say the gates in the same sentence as the capability: Concierge 'early access to some users', Muse 'first to people who ask Muse to make a call', and Instinct itself behind 'a private access group' plus waitlist or invite (the site's own words). Quote the X posts only up to their embedded ellipses — 'get on…' and 'helped make…' end the record; introduce the three Concierge use cases as 'per TechCrunch's paraphrase.' 'Reportedly' stays welded to the $10B-valuation talks; Sensor Tower's 'over', 'first five days', and 'at one point' stay welded to their numbers, each with 'per Sensor Tower.' What Muse is comes from Meta's September 8 launch post; what Muse newly does comes from this week's report — never the reverse; the launch post contains no calling announcement. Label the parity line as TechCrunch's characterization.
Usable claims
- Two rival AI assistants added phone calling the same week, per TechCrunch's Thursday report (Ivan Mehta, 6:46 AM PDT, September 17, 2026). The market frame, the report's words: 'Text-based AI assistants are a hot market, with Instinct, Wajo, Town, Ollie, and Meta's Muse agent, which launched last week , vying for users' attention and tasks.' Instinct's move: 'it's added a new calling feature that lets your agent place calls and act as a concierge.' Per the report, 'Instinct's founder Noah Shinn announced on X that the buzzy startup's new calling feature, Instinct Concierge, can do things like book restaurants that don't take online reservations, get on your dentist's cancellation list, or sort out problems with your cable bill.' — that use-case list is the article's paraphrase of the post. Rollout: 'The feature is rolling out in early access to some users, with plans to expand to all users in time, he said.' Shinn's X post, as embedded in the report (September 16, 2026): 'Your Instinct can now handle phone calls. Introducing Instinct Concierge – a white glove service meant to handle high-touch cases, such as making phone calls, high-end service booking, and more.' — the embed truncates the next sentence at 'get on…', so no further post text exists in this run's record. Muse's move: 'Though Instinct is shipping fast, Meta's Muse AI assistant is keeping up. It also just gained the ability to place calls to U.S. businesses.' The gating mechanics, the report's words: 'The tech giant said the feature will first roll out to users who ask Muse to make a call. (This essentially serves as a feature request mechanism, and Meta said that whenever Muse launches new features, they'll be available first to people who ask Muse about them.)' Fox's X post, as embedded (September 16, 2026): 'we just expanded the @Muse beta for outbound calls to US businesses, prioritizing folks who'd asked their Muse to let us know they wanted it first :)' […] 'phone calling was one of our top requests and your feedback helped make…' — the two spans are separated by one elided sentence of the post, and the embed's own ellipsis marks where the post text ends. The outlet's competitive read, its characterization: 'These calling features arrive as a number of rivals had been touting their ability to make calls as an advantage over Instinct; but with Instinct's Concierge feature rollout, they're all now in parity with each other on this front.' Instinct's site (opened this run) says the product is 'trained to use a phone and a computer. You can text or call it.' and is 'currently available to a private access group as we're scaling up compute' with waitlist or member-invite access. Muse's identity comes from Meta's own September 8 launch post: 'Muse is a personal AI agent. It doesn't just answer questions, it actually does the work.', rolling out 'in the US on iOS, Android, and muse.ai', and 'free for most of what people need, with subscription plans for people who want to do more.' Funding and traction, attributed: 'Instinct raised $350 million at a $2.5 billion valuation last month', and it is 'reportedly in talks to raise $1 billion at a $10 billion valuation'; per 'Analytics firm Sensor Tower', Muse 'garnered over 730,000 downloads in the U.S.' versus the Meta AI app's '707,000 times post-launch in the first five days', and 'even reached the No. 2 position in the top charts at one point.'
Evidence pipeline
From the news
Breakdown
The biggest structural risk in this story is compressing three evidence layers — and a numbers lane of their own — into one sentence. Layer one: TechCrunch's report is the only fully opened narrative carrier — the capability lines, the gating lines, and the use-case list (restaurant, dentist, cable bill — a paraphrase) all live here, and 'parity' is its characterization, quotable only with the outlet's name. Layer two: the two X posts exist only as embeds and both truncate mid-sentence — the ellipses at 'get on…' and 'helped make…' are the physical boundary of the text; quoting past one fabricates post content. Both embeds are dated September 16, 2026 (Wednesday), the same day. Layer three: the two company surfaces each own one lane — instinct.com describes the product ('You can text or call it') and the gate ('a private access group', waitlist or invite), which is the honest answer to 'can I use it now?'; Meta's September 8 launch post defines what Muse is ('a personal AI agent') and its pricing, and is grep-verified to contain no phone, call, or waitlist anywhere — pinning the calling ability onto that post is the easiest mistake on the story. The numbers run their own layer: $350M at $2.5B is stated; $1B at $10B is 'reportedly in talks'; Sensor Tower's 'over 730,000', 'first five days', and 'at one point' each stay welded to their figure. Editor's rule: gates ship in the same sentence as capabilities, the ellipsis is the boundary, every layer names its own source, and no qualifier travels alone.
Sources
Risks
- Say the gates in the same breath as the capability: Concierge 'in early access to some users', Muse 'first to people who ask Muse to make a call', and Instinct itself on a waitlist or member invite. Quote the X posts only up to their embedded ellipses, and introduce the three Concierge use-cases as 'per TechCrunch'. Keep 'reportedly' attached to the $10B talks, and the three Sensor Tower qualifiers attached to their numbers. When you explain what Muse is, that's Meta's September 8 post — when you explain what Muse newly does, that's this week's report, never the launch post. And give TechCrunch its credit for the parity line.
Demo ideas
- The 'who calls first' two-lane flowchart: Instinct (Concierge early access → site waitlist) against Muse (first to those who asked → feature-request mechanism), nodes quoting verbatim lines up to the ellipses
- Numbers attribution card: left column 'stated' ($350M at $2.5B), right column 'reportedly' ($1B at $10B talks), plus the Sensor Tower strip with its three qualifiers — making the qualifiers themselves the content