Press kit

CallForMe app icon

CallForMe lets your AI agent call businesses for you. Add one URL to Claude, ChatGPT, Codex, Cursor or any agent that speaks MCP, and it can phone a US or Canadian business to book a table, get a quote, cancel a membership or wait on hold, then report back with a transcript and the answers.

In one line

CallForMe lets your AI agent call businesses for you.

Short description

CallForMe gives AI agents a phone. You add one URL to your agent and click Connect. When you ask for an errand that needs a phone call, a voice assistant places the call, works the phone menu, waits on hold, checks with your agent when the business asks something, and brings back the answer with a full transcript.

Longer description

Most errands still end in a phone call: a quote from a mechanic, a table for six, a gym that only cancels by phone. AI agents can find the number but can't dial it. CallForMe is the phone for those agents.

It is a remote MCP server at https://callforme.tel/mcp, plus a REST API and an agent skill for agents that only have a shell. The user's own agent stays in charge. It tells CallForMe who to call and what to get done. The voice assistant says who it is calling for in its first sentence, says it is an AI whenever asked, handles menus and hold music, and when the business asks for something the agent didn't provide, it asks the agent mid-call and passes the answer back into the live conversation. Several calls can run at once, so getting three quotes takes the same time as getting one.

Pricing is $1 per answered call, including the first 2 minutes, then 40 cents a minute. Calls nobody answers are free. There is no subscription.

Facts

Product CallForMe
Company Great Work LLC
Launched September 29, 2026
Website https://callforme.tel
MCP server https://callforme.tel/mcp
Works with Claude (web, desktop, Claude Code), ChatGPT, Codex, Cursor, Gemini CLI, VS Code, Windsurf, Zed and other MCP clients, plus any agent that can make HTTP requests
Pricing $1 per answered call, first 2 minutes included, then 40 cents a minute. Unanswered calls are free
Calls come from +1 (469) 966-0062
Where it calls Businesses in the US and Canada
Disclosure Every call opens "an assistant calling for (name), on a recorded line"; it says it is an AI whenever asked and never claims to be a person

Coverage

Independent checks

Places outside Great Work that reviewed or tested CallForMe, so you can check them yourself:

  • ChatGPT apps directory: reviewed by OpenAI and listed since October 20, 2026.
  • Official MCP Registry: tel.callforme/callforme, published under a namespace proven by a DNS record on callforme.tel.
  • AllMCPs: the directory ran the server itself and marks the listing "Verified working" (checked October 30, 2026).
  • Source for the agent skill: github.com/callforme-tel/callforme, MIT licensed.
  • Live status: callforme.tel/status, from the same server that places calls.

What we've learned so far

From customer calls in October 2026, the first month (our demo and test lines excluded). Small samples, so treat them as early signals:

  • The first sentence decides whether a business stays on the line. When calls opened with "this is an AI assistant", about a third of businesses hung up within 45 seconds. Opening with "an assistant calling for (name), on a recorded line", and saying it's an AI whenever asked, brought that to about one in eight.
  • Most calls are short. More than half of answered calls (29 of 53 through November 3, 2026) finished inside the included 2 minutes and cost exactly $1.
  • Businesses ask things the agent didn't know. A visit type, the last digits of a VIN, a callback number. The question goes back to the user's agent mid-call while the business waits, and that round trip is most of what makes a call succeed.
  • Identity checks are the wall. Airlines, banks and loyalty programs send a code to the account holder, so those calls work best when the assistant patches the user in once a person answers.
  • Big customer service lines are slow to reach, when they let you through at all. On November 2 and 3, 2026, our assistant called 34 large customer service lines and we timed each one from pickup to the first live person, reading every time off the call transcript. 25 put us through: American Express took 1 minute 2 seconds, USPS 23 minutes 4 seconds, median 3 minutes 38 seconds. The other 9 never connected us to anyone, and banks and card issuers were the hardest: 4 of 7 never put us through (two wanted account details first, one hung up in its queue, one went silent). Phone carriers were the fastest to answer (median 1 minute 24 seconds), shipping lines the slowest (15 minutes 38 seconds). One call per company, so it's a snapshot: the table, the numbers we dialed and the path that worked.
  • The transcript can be right while the audio is wrong. Our voice wrote phone numbers as numerals and the speech engine ran the digits together: transcribing our own recordings, "214-555-0142" came back as "2455-042". The assistant now says numbers digit by digit, and we check recordings, not just transcripts.

Try it before you write about it

The free demo lines are fictional businesses that answer the phone. No card, no signup. Add https://callforme.tel/mcp to your agent and say:

Call Maple Street Pizza at (469) 770-7412 and ask if they have a gluten-free crust and how long a large pepperoni takes for pickup.

All demo lines are listed at https://callforme.tel/demo. If you are reviewing CallForMe and want to call real businesses, email hello@greatwork.company and we'll put free credit on your email address.

Logos, images and video

You can use these in coverage of CallForMe. Please don't change the icon's colors or proportions.

File What it is Format
icon.svg App icon, vector SVG
icon-512.png App icon 512 x 512 PNG
icon-192.png App icon, small 192 x 192 PNG
og.png Social card 1200 x 675 PNG
demo-poster.png Still from a demo call 1280 x 720 PNG
reservation-wide.mp4 Video: an agent books a table by phone MP4, 2 MB
media/demo-brake-quote.mp3 Audio: a full call to a demo auto shop (brake quote; the shop asks if it's a robot) MP3, 0:57, 340 KB
datasets/hold-times.csv Data: the hold-times study, one row per company (number dialed and its source page, seconds to a person, menu type, reason if no person), CC BY 4.0 CSV
datasets/hold-times.json Data: the same hold-times study as JSON JSON
hold-times-chart.png Chart: time to reach a person at each customer service line we measured 1200 x 1200 PNG
brake-quotes-wide.mp4 Video: three shops called at once for brake quotes MP4, 2 MB

Useful links

Contact

hello@greatwork.company