Some of us would rather do almost anything than make a phone call. Rescheduling a dinner booking, chasing the dry cleaner, checking whether a shop has something in stock. They’re small tasks, but they somehow end up sitting on our to-do list for days.
AI Rudder, a Singapore-headquartered company that says it has handled more than 80 billion AI calls, wants to take those calls off your hands. It has launched CALL-E, a service that lets AI agents make real phone calls on your behalf.
Where it works
CALL-E plugs into the AI tools many people already use:
- ChatGPT
- Personal AI agents such as Muse, OpenClaw and Hermes Agent
- AI coding tools such as Claude Code, Codex and Cursor
You give your AI agent the errand, it hands the call to CALL-E, and CALL-E reports back once it’s done.
According to AI Rudder, personal AI agents are taking off fast. It says Meta’s Muse passed five million US downloads in its first three weeks, and that more users want their agents to handle phone calls for them.
Why phone calls are hard for AI
Chatting with AI by text is one thing. Talking on the phone is much messier.
AI Rudder CEO Teng Ren gives a simple example. Your AI calls a restaurant to move a booking. The host says “mm-hmm”, and the AI stops talking as if she’d asked a question. Then she says “Tuesday at... let me check”, and the AI jumps in with “Great, Tuesday it is”, locking in a time she never actually gave. At that point, she knows she’s talking to a bot and hangs up.
To handle this, AI Rudder built a model called Helm Swift that works out when someone is actually interrupting and when they’re just saying “yeah”. The company says it has the joint-lowest false-positive rate on the TurnBench Interruption leaderboard, so it rarely stops talking just because the other person makes a small acknowledging noise.
Four safeguards built in
To its credit, AI Rudder seems to have thought about what could go wrong:
1. Approval before every call
A phone call can’t be “unsent”. CALL-E only places a call after you approve the exact plan, so it never calls anyone without your go-ahead.
2. Proof after every call
After each call, CALL-E tells you what was agreed, how confident it is that the task was done, and the exact lines from the conversation that back it up. If it says “Booked for 7pm”, you can check it actually heard 7pm.
3. It says who is calling
By default, every call introduces itself as CALL-E, a voice agent. You can change what it says, but a pre-check stops it from pretending to be authorities like the FBI or 911.
4. Security and records
CALL-E runs on the same technology as AI Rudder’s enterprise products, which are built to meet SOC 2 Type 2 and ISO 27001, as well as GDPR and PDPA requirements. Every call is logged from plan to result, with your approval and the evidence attached, so a disputed call can be traced.
Regulators are heading the same way. In the US, the FCC ruled in 2024 that AI-generated voices count as artificial voices under phone call laws, so they need consent and caller identification. In Europe, the EU AI Act requires that people be told when they’re talking to an AI.
A word of caution: AI calls, spam and scams
This is useful technology, and CALL-E’s safeguards are a good start. But AI phone calls also arrive at a tricky time. Here in Malaysia, phone scams are already a daily headache, from Macau scam calls to fake bank and courier calls. AI voices that sound convincingly human could make that problem worse in the wrong hands.
So whether you’re using an AI agent to make calls, or receiving one, a few sensible habits go a long way.
If you’re using an AI agent to make calls:
- Stick to simple, low-risk errands. Booking changes, opening hours and stock checks are fine. Anything involving money, contracts or legal matters is better handled by you.
- Never let it share sensitive information. Keep your IC number, bank details, card numbers, passwords and OTPs out of any task you give an AI agent.
- Always let it introduce itself as AI. Don’t change the greeting to make it sound like you. Being upfront is fair to the person on the other end, and it builds trust.
- Only call people who expect it. Use it for businesses you deal with, not for cold calling, mass calling or marketing. That’s how spam starts.
- Read the report afterwards. The proof is there for a reason. Check that what was agreed is actually what you wanted.
If you receive a call from an AI agent:
- Don’t share personal or financial details with any caller you didn’t expect, AI or human.
- Never give out an OTP or TAC over the phone. No real bank, courier or government agency will ask for one.
- Verify before acting. If an AI says it’s calling on behalf of someone you know or a company you use, hang up and call them back on a number you already have.
- Be extra wary of urgency. “Pay now or your account will be frozen” is a classic scam line, whether a person or an AI says it.
Availability
CALL-E is available now at heycall-e.com, and new users get 100 free credits to try it out.
It’s a clever idea, and the built-in safeguards show AI Rudder is taking the risks seriously. Just remember that an AI making calls for you is still speaking in your name. Use it for the boring errands, keep your sensitive details to yourself, and stay sceptical whenever an unexpected caller starts asking for information.
Leave a comment