A turnkey, production-ready agent layer between your servers and your customers' thumbs.
What it is
/ˈreɡjələr/ noun
a customer who comes back — someone the staff knows by name and never has to re-explain themselves.
Regulars AI — the agent layer between your servers and your customers' phones. We give you a phone number they text from iMessage, RCS, SMS, or WhatsApp.
an AI agent talks to your existing API to answer, remembers every conversation, and replies in your brand's voice — fully white-label.
iMessage, RCS, SMS, WhatsApp
You pick an area code, we provision the number. Messages flow over iMessage, RCS, SMS, or WhatsApp depending on the customer's device and your plan. We handle carrier registration, compliance, and deliverability so your number stays live and your messages land. You never have to touch carrier infrastructure.
The phone number is the auth
The messaging channel is the auth. A customer's phone number is one of the most stable identifiers they have. Every request to your API includes it alongside the channel, user ID, and a signature your server can verify. Signed per RFC 9421 with HMAC-SHA256.
No AI stack for your team to maintain
Incoming messages are queued, deduplicated, and processed in order. Conversation state is retrieved, context is hydrated and cached, guardrails enforced, intent parsed, tools called, and a response delivered. The full orchestration happens in seconds with no queue for your team to monitor.
Every conversation picks up where it left off
Message history and conversational context are stored and managed on our side. Customers text the same number, land in the same thread, and talk to an agent that remembers.
Repeat queries resolve without hitting your servers
Responses from your API are cached and intelligently reused across conversations. Customers get lower latency replies, your infrastructure gets fewer requests.
Filter what goes in and what comes out
Guardrails run on both user inputs and agent responses, catching abuse, injection attempts, PII, and off-topic requests before they reach your API or the customer. Pattern matching, keyword filters, AI-based evaluation, and PII redaction are all configurable per agent. Input guardrails run in parallel with context hydration so they don't add latency.
Not every conversation starts with the customer
Events let your API initiate a message when something happens on your end: an item restocks, an appointment is tomorrow, an order ships. Your API sends the signal, the agent crafts and delivers the message.
When the message needs to be exact, your API sends it directly
Your API can bypass the agent entirely. Throw a structured error with a message and we deliver it straight to the customer. No agent interpretation, no rewriting. Onboarding flows, payment links, account setup.
You pay per message, not per token
No estimating context windows, no surprise bills when conversations run long. One message in, one message out, one price. Usage scales linearly and predictably.
End-to-end conversations across any flow your business runs. These are just a few examples.
Easier than onboarding a human employee. No infrastructure to stand up, no models to host, no SDK to learn. If your API can serve a mobile app, it can serve Regulars.
We provision a real phone number in the area code you pick. Customers text it from iMessage, RCS, SMS, or WhatsApp.
Point the agent at HTTP endpoints that return live state. We fetch, cache, and template them into every response.
Tell the agent how to act: tone, scope, escalation rules, edge cases. Drop in example flows. Reference resources inline for live context.
Expose actions as HTTP endpoints or an MCP server. Bookings, refunds, KB lookups, status changes. Anything your API can do, the agent can do on request.
Filter user inputs and agent outputs for abuse, injection, PII, and off-topic requests. Block or redact before anything reaches its destination.
If prompting hits its limit, train on your transcripts. Pin the voice. Make the edge cases stick.
Every request signed per RFC 9421 with HMAC-SHA256 — an open IETF standard with libraries in every major language.
For less than the cost of a single employee's shift, every customer gets the regular treatment, over the messaging channels they prefer.
For developers and testing
100 messages capped
Pattern + keyword guardrails
Low request priority
WhatsApp at global scale
5,000 messages
3 inference-based guardrails
Native messaging channels
5,000 messages
iMessage + RCS with SMS fallback
3 inference-based guardrails
We handle everything
As many messages as you need contact us for volume pricing
All channels iMessage, RCS, SMS, WhatsApp
Unlimited guardrails tuned to your policies
Managed onboarding we set it up with you
Implementation & training consulting