OpenHelm Mail · Outbound
Outreach that handles what comes back
Write to a list of suppliers, prospects or customers, each message built from that row rather than a find-and-replace. Then the replies wake an agent that reads them and moves each conversation on.
The situation.
Sending the first message is the easy half. The work is in what comes back: reading fifty answers, pulling the numbers out of them, chasing the people who said nothing, and knowing which ones need you personally.
How it works.
Personalised from the row, not templated
Each message is written from that recipient’s own data and whatever the agent can verify about them. If it cannot verify a detail, it leaves it out rather than guessing.
The replies drive the work
A reply wakes the agent that sent the original, on the same thread. It reads what came back, opens any attachment, records the outcome, and either answers, advances, or escalates.
One-click unsubscribe, built in
Every message that opens a conversation carries a working one-click unsubscribe, which is what Gmail and Yahoo require of bulk senders. An opt-out is permanent and cannot be reversed by the sender.
Approve the first few
Hold agent mail for approval while you calibrate the tone, then switch the inbox to send by itself. The queue shows you exactly what would have gone out.
Separate budgets, so nothing starves
Marketing, agent and transactional mail each have their own daily allowance per address. A campaign cannot consume the budget your password resets need.
Bounces and complaints handled
A hard bounce or a complaint suppresses that address automatically, and an address that keeps bouncing pauses the inbox rather than burning its reputation.
A new domain is warmed up, not opened at full volume
Start a warm-up on a fresh sending domain and its daily ceiling rises on a schedule instead of jumping to full volume, which is what gets a new domain throttled by the receiving side. Your transactional mail is exempt, because a password reset is not volume you chose to send.
What you actually do. Once.
Add your list
A table of recipients with whatever you know about each one. The agent works from the rows and will not invent a recipient that is not there.
Send from your own domain
Verify a domain so the mail comes from you, with its own DKIM and its own reputation, separate from everyone else on the platform.
Describe the objective
What you want from each conversation, and what should always come to a person. One prompt, not a sequence builder.
Let the replies run it
Each answer wakes the agent on that thread. Silence is chased on a cadence you set, and a clear no closes the row and suppresses the address.
Questions
Is this compliant bulk email?
Messages that open a conversation carry one-click unsubscribe headers and a working opt-out page, which is what Gmail and Yahoo’s bulk-sender rules require. If the platform cannot mint a working unsubscribe link, the send is refused rather than delivered without one.
Do replies really trigger work, or is something polling?
They trigger it. The reply is delivered to us and starts the run, carrying its thread. You do not connect a personal mailbox and there is no poll to tune.
Can it read a quote that came back as a spreadsheet?
Yes. It can list and download what is attached to a thread and work from the file.
What about open and click reporting?
Delivery, bounces and complaints are recorded, and opens and clicks are stored per message where the provider reports them. There is no campaign reporting surface over that yet, and we would rather say so than show you a dashboard we have not built. An open is also a pixel load: privacy-protecting mail clients fire it whether or not a person read anything, which is why we never present it as attention.
Can I send from a brand-new domain?
Yes, but warm it up. Starting a warm-up raises the daily ceiling on a schedule rather than letting the first campaign be the most damaging thing you do with the domain. An established domain you are migrating should not ramp, and is not made to.
What stops it emailing someone twice?
Every send can carry an idempotency key, so a retried request returns the original message instead of sending a second one.
Give your agents a real email address.
Inboxes on your own domain, replies that wake the agent handling them, and a person in the loop wherever you want one.