Timing Beats Volume: How We Built an AI Worker That Knows When to Reach Out
Generic outreach is dying. Here's the AI research worker we built to spot the moments your contacts actually want to hear from you — and draft the message.
Almost every business has the same outreach problem, and almost everyone solves it the same broken way: buy a list, write a generic message, send it to a few thousand strangers, and hope. It barely works anymore. Spam filters punish bulk mail, and the people on the other end have seen the template a hundred times. Response rates to cold, untargeted outreach are famously low — and they keep falling.
The thing is, the problem was never the message. It was the timing. A note that lands the same week someone changed jobs, opened a second location, or publicly asked for exactly what you sell doesn’t read like a pitch. It reads like good attention. The trouble is that those moments are scattered across the open internet, they happen on no schedule, and a small team can’t possibly watch for all of them by hand.
So we built an AI worker to do exactly that. We pointed it at our own market first, and it has been quietly doing the watching ever since. This is what it does, why it works, and why the same capability is one of the most useful AI projects most businesses could stand up.
The job: watch the world, surface the moment
Think of a sharp business-development assistant who reads everything — local news, industry announcements, public posts, security advisories — and whose entire job is to tap you on the shoulder and say, “You should reach out to this person today, and here’s why.” Not a thousand names. A short, ranked list of genuinely good moments, each one with a reason attached.
That’s the worker we built. It doesn’t blast anyone. It doesn’t replace the people who do the talking. It runs in the background, watches for the handful of real signals worth acting on, and hands a human a ready-to-go draft for each one. The person spends a focused half-hour a day reviewing what the AI found, editing the language to sound like themselves, and clicking send — or deciding the moment isn’t right and moving on.
The promise is simple: fewer messages, far better timing, and every one of them grounded in something specific the recipient actually cares about right now.
How it works, in four moves
Under the hood, the worker runs the same loop a great researcher would, just continuously and at machine speed.
- Watch. It pulls from a range of public sources on a steady cadence — news and announcements, leadership changes, expansion and hiring activity, security advisories, and public posts where someone openly asks for a recommendation. Everything funnels through a single intake so new sources can be added without rebuilding the machine.
- Understand. A language model reads each raw item and pulls out what matters: which company, which person, what happened, and how time-sensitive it is. A press release becomes structured facts instead of a wall of text.
- Match. It connects each event back to the people and organizations you already know — the contacts in your CRM. A signal that doesn’t tie to anyone you care about is set aside. A signal that lands on a real relationship gets promoted.
- Prepare. For the moments that clear the bar, it drafts the outreach: a short subject line and a few sentences that open by referencing the specific thing that happened, then make a natural, low-pressure ask. The draft arrives with its reasoning and a link to the source, so the person reviewing it can see exactly what triggered it.
By the time a human looks, the research is done, the message is written, and the only thing left is the part that should always belong to a person: judgment.
Every claim shows its work
The fastest way to lose trust in an AI is to let it assert things it can’t back up. So we built the worker around a rule: every recommendation cites its source. If the draft says a contact just stepped into a new leadership role, the person reviewing it can click straight through to the announcement that says so. If it says someone publicly asked for a provider like yours, the original post is one click away.
This matters for two reasons. First, it keeps the human in genuine control — you’re approving a real, verifiable moment, not trusting a black box. Second, it keeps the outreach honest. Nobody sends a message built on a hallucinated fact, because the fact is always sitting right there next to the draft, ready to be checked.
The hard part isn’t finding signals — it’s restraint
Here’s the lesson that separated this tool from a spam cannon: knowing when not to reach out is as important as knowing when to.
Take an obvious example. When a company suffers a security incident, it shows up loudly in public sources, and the naive move is to pounce — “looks like you could use a new IT provider.” That’s tone-deaf, and people remember it. So we taught the worker the opposite instinct. It flags those companies, but instead of pushing a draft to the top, it sets them aside with a reminder to revisit weeks later, once the dust has settled and they’re genuinely evaluating options. The same patience applies to plenty of other signals that look urgent but aren’t.
We also made the negative decisions count. When a person reviews a suggestion and decides it’s noise, they say why in a word or two, and that feedback flows back into the system. Over time it learns what a given team considers a waste of time and stops surfacing it. The worker gets quieter and sharper the more it’s used — which is exactly backwards from how most automated outreach goes.
This kind of judgment — patience, context, learning from a “no” — is what makes an AI worker feel like a teammate instead of a tool. It’s also why we think of these as AI virtual employees rather than software features: they own an outcome, they exercise restraint, and they answer to a person.
We built it for ourselves first
We didn’t read about AI-driven intelligence in a vendor deck and turn around to resell it. We built this worker to solve our own outreach problem, pointed it at our own market, and let it run as part of how we actually work. That’s the bar we hold ourselves to as a Managed Intelligence Provider: every AI capability we offer, we operate in our own business before we ever propose it for yours.
That matters because the gap between a slick demo and a system that survives daily use is enormous. The demo never has to deal with a source that goes quiet, a name that matches two different companies, or a signal that looked urgent and turned out to be nothing. Ours does, every day, because it’s a worker on the job — not a science project.
The human stays in charge, and the safety is built in
A fair worry: “If an AI is deciding who to contact, am I handing over the relationship?” No. The worker handles the grinding middle — the watching, reading, matching, and first-draft writing. It never sends anything on its own. Nothing leaves the building without a person reading it, adjusting the wording, and choosing to send. The AI clears the runway; your team still flies the plane.
We also built the brakes in from the start. The system runs in a safe mode by default, where messages route to a test inbox instead of real contacts until someone deliberately turns live sending on — and turning it on takes a conscious, confirmed action, not an accidental click. Every send, reply, and decision is logged, so there’s always a clear record of who reached out to whom and why. For outward-facing automation, that kind of reversibility and audit trail isn’t a nice-to-have; it’s the whole reason you can trust the thing to run.
Why this can — and probably should — run privately
There’s one more reason this capability belongs to your business rather than a generic cloud service: the raw material is your contact list and your relationships. That’s some of the most sensitive data a company owns, and the last thing you want is to pump your entire CRM through a third-party AI service to get it sorted.
It doesn’t have to work that way. The same worker can run on Local AI — a private AI server your business owns, sitting on your own hardware — so the reasoning happens inside your network. Your contacts, your pipeline, and the messages your team is about to send never leave the building. You get the intelligence without renting out your relationships, and without a per-message bill that grows every time the system gets more useful.
That’s the pattern behind everything we build: AI layered on top of dependable managed IT, not in place of your people. The core — your network, your security, your backups — stays rock-solid. The AI worker just removes the part of business development that no small team can do well by hand: being in the right place at the right moment, every time.
Could we build this for your business?
Almost certainly, and the shape of it would look familiar even if the signals were different. A contractor cares about new permits and projects breaking ground. A professional firm cares about leadership changes and expansions in the industries it serves. A supplier cares about who just won a contract that means new demand. The principle is identical: somewhere in the public record, the moments when your next customer is most ready to hear from you are already happening. The only question is whether anyone is watching for them.
We built our intelligence worker to do that watching, on infrastructure we own, with a human firmly in the loop. The same thing can run inside your business — turning the scattered noise of the open internet into a short daily list of well-timed, well-grounded conversations worth having.
Want to know what your first AI worker would be? Start with a business IT assessment and we’ll look at how your team finds and reaches new business today, where the timing is slipping through the cracks, and whether a private AI research worker is the right first step.
Timing Beats Volume: How We Built an AI Worker That Knows When to Reach Out
Generic outreach is dying. Here's the AI research worker we built to spot the moments your contacts actually want to hear from you — and draft the message.
Almost every business has the same outreach problem, and almost everyone solves it the same broken way: buy a list, write a generic message, send it to a few thousand strangers, and hope. It barely works anymore. Spam filters punish bulk mail, and the people on the other end have seen the template a hundred times. Response rates to cold, untargeted outreach are famously low — and they keep falling.
The thing is, the problem was never the message. It was the timing. A note that lands the same week someone changed jobs, opened a second location, or publicly asked for exactly what you sell doesn’t read like a pitch. It reads like good attention. The trouble is that those moments are scattered across the open internet, they happen on no schedule, and a small team can’t possibly watch for all of them by hand.
So we built an AI worker to do exactly that. We pointed it at our own market first, and it has been quietly doing the watching ever since. This is what it does, why it works, and why the same capability is one of the most useful AI projects most businesses could stand up.
The job: watch the world, surface the moment
Think of a sharp business-development assistant who reads everything — local news, industry announcements, public posts, security advisories — and whose entire job is to tap you on the shoulder and say, “You should reach out to this person today, and here’s why.” Not a thousand names. A short, ranked list of genuinely good moments, each one with a reason attached.
That’s the worker we built. It doesn’t blast anyone. It doesn’t replace the people who do the talking. It runs in the background, watches for the handful of real signals worth acting on, and hands a human a ready-to-go draft for each one. The person spends a focused half-hour a day reviewing what the AI found, editing the language to sound like themselves, and clicking send — or deciding the moment isn’t right and moving on.
The promise is simple: fewer messages, far better timing, and every one of them grounded in something specific the recipient actually cares about right now.
How it works, in four moves
Under the hood, the worker runs the same loop a great researcher would, just continuously and at machine speed.
- Watch. It pulls from a range of public sources on a steady cadence — news and announcements, leadership changes, expansion and hiring activity, security advisories, and public posts where someone openly asks for a recommendation. Everything funnels through a single intake so new sources can be added without rebuilding the machine.
- Understand. A language model reads each raw item and pulls out what matters: which company, which person, what happened, and how time-sensitive it is. A press release becomes structured facts instead of a wall of text.
- Match. It connects each event back to the people and organizations you already know — the contacts in your CRM. A signal that doesn’t tie to anyone you care about is set aside. A signal that lands on a real relationship gets promoted.
- Prepare. For the moments that clear the bar, it drafts the outreach: a short subject line and a few sentences that open by referencing the specific thing that happened, then make a natural, low-pressure ask. The draft arrives with its reasoning and a link to the source, so the person reviewing it can see exactly what triggered it.
By the time a human looks, the research is done, the message is written, and the only thing left is the part that should always belong to a person: judgment.
Every claim shows its work
The fastest way to lose trust in an AI is to let it assert things it can’t back up. So we built the worker around a rule: every recommendation cites its source. If the draft says a contact just stepped into a new leadership role, the person reviewing it can click straight through to the announcement that says so. If it says someone publicly asked for a provider like yours, the original post is one click away.
This matters for two reasons. First, it keeps the human in genuine control — you’re approving a real, verifiable moment, not trusting a black box. Second, it keeps the outreach honest. Nobody sends a message built on a hallucinated fact, because the fact is always sitting right there next to the draft, ready to be checked.
The hard part isn’t finding signals — it’s restraint
Here’s the lesson that separated this tool from a spam cannon: knowing when not to reach out is as important as knowing when to.
Take an obvious example. When a company suffers a security incident, it shows up loudly in public sources, and the naive move is to pounce — “looks like you could use a new IT provider.” That’s tone-deaf, and people remember it. So we taught the worker the opposite instinct. It flags those companies, but instead of pushing a draft to the top, it sets them aside with a reminder to revisit weeks later, once the dust has settled and they’re genuinely evaluating options. The same patience applies to plenty of other signals that look urgent but aren’t.
We also made the negative decisions count. When a person reviews a suggestion and decides it’s noise, they say why in a word or two, and that feedback flows back into the system. Over time it learns what a given team considers a waste of time and stops surfacing it. The worker gets quieter and sharper the more it’s used — which is exactly backwards from how most automated outreach goes.
This kind of judgment — patience, context, learning from a “no” — is what makes an AI worker feel like a teammate instead of a tool. It’s also why we think of these as AI virtual employees rather than software features: they own an outcome, they exercise restraint, and they answer to a person.
We built it for ourselves first
We didn’t read about AI-driven intelligence in a vendor deck and turn around to resell it. We built this worker to solve our own outreach problem, pointed it at our own market, and let it run as part of how we actually work. That’s the bar we hold ourselves to as a Managed Intelligence Provider: every AI capability we offer, we operate in our own business before we ever propose it for yours.
That matters because the gap between a slick demo and a system that survives daily use is enormous. The demo never has to deal with a source that goes quiet, a name that matches two different companies, or a signal that looked urgent and turned out to be nothing. Ours does, every day, because it’s a worker on the job — not a science project.
The human stays in charge, and the safety is built in
A fair worry: “If an AI is deciding who to contact, am I handing over the relationship?” No. The worker handles the grinding middle — the watching, reading, matching, and first-draft writing. It never sends anything on its own. Nothing leaves the building without a person reading it, adjusting the wording, and choosing to send. The AI clears the runway; your team still flies the plane.
We also built the brakes in from the start. The system runs in a safe mode by default, where messages route to a test inbox instead of real contacts until someone deliberately turns live sending on — and turning it on takes a conscious, confirmed action, not an accidental click. Every send, reply, and decision is logged, so there’s always a clear record of who reached out to whom and why. For outward-facing automation, that kind of reversibility and audit trail isn’t a nice-to-have; it’s the whole reason you can trust the thing to run.
Why this can — and probably should — run privately
There’s one more reason this capability belongs to your business rather than a generic cloud service: the raw material is your contact list and your relationships. That’s some of the most sensitive data a company owns, and the last thing you want is to pump your entire CRM through a third-party AI service to get it sorted.
It doesn’t have to work that way. The same worker can run on Local AI — a private AI server your business owns, sitting on your own hardware — so the reasoning happens inside your network. Your contacts, your pipeline, and the messages your team is about to send never leave the building. You get the intelligence without renting out your relationships, and without a per-message bill that grows every time the system gets more useful.
That’s the pattern behind everything we build: AI layered on top of dependable managed IT, not in place of your people. The core — your network, your security, your backups — stays rock-solid. The AI worker just removes the part of business development that no small team can do well by hand: being in the right place at the right moment, every time.
Could we build this for your business?
Almost certainly, and the shape of it would look familiar even if the signals were different. A contractor cares about new permits and projects breaking ground. A professional firm cares about leadership changes and expansions in the industries it serves. A supplier cares about who just won a contract that means new demand. The principle is identical: somewhere in the public record, the moments when your next customer is most ready to hear from you are already happening. The only question is whether anyone is watching for them.
We built our intelligence worker to do that watching, on infrastructure we own, with a human firmly in the loop. The same thing can run inside your business — turning the scattered noise of the open internet into a short daily list of well-timed, well-grounded conversations worth having.
Want to know what your first AI worker would be? Start with a business IT assessment and we’ll look at how your team finds and reaches new business today, where the timing is slipping through the cracks, and whether a private AI research worker is the right first step.