Product 02 · The LinkedIn operating system
Shuttle
Scale the accounts.
Not the headcount.
Read the constraint before you buy your way around it.
The reframe
Everyone was buying
£40,000 accounts.
The platform caps outbound connection requests at roughly two hundred a week — per account. Not per person. Not per company.
Everyone knows there is a limit. Almost nobody reads it closely enough to notice where it lands. So the entire industry scales outbound the same way: hire another person, get another account, get another thirty or forty approaches a day, and accept another private pile of untracked conversations along with it.
But thirty or forty approaches a day is nowhere near a full day's work. What that salary was really purchasing was an account to attach the effort to.
If the constraint is accounts, the answer isn't headcount. It's identities: nine real, properly warmed accounts, one per market, all operated from one screen. A multi-year hiring plan became a configuration.
Screen 01
Nine identities.
One person at the screen.
The business argument, made visible in one frame.
| Identity | Market | Account | Weekly cap | Sent this week | Status |
|---|---|---|---|---|---|
AKAyaan | United Kingdom | 187528 | 200 | 184 | Active |
NRNidhi | Ireland | 190707 | 200 | 176 | Active |
NSNishika | Europe | 189987 | 200 | 191 | Active |
PYPiyush | Greece | 206852 | 200 | 158 | Active |
TKToshak | APAC | 192377 | 200 | 200 | Active |
SLSunil | Thailand | 208298 | 200 | 143 | Active |
MDMudit | Austria & Canada | 197754 | 200 | 167 | Active |
RHRahul | Sweden & Denmark | 204617 | 200 | 172 | Active |
AVAditi | New Zealand | 210813 | 200 | 149 | Active |
- What this is
- One market per identity: UK, Ireland, Europe, Greece, APAC, Thailand, Austria & Canada, Sweden & Denmark, New Zealand.
- Why it exists
- Nine markets used to mean nine silos, or nine hires. Assigning a real, warmed identity to a market keeps outreach local and credible — the person a Stockholm studio hears from is the person who covers Stockholm.
- Business outcome
- Nine markets run as one operation, and opening the tenth is a configuration rather than a recruitment round.
- What this is
- A weekly cap of 200, shown per account, with the week's real usage beside it.
- Why it exists
- This is the number the entire product is built around. It is a platform limit and it applies to the account, so the tool respects it per identity rather than trying to outrun it. Nothing here attempts volumes the platform cannot actually deliver.
- Business outcome
- Nine warmed accounts, none of them pushed into a restriction. The asset that takes years to build is protected rather than burned for one quarter of throughput.
- What this is
- 1,800 requests a week of capacity, and 18,400 approaches made so far — nine accounts multiplied by the platform's own ceiling.
- Why it exists
- One person working manually was capped near 200 a week, because they had one account. Reaching 1,800 the conventional way meant roughly ten hires and ten salaries.
- Business outcome
- The next increment of outreach volume stopped costing a headcount. Growth became a configuration change.
- What this is
- Connection status for every identity, in one badge.
- Why it exists
- A disconnected sending account fails quietly: sequences simply stop, and nobody finds out for days. The state of every identity is on the screen rather than in a log.
- Business outcome
- A dead account is noticed the same morning, not the following month when somebody asks why one market went silent.
But volume alone
would have made it worse.
Multiply outreach by nine and change nothing else and you don't get nine times the pipeline. You get nine times the replies, arriving into a system that was already failing to keep up with one person's worth.
A dropped conversation with a senior designer isn't a lost email. It's a damaged relationship in a small industry where everybody talks — and at nine times the volume it stops being an accident and becomes a policy.
So the volume and the memory had to be solved together, or not at all. That is why this is an inbox, not a sender.
Screen 02
Every reply, in one place.
Not a notification feed. Nine identities' worth of conversations, separated from the noise and sorted by intent.
- What this is
- 4,180 conversations, pre-sorted by what the prospect meant rather than when they wrote.
- Why it exists
- A social platform's own feed mixes a live prospect with a recruiter, a work anniversary and a newsletter. There is no filter for “this is somebody responding to outbound”, and at nine identities' worth of volume that feed becomes unusable.
- Business outcome
- The morning starts with 486 warm leads in one queue, not with a scroll.
- What this is
- A “Not interested” tab, given the same weight as the good news — 212 of them.
- Why it exists
- A clear no is valuable information. It stops the follow-up queue, protects the relationship, and prevents the same studio being approached again next quarter by a colleague who never knew.
- Business outcome
- Rejections are recorded rather than forgotten, which is what makes a shared prospect database safe to operate across nine markets at once.
- What this is
- Journey — one prospect's whole history in date order: messages, replies, campaigns entered, status changes.
- Why it exists
- Across nine identities and dozens of campaigns, “what have we already done with this person?” becomes genuinely hard — and getting it wrong means approaching a senior designer twice, differently, from two different names.
- Business outcome
- Nine times the volume stayed survivable, because the system remembers so the person doesn't have to.
- What this is
- An AI draft button that drafts, and never sends.
- Why it exists
- The blank page is the biggest reason a warm reply sits unanswered for a day, and at this volume a day becomes a week. But nothing is ever sent by anything other than a person deciding to send it.
- Business outcome
- A six-hour average response time held while volume multiplied — with a hard line kept: no autonomous agent speaks to a senior designer on the brand's behalf.
Screen 03
“Come back in March”
used to be a feeling.
Now it's a date, queued per sending identity, and re-checked every time the screen is opened.
| Person | Market | Sending as | Last contact | Due | |
|---|---|---|---|---|---|
ELElke Lindqvist |
Nordiska Rum | Rahul | 02 Apr 2026 | 39 days overdue | Follow up |
TATheodora Alexiou |
Alexiou Architects | Piyush | 18 Apr 2026 | 23 days overdue | Follow up |
JWJames Whitcombe |
Whitcombe Design | Aditi | 30 Apr 2026 | 11 days overdue | Follow up |
DMDiarmuid Moran |
Moran Interiors | Nidhi | 06 May 2026 | 5 days overdue | Follow up |
HCHarriet Colefax |
Colefax & Vane | Ayaan | 09 May 2026 | Due today | Follow up |
- What this is
- How overdue each conversation is, stated plainly, worst first — 340 of them.
- Why it exists
- In a six-month buying cycle, “send me something in March” is the single most common thing a prospect says, and the single most commonly forgotten. Before this screen there was no queue — only a memory and a recurring feeling that something had been missed.
- Business outcome
- Long conversations survive their own gaps instead of dying in them. The four-month wait became an asset rather than a leak.
- What this is
- Which identity owes the follow-up — not merely that one is owed.
- Why it exists
- The prospect knows one person. A follow-up arriving under a different name than the original message is worse than no follow-up: it reveals that the conversation was never really with a person at all.
- Business outcome
- Nine identities are operated by one person without any prospect ever noticing that the operation is larger than the relationship.
- What this is
- “Re-checked on open — closed leads drop out automatically.”
- Why it exists
- A queue that was correct when it was written is not the same thing as a queue that is correct now. If somebody replied yesterday and was marked not interested, their follow-up should vanish rather than nag.
- Business outcome
- The list can be worked top to bottom without thinking. Knowing when to stop matters more than knowing when to send.
What changed
Same output.
One tenth the cost.
Not a productivity gain. A different shape of business.
Before · one account per person
200
approaches a week, per salary
Every additional two hundred approaches cost roughly a full salary — and arrived with its own private pile of untracked conversations.
After · nine identities, one operator
1,800
approaches a week, same salary
And an inbox built at the same time as the volume, so nine times the replies did not become nine times the dropped relationships.
The cap is per account. Not per person. That one reading turned a hiring plan into a configuration.
| Before | After | |
|---|---|---|
| Sending identities | 1 per person | 9 across 9 markets |
| Weekly outreach ceiling | ~200 | ~1,800 |
| Total approaches made | Uncounted | 18,400, all attributed |
| Cost of the next 200 a week | One salary | One account |
| People required for that volume | ~10 | 1 |
| Conversations held | Whatever fits in memory | 4,180 |
| Meetings booked | Untracked | 214 |
| The salesperson's day | Sending | Judging and meeting |
The salesperson stopped being the mechanism.
The accounts do the volume. The person reads the replies that matter, judges which prospects are real, follows up properly, and takes the meetings. Sales capacity didn't just increase — it moved up the value chain.
Try it
Scale it the normal way first.
Drag the team to ten and watch the cost. Then read the line that unlocks, and try it the other way.
The account-scaling calculator
Ten seconds. Every founder who has ever debated a headcount request should finish this thinking about their own team.
People
Sending accounts
Locked — the way it is normally done.
But volume alone would have made it worse.
Nine times the outreach is nine times the replies — arriving into a system that was already dropping conversations at one times.
Architecture, for the people who ask
Shuttle runs on top of the existing outreach platform rather than replacing it. The warmed accounts are the asset, and they cannot be migrated, purchased, or rebuilt quickly. What Shuttle owns is everything the platform doesn't: the unified inbox across identities, intent classification, the follow-up engine, the pipeline, and the permanent record.
Much of that platform's API is undocumented, so the integration was reverse-engineered and then confirmed with the vendor's own engineers — including two endpoints that appear in no published documentation and now carry the product's most important numbers.
A competitor was demoed in full and rejected. They asked for account credentials while claiming an official outreach API that does not exist, and cited volumes roughly a hundred times above the platform's real ceiling. When a vendor's central technical claim cannot be true, none of their other claims are evidence of anything. Handing over nine irreplaceable identities on that basis was not a procurement decision; it was a bet.
Next product
Warp
Every email personal. Every one remembered.