Demo outbound build · archive-ark · built 19 Oct 2026

An outbound build that reads the company's own words first.

Most outbound builds start with a list and a template. This one starts with a context file: what the company's own people say about who buys, what triggers the purchase and what hurts. Every email then has to trace back to it.

The example company is Pylon, a B2B support platform. I don't have Pylon's sales calls, so the context file is built from six public interviews with its founders. For your team, it would be built from your recorded calls.

Unsolicited demo. Not affiliated with, commissioned by or reviewed by Pylon. Every fact comes from public interviews and public job posts. Sample emails are signed by a placeholder rep, not a real person. The accounts in the table are described, not named, on this public copy.

Part 1

Context file

What the founders' own words say about buyer, trigger and pain.

Part 2

Signal pipeline

503 companies scanned for the one trigger the context file names.

Part 3

Table

8 accounts kept, 16 removed, and the reason for every row.

Part 4

Sequence

The same account written two ways: a template, and from the context file.

Part 1

Context file: what the founders say on tape

Six interviews, about 77,000 words of auto-captions, mined for buyer, trigger, pain, competition and claims to avoid. Each line links to the second it was said. Each one ends with the rule it sets for the copy.

Who buys

"the up to like 500 employees generally uh customer persona"

Pylon co-founder · Pylon: Reimagining B2B Customer Support, 38:05

"they're usually the first buyer" (the support team, before customer success, solutions and account management)

Marty Kausas · Supercharging B2B Support, 15:40 RuleB2B, up to about 500 people. Open with the support or CS seat, not the CFO.

Trigger

"a lot of B2B companies were starting to talk to their customers [in] shared Slack channels, Microsoft Teams channels, Discord, WhatsApp, all that stuff instead of just email"

Marty Kausas · Supercharging B2B Support, 2:15 RuleThe signal is a company that already supports customers in Slack. Funding alone is not the trigger.

Core pain

"shared Slack channels you literally just cannot track the tickets in there"

Marty Kausas · Supercharging B2B Support, 3:16

"a shared Slack channel that you have with a customer. There's no clear start or end."

Marty Kausas · Modern B2B Support, AI KB, 21:52

"their important customers that they actually cared about were all in shared slack channels so they had zero coverage for their important customers"

Marty Kausas, on Hightouch (Series B, about 70 people) before Pylon existed · Modernizing B2B support, 6:40 RuleName the untracked channel. The pain is lost visibility on the biggest accounts, not slow agents.

Competition

"the company that we go up against most is Zendesk"

Marty Kausas · Modernizing B2B support, 19:01

"Intercom doesn't support that. Fresh[desk] doesn't support that. Salesforce even though they own Slack don't support that"

Marty Kausas · Supercharging B2B Support, 3:07 RuleWhen a job post names Zendesk or Intercom, the angle is the missing channel, not a feature list.

Price

"since it's an existing spend ... We're just replacing that cost"

Marty Kausas · Supercharging B2B Support, 12:27 RuleNever promise savings. They price against the tool you already pay for.

AI claims

"these are not functionalities like autoresolving like a ticket because the B2B nature and the very unique context nature of these tickets"

Pylon co-founder · Pylon: Reimagining B2B Customer Support, 34:34

"like 90% of the time the answer is right but there's cost when it's wrong"

Marty Kausas · How I Invest, 27:44 RuleNo "AI deflects 50% of tickets". The founders say the opposite on the record.

Not in public material

Objections, lost-deal reasons, the words buyers use on discovery calls, which persona says yes. Interviews are the founders talking. These answers live in a team's recorded sales calls, and that's the first thing I'd ask your team for.

Part 2

Signal pipeline: companies whose job posts admit the pain

The context file says the trigger is support already running in customer Slack channels. Companies write that down in their own job posts. So the pipeline reads job posts, not funding news.

503B2B companies in the source lists

189at least, with a public job board I could read

24with a customer-facing role that mentions Slack

16removed: post names Pylon, false match, or too big

8kept for the table

Source list. 414 active B2B YC companies with 10 to 500 people that are hiring, plus 117 Series A to C companies from my own weekly funding scan (28 overlap).
Read every open role from the public Ashby, Greenhouse and Lever job-board APIs. This found 20 companies. A Google search of job boards, run first, found only 4 more.
Role filter: support, success, solutions, implementation, deployment and forward-deployed roles only.
Phrase filter: the post has to put customers in Slack ("shared Slack channels", "Slack Connect", "in their Slack", "via Slack"). The matching sentence is saved as the evidence.
Suppression: any post that already names Pylon is removed, as the tool they use or one candidates should know. Seven posts.
Human pass: sentences where Slack is a former employer or an internal tool are removed. Eight false matches.

Delivered as a Clay table in your workspace. This demo ran on my own scripts; in Clay the same steps are an HTTP API column per job board, a formula filter and an AI column for the human pass.

Part 3

Table: 8 accounts, each with its own evidence

Tier A: the role is open now and the post puts support in customer Slack. Tier B: the same signal, but the post is older or the role is forward-deployed rather than support. Headcounts are from the YC directory or the post and can lag. Contacts are not enriched in this demo; for a real build, this is where email lookup runs.

Accounts are described rather than named on this public copy. Job posts were read on 19 Oct 2026, and some roles will have closed since.

Account	People	Role · posted	Evidence (their words)	Pain line it matches	Write to	Tier
Account 1 Open-source observability · SF HQ, role in India	38	Founding Customer Success Lead 2026-10-01	"decide what each tier gets: a named owner, a shared Slack channel, regular business reviews, or pooled coverage"	No clear start or end	Founders (post: "work directly with the founders")	A
Account 2 Identity security	not stated	Senior Customer Success Engineer 2026-10-14	"Take tier 2/3 issues from tickets and shared Slack channels"	Cannot track the tickets in there	Head of Customer Success	A
Account 3 Developer build infrastructure	not stated	Customer-Facing Software Engineer 2026-10-02	"a global customer base across the US, EU, Asia and Australia, primarily async via Slack and email"	Cannot track the tickets in there	Founders	A
Account 4 AI sales training · SF	30	Technical CSM, Mid-Market 2026-09-23	"Responsiveness: Fast on Slack, tickets, and email"	Response times nobody can measure	Head of Customer Success	A
Account 5 Trust and safety software · NYC HQ, role in London	30	Forward Deployed Engineer 2026-09-28	"part of their engineering team, in their Slack, on their calls"	Zero coverage for important customers	CEO	B
Account 6 Speech AI API · SF	not stated	Senior FDE, Strategic Accounts 2026-08-31	"comfortable on a customer's Slack channel walking their team through an API integration"	Zero coverage for important customers	VP Customer Success	B
Account 7 HR integrations API · Berlin/NYC	77	Developer Support Engineer (NYC) 2026-02-04 · old	"Assist customer engineers via Slack, email & occasional video calls"	Cannot track the tickets in there	Head of Support	B
Account 8 Enterprise AI platform · NYC	14	Customer Deployment Engineer 2026-04-11 · old	"Live in their slack channels and meet them at their desks"	Zero coverage for important customers	CEO	B

7 removed: the post already names Pylon

Seven posts name Pylon, either as the support tool the company runs or as one candidates should know. For a real build these go to an expansion list, worth checking before any outbound.

8 removed: false matches

2 posts: Slack named as a past employer of the team
3 posts: Slack as an internal tool
1 post: a customer community, not support
1 post: a partnerships role
1 post: a job aggregator reposting another company's role

1 removed: size

The directory says 65 people, but the company hires support in five cities and runs Zendesk. Directory headcount lags; it's far past the band.

Part 4

Sequence: the same account, written two ways

Left is a template-first build: enrichment, then a template. I wrote it to show the pattern; it is not Pylon copy. Right is written from the context file. The notes under each email show where every line came from.

Template-first build

Template

Illustrative, written by me, not Pylon copy · To: Account 1 founder · Subject: Scaling support at {company}

Hi {first_name},

Congrats on {company}'s 4x revenue growth! I noticed you're hiring a Founding Customer Success Lead, which usually means support volume is growing fast.

Pylon is the AI-native support platform for B2B companies. Our AI agents can deflect up to 50% of tickets and save your team hours every week.

Worth a quick 15-minute chat this week?

"Deflect up to 50%" contradicts the founders, who say they don't auto-resolve B2B tickets.
"Save your team hours" sells savings. Pylon prices against the tool you already pay for.
The hire is real, but the email never says what the post says.

Context-aware build

Email 1 · day 0

To: Account 1 founder · Subject: which tier gets the Slack channel

Your Founding Customer Success Lead post asks the hire to decide which accounts get "a shared Slack channel."

Once a top account has one, support questions, feature asks and renewal hints land in the same thread. There's no ticket, and no clear start or end.

Most help desks were built for email, not shared channels. Pylon was built for them: each message can become a ticket the CS lead sees across every account.

Worth a look before the tiers are set?

[Pylon rep]

Hook: the account's own post, quoted.
"No clear start or end": Marty Kausas, 21:52.
Help-desk line: the channel gap the founders describe, kept general rather than naming competitors (a real client confirms it before it ships).
No AI claim, no savings claim (context-file rules).

Email 2

Day 3 · new move: their health-score plan

Same thread

The same post wants health scores built from "support tickets and engagement" in your warehouse.

If your biggest accounts ask for help in Slack, those requests never become tickets, so the score misses the accounts that talk to you most.

Who owns that data until the hire starts?

Quote: the account's post, section "Account coverage and health".
Matches the Hightouch story: "zero coverage for their important customers" (6:40).

Email 3

Day 7 · new move: a story, no ask

Same thread

One story from our CEO's interviews: Hightouch, at about 70 people, ran Intercom for self-serve users while its biggest customers sat in hundreds of shared Slack channels. The team marked messages unread to remember them.

Your post says most revenue is usage-based and larger accounts drive growth. Same split.

Is that where your top accounts are?

Story: Marty Kausas, Modernizing B2B support, 5:44 to 7:00. Presented as the discovery story, not as a customer claim.
"Most of our revenue is usage-based": the account's post, verbatim.

Email 1 for two more rows

Account 2

Tier A

To: Head of Customer Success · Subject: two queues for tier 2/3

Your Senior Customer Success Engineer post has the hire taking "tier 2/3 issues from tickets and shared Slack channels."

That's two queues, and only the ticket one has response times. The Slack half is tracked by whoever remembers it.

Pylon puts both in one inbox, so an enterprise thread in Slack gets the same clock as a ticket.

Which queue runs slower today?

Account 4

Tier A

To: Head of Customer Success · Subject: fast on Slack, measured how?

Your Technical CSM post asks for someone "Fast on Slack, tickets, and email."

Two of those come with a timestamp you can report on. In a shared Slack channel there's no ticket, so there's no response time to measure either.

Pylon turns Slack messages into tickets, so "fast" gets a number.

How do you measure Slack speed now?

For your team

Want this built on your own calls?

This demo had six public interviews to work from. Your team has something better: recorded sales calls, where buyers say what they need, what they object to and why deals are lost.

I build the context file from those calls, your CRM notes and your docs, in a repo your team owns. Your outbound and your AI then draft from it, so the copy uses your buyers' words and avoids the claims your own team would never make.

The first step is a short call to see whether your calls are enough to start.

Book a 20-minute call → Back to the site