The workflows

Integrations assembled into something that runs

An integration on its own is a connection. A workflow is a decision about what happens to a record from the moment it appears to the moment a human acts on it. Eight worked sales stacks, one page each, every one naming the tier it needs and the limit it works around.

They run from three tools to five. The shortest, champion job-change alerts, is a monthly sweep that produces named accounts rather than a filled field, which is why it is the easiest one here to get signed off. The longest crosses five tools and four layers before anything reaches a human.

8worked workflows
3build in an afternoon
1needs no data layer
0need a marketplace app
All eight

Pick by the problem, not by the tools

Each workflow is a chain of integrations from this directory, filtered the same way, because a workflow is only as buildable as its weakest connection. The integration matrix is the same records seen from the other side, every tool against every other, if you would rather start from what you already own.

The data layer

Whichever of these you connect, the integration moves the record. We fill it.

Start for free
Showing 8 of 8
HubSpot→Reverse Contact→Slack

Inbound speed to lead

A form submission enriched, scored, routed and announced in Slack before the visitor closes the tab.

CRMDataCollaboration
Under 60sform to a named owner
3tools in the chain
Hoursto build
HubSpot→Clay→Reverse Contact→n8n

The CRM data layer

Stop maintaining records by hand. A scheduled loop that fills gaps, respects what reps own and never overwrites a human.

CRMOrchestrationData
Weeklyrefresh cycle
4tools in the chain
Daysto build
LinkedIn Sales Navigator→Clay→Reverse Contact→HubSpot→Smartlead

Signal to sequenced contact

A Sales Navigator search becomes an enriched, deduplicated, CRM-checked list in a live campaign.

ProspectingOrchestrationData
5 stepssearch to first touch
5tools in the chain
Daysto build
Reverse Contact→HubSpot→Slack

Agent-written account briefs

Two MCP servers, no pipeline. An agent researches the account and writes the brief onto the record.

DataCRMCollaboration
0 linesof glue code
2MCP servers, no pipeline
Hoursto build
HubSpot→ClickUp→Slack

Closed won to delivery

The deal closes and the work starts itself, with the terms attached and nobody retyping a thing.

CRMCollaboration
0fields retyped
3tools in the chain
Hoursto build
HubSpot→Reverse Contact→Slack

Champion job-change alerts

Your best customer contact starts somewhere new. That is a warm pipeline nobody is watching.

CRMDataCollaboration
Monthlysweep of your champions
3tools in the chain
Hoursto build
HubSpot→Reverse Contact→n8n

Deduplication and identity resolution

Four records for one person, and every report quietly counting them as four.

CRMDataOrchestration
Confidencedecides, not a fuzzy match
3tools in the chain
Daysto build
Salesforce→Clay→Reverse Contact

The Salesforce data layer

The same enrichment loop, against an org with validation rules, triggers and an API allocation.

CRMOrchestrationData
Sandboxfirst, always
3tools in the chain
Daysto build
How to build one

The order that stops a workflow quietly failing

Every workflow on this page was assembled in the same order, and it is not the order people reach for. The temptation is to start with the exciting part, which is nearly always enrichment or AI, and to leave the boring parts for later. The boring parts are what make the exciting part get used.

Start with the write path. Decide which properties each source may touch before a single record moves, and give every external source its own fields rather than the ones your reps type into. This takes an afternoon and it is the difference between a pipeline that runs for years and one that gets switched off in month three after an argument about a job title.

Then build the notification. A workflow nobody hears about is a workflow nobody trusts, and the fastest way to lose the room is to enrich three thousand records perfectly and have a rep discover it by accident. One trigger, one channel, one named owner.

Only then add the enrichment, and measure it before you scale it. Run five hundred rows, count what came back, and decide from that number rather than from a vendor's global match rate. Nobody publishes a figure you can hold them to, including Reverse Contact, whose 70%+ identity match rate is a global average rather than a promise about your particular list.

Last, put it on a schedule. A backfill is a project. A loop is a data layer, and the difference is one setting and a decision about how often a record goes stale.

Write path first
Which properties may each source write to, and what happens when it disagrees with a human. Decide before anything moves, not after.
Notification second
One trigger, one channel, one owner. If nobody is accountable for acting on it, it is a log entry rather than an alert.
Enrichment third, measured
Five hundred rows through the real pipeline, then count. Your match rate on your list is the only figure that predicts your bill.
Schedule last
A loop with a refresh cadence is a data layer. A one-time backfill is a project that decays from the day it finishes.
For agents

Every workflow ships a skill file

Each one carries the API surface of every tool in its chain, the build order, the documented limits and the checks that prove it worked. Plain markdown with YAML frontmatter, so it loads into Claude Code, Codex, Cursor or anything else that reads a file or fetches a URL. An agent that would rather query the dataset than read a skill has the MCP server instead, over the same records.

The point of these is not that an agent can write the code. It is that an agent starting from a blank prompt does not know that HubSpot's webhook action lives in Data Hub, that Reverse Contact's live endpoints return a task rather than a record, or that a Clay action bills on the attempt. Those are the facts that decide whether the build works, and they are in the file.

01

hubspot-speed-to-lead

Build an inbound speed-to-lead pipeline: a HubSpot form submission enriched by Reverse Contact, scored and routed on the enriched fields, and announced to a named owner in Slack inside a minute.

4 APIs →
02

crm-data-layer-loop

Build a scheduled enrichment loop over a CRM: pull the records missing a field, fill the gaps with Reverse Contact, write back without overwriting anything a human owns, and run it on a cadence.

4 APIs →
03

outbound-list-suppression

Build an outbound list pipeline that enriches a prospect list, checks every row against the CRM for customers, open deals, recent contact and existing owners, and pushes only the survivors into a sequencing tool.

5 APIs →
04

mcp-account-brief

Connect the Reverse Contact and CRM MCP servers to an AI client and have it research an account and write a pre-call brief onto the CRM record.

2 APIs →
05

closed-won-to-delivery

Build a closed-won handover: a won deal creates delivery work with the line items attached, and work status reports back to the CRM record.

3 APIs →
06

champion-job-change-alerts

Build a monthly sweep that detects when a past customer contact changes company and turns it into pipeline: re-resolve the champion list, diff against the CRM, and alert the account owner on confirmed moves.

4 APIs →
07

crm-identity-dedupe

Build a CRM deduplication pass that resolves records to an external identity rather than fuzzy-matching names, merges only above a confidence threshold, and runs a dry run first.

3 APIs →
08

salesforce-enrichment-loop

Build a scheduled enrichment loop against Salesforce, designed around the org rules that reject writes: validation rules, required fields, duplicate rules, triggers and the shared API allocation.

3 APIs →

Served as plain text, so you can read one in a tab, save it, or hand the URL straight to an agent. Every figure in them carries the date it was verified.

Seven of these eight run on the same data layer

Reverse Contact resolves an email, a domain or a profile URL into a verified people and company record, through one API, one schema and a confidence score you can branch on. No marketplace app, four routes into HubSpot, and seven into the rest of the stack.

7 days of evaluation access, no card required.

FAQ

Questions about building these

The ones that come up once someone has decided to build rather than to buy.

Which workflow should I build first?

Inbound speed to lead, if you have any inbound at all. It is three tools, it builds in an afternoon, and it produces a visible result the same week, which matters more than it should when you are asking for budget for the next one. The CRM data layer has the highest return and is a harder sell, because its output is the absence of a problem.

Do I need Clay or n8n, or can I do this in the CRM?

It depends on one thing: whether you own HubSpot Data Hub Professional. Calling an external API from inside a HubSpot workflow requires the webhook or custom code action, and both live in Data Hub rather than Sales Hub. If you have it, build in the CRM and skip the middle layer. If you do not, Clay or n8n is cheaper than the hub upgrade by a wide margin, and that is the normal case rather than the exception.

What does one of these cost to run?

The tooling is mostly cheap or free: n8n self-hosts at no cost, Slack and HubSpot's free tiers carry real work, and Clay has a free plan. The metered line is enrichment, and we cannot price it here honestly, because Reverse Contact publishes a $100 pay-as-you-go entry and no public per-lookup rate. Ask for that figure against your own expected volume, then multiply by the rows you actually intend to enrich rather than by your database size.

Does every workflow need an enrichment step?

No, and one of these deliberately has none. Closed won to delivery moves data your own team already entered during the deal, so pointing an enrichment tool at it would add cost without adding anything. Not every gap in a sales stack is a data problem, and a directory published by a data company should be willing to say so.

How long before one of these breaks?

Roughly when a vendor changes a tier. The connections themselves are stable; what moves is what your plan is allowed to do, and that changes without an announcement. The second failure mode is silent: a Slack channel gets noisy, someone mutes it, and the workflow keeps running while nobody reads the output. Both are worth a quarterly check that takes ten minutes.

Can an agent replace one of these?

Part of one. An agent with access to an MCP server for data and one for the CRM can do the research-and-write-a-brief job that used to need a pipeline, which is exactly what the agent workflow here is. It does not replace a scheduled two-way sync, because that is a pipeline and it should stay one. What it replaces is the one-off task somebody was about to automate badly.

Tier requirements, sync behaviour and limits read from each vendor's own documentation in September, October and November 2026. Vendors change these without notice.

Reverse Contact appears in this directory as a data provider and publishes it. Every other tool is here on merit and nobody pays for placement.

Ask an assistant