> ## Documentation Index
> Fetch the complete documentation index at: https://docs.agentweb.pro/llms.txt
> Use this file to discover all available pages before exploring further.
> ## Agent Instructions
> ---
> About this documentation: Emma is the AI marketing agent built by AgentWeb (the product is also marketed as HireEmma; AgentWeb, Emma, and HireEmma refer to the same product). Emma learns a company's brand, ICP, and positioning, then executes marketing work inside that company's own accounts: research, ad campaigns on Google Ads, Meta, and LinkedIn Ads, LinkedIn and email outbound, organic content, CRM lead enrichment and ICP scoring, and landing pages deployed to the customer's domain. Actions that spend money or post publicly require the user's confirmation.
> For what Emma can read, draft, publish, and spend on per channel, see https://docs.agentweb.pro/capabilities. For an overview of the product, see https://docs.agentweb.pro/what-is-emma. For terminology (skill, work plan, scheduled chat, template), see https://docs.agentweb.pro/concepts.
# The pipeline system
> How founder-led outbound, organic content, paid ads, and events compose into one compounding pipeline — what order to turn them on in, who owns each step, and where pipelines break.
Every other playbook here covers one motion. This page covers how they compose into a pipeline that keeps producing after you stop pushing it.
The distinction matters. A motion is a thing you run — founder outreach, a content cadence, a paid campaign, an event. A pipeline system is what happens when those motions share the same brand context, the same lead spine, and the same measurement, so each one makes the next one cheaper.
## The shape
```text theme={null}
SOURCES SPINE OUTCOME
Founder-led outbound ---+
|
Organic content ---+----> CRM lead list ----> Enrich
| |
Paid ads ---+ v
| ICP score 0-100
Events ---+ |
v
Prioritized outreach
|
v
Meeting prep --> Close
|
<---- what you learned -------------+
(back into Market Strategy)
```
Four sources, one spine. The value isn't in any single arrow — it's that a lead entering from a paid ad and a lead entering from a LinkedIn connection request land in the same list, get scored against the same ICP, and show up in the same meeting-prep flow.
## Who owns what
Emma executes; you supply judgment at the points where judgment is the actual input.
| Step | Emma does | You do |
| -------------- | -------------------------------------------------- | ----------------------------------------------------------------------------------- |
| Define the ICP | Drafts personas, researches competitors | Validate against real conversations — store only ICPs you've sold to or interviewed |
| Source leads | Searches, imports, discovers matching prospects | Set the criteria, approve the outreach |
| Enrich | Researches and fills fields, with a 2-lead preview | Confirm the preview looks right before the batch runs |
| Score | Scores 0–100 against Market Strategy with a reason | Decide the threshold that means "worth personal outreach" |
| Outreach | Drafts personalized messages in your voice | Approve before anything sends |
| Meeting | Produces a research-backed prep playbook | Have the conversation |
| Learn | Nothing, unless you tell her | Feed what you heard back into Market Strategy |
That last row is the one teams skip, and it's the one that makes the system compound. Emma re-reads Market Strategy on every run — so a sharpened ICP improves lead scoring, ad targeting, content angles, and outreach copy simultaneously. A stale one degrades all four at once.
## Turn the motions on in this order
Running all four at once is the most common way to burn a quarter. Each stage produces the input the next one needs.
Complete [brand onboarding](/onboarding/brand-onboarding): Brand Guidelines, founder profile, Market Strategy, and a real Knowledge Base.
Everything downstream reads from here. If output feels generic later, this is almost always why — not the prompt.
**Done when:** you have 3–5 behavior-based ICP personas, a founder voice description, and a few real posts or case studies in the Knowledge Base.
Run the [founder-led sales playbook](/playbooks/founder-led-sales) and the [LinkedIn content workflow](/playbooks/linkedin-content-workflow) together, not sequentially.
They're a single loop: an ICP-dense first-degree network makes organic content land with actual buyers, and content that lands makes the next connection request likelier to be accepted. Neither works well alone — posting into a network of former colleagues reaches nobody who'd buy, and connecting without posting gives them no reason to remember you.
**Done when:** you're having real conversations weekly, and you can state what buyers respond to in one sentence.
Put [measurement](/playbooks/measurement) in place *before* you spend money, not after. Without a lead-source taxonomy you can't tell which motion produced pipeline, and every later decision is a guess.
Layer in [event-led sales](/playbooks/event-led-sales) when you have something worth gathering people around. Events are the highest-intent source and the fastest to decay — they reward a system that already exists.
**Done when:** you can answer "which motion produced last month's meetings?" from the CRM rather than from memory.
Paid amplifies a message; it doesn't find one. Start [paid ads](/playbooks/paid-ads) only when organic has told you which claim resonates, which ICP converts, and what a qualified lead is actually worth to you.
Over-investing in paid before organic is proven is the most expensive sequencing mistake early-stage teams make. Paid will faithfully scale a message that doesn't work.
**Done when:** cost per *qualified* lead — not cost per lead — is stable enough to budget against.
## The three loops that compound
A pipeline sustains itself when output feeds back as input. Three loops do most of the work.
**Network to content to network.** LinkedIn shows your posts primarily to first-degree connections. Every accepted ICP connection widens the audience for the next post; every post that lands makes the next request likelier to be accepted. This is the cheapest loop you have and the one to start first.
**Paid to landing page to CRM to ICP score to paid.** Ads drive traffic to a page Emma built and instrumented; the form drops leads into a CRM list carrying UTM attribution; Emma scores them against your ICP. That score is what turns raw lead volume into a targeting signal — without it you optimize toward cheap leads instead of good ones.
**Conversations to Market Strategy to everything.** What you hear in meetings is the highest-quality market research you will ever get. Writing it back into Market Strategy propagates it to lead scoring, ad copy, content angles, and outreach on the very next run.
## Making it sustained rather than sporadic
The mechanism is [scheduled chats](/concepts) — a prompt Emma re-runs on a cadence, with results sent somewhere you'll actually read them.
A reasonable standing set for a founder running Stages 1–2:
| Cadence | What runs |
| ------------------- | ---------------------------------------------------------------------------- |
| Weekday morning | Find new ICP leads, draft connection requests, email you the list for review |
| Weekly | Pull last week's channel metrics and lead-source breakdown into one summary |
| Before each meeting | Prospect decode playbook for whoever you're meeting |
| Monthly | Competitor and positioning refresh against Market Strategy |
The full cadence — what each schedule should contain, what the Monday review covers, and what must never be scheduled — is in [operating rhythm](/playbooks/operating-rhythm).
**Scheduled runs don't stop to ask.** They proceed on a best-effort basis rather than pausing for confirmation, so anything that sends, posts, or spends should be restructured to end at a draft before you put it on a cadence.
They also don't notice when they stop finding anything. A schedule pointed at a source that goes quiet keeps producing empty output indefinitely. Tell Emma what to do when there's nothing to report, and review your schedules monthly.
How far to automate, and what evidence should earn each step, is covered in the [autonomy ladder](/playbooks/operating-rhythm#the-autonomy-ladder).
## Where pipelines break
| Failure | What it looks like | Fix |
| -------------------------------- | ------------------------------------------------------------------------------------------ | ---------------------------------------------------------------------------------------------------------- |
| **Motions in silos** | Ad leads in one list, event leads in another, LinkedIn leads in a third, no shared scoring | One lead spine. Same ICP scoring across every source. |
| **No source taxonomy** | "Paid is working" with no way to prove it | [Measurement](/playbooks/measurement) — a lead-source field and a UTM convention, defined before you spend |
| **Paid before message** | Rising spend, flat qualified pipeline | Go back to Stage 1. Paid scales a message; it can't find one. |
| **Stale Market Strategy** | Output drifts generic across every channel at once | Update after conversations, not quarterly |
| **Set-and-forget schedules** | Empty reports nobody reads | Monthly schedule review; explicit stop conditions on fetching steps |
| **Never converting what worked** | Re-explaining the same workflow every month | Say "make this as a workplan" while the thread is still fresh |
## Next steps
* [Founder-led sales](/playbooks/founder-led-sales) — Stage 1 outbound, from profile audit to meeting prep.
* [The content engine](/playbooks/content-engine) — Stage 1 organic, run as a sustained cadence.
* [Operating rhythm](/playbooks/operating-rhythm) — the weekly and monthly cadence that keeps all of this running.
* [Measurement](/playbooks/measurement) — Stage 2. Do this before Stage 3.
* [Paid ads](/playbooks/paid-ads) — Stage 3, and how to know you're ready for it.
* [Event-led sales](/playbooks/event-led-sales) — the highest-intent source, with the shortest window.