Rocket SaaS is building an AI-powered outbound motion

Rocket SaaS runs marketing programs for B2B teams. One channel they use for new client acquisition is cold email. But email outbound has gotten harder. Everyone is sending more messages, so personalisation has to be genuinely relevant, not cosmetic.

The problem was simple:

  • Manual website audits do not scale
  • LLM only outputs can be inconsistent, with formatting changes and variable structure
  • Outbound workflows break when the output is not predictable


Rocket SaaS wanted a system that could run in the background, produce the same output structure every time, and be improved continuously without rebuilding the whole workflow.


Why Rocket SaaS built this in Rightbrain, not directly in Clay

Clay can already call models directly. Rocket SaaS chose Rightbrain because it made the workflow:


  • More consistent: structured outputs every time, so emails do not break when the response changes
  • Cheaper: more cost effective model usage, rather than burning Clay credits on expensive defaults
  • Easier to optimise: compare models quickly, swap to cheaper ones once the prompt is dialled in
  • More reliable: set a backup model so the workflow does not fail if one provider is down


“We could call an LLM directly, but Rightbrain made the outputs consistent and easy to improve. That matters when it’s feeding emails automatically.”
Joe McLaughlin, Rocket SaaS


Building a website audit Task with a scoring matrix

Rocket SaaS already had an internal landing page builder with a strong prompt based on best practice. They reverse engineered that thinking into a website audit scoring matrix.

The Rightbrain Task takes a prospect’s website and returns:


  • A score against the landing page best practice criteria
  • Three clear bullet point recommendations to improve the page
  • A structured output format that is identical every time


That predictable structure is what makes it safe to pipe directly into outbound messaging.


From Clay to Rightbrain to Instantly, fully automated

The workflow runs end to end without manual steps:


Clay creates or updates contacts from engagement

As Clay identifies new contacts engaging with Rocket SaaS ads, contacts are created in the workflow.


Clay calls Rightbrain via HTTP enrichment

Clay triggers the Rightbrain Task using an API call in the background.


Rightbrain returns a structured audit output

The Task outputs a consistent set of fields, like score plus three recommendations.


Clay populates fields and pushes to Instantly

Those fields are mapped into Clay columns and pushed into Instantly, where the audit insights become personalised lines inside outbound emails.


Model testing and prompt optimisation without breaking the system

Rocket SaaS started with a higher cost model to get quality right, then:

  • Tightened the prompt based on real outputs
  • Swapped to cheaper models to reduce cost and improve speed
  • Added fallback models so the enrichment does not fail if one model endpoint is unavailable


This makes the system scalable and keeps performance improving over time.

“We started expensive to get the output right, then used Rightbrain to test alternatives and bring cost down fast, without losing consistency.”
Joe McLaughlin, Rocket SaaS


Impact (so far)

This is early, but already proving the motion works:


  • Fully automated outbound flow that runs continuously
  • Good open rates, and first replies coming in
  • A repeatable framework for adding micro intelligence into outreach, without adding manual work


If you want, we can add metrics here like audits per day, average cost per audit, reply rate lift, and time saved.


What’s next: more Tasks across outbound and the agency

Rocket SaaS is using this as the first anchor use case, then expanding:


1) New outbound campaigns powered by Rightbrain Tasks

More variations of audit depth, angles, and triggers to level up automated outbound.


2) A brand and QA checker for client work

A Rightbrain Task trained on a client’s brand guidelines that can:


  • Grade assets before publishing
  • Check grammar and US or UK spelling
  • Validate required keywords and rules
  • Block anything below a threshold score for review


3) A full workflow audit to find the best places for automation

Rightbrain’s team helps identify where moments of micro intelligence will create the biggest leverage, then turn them into production ready Tasks.



“We could call an LLM directly, but Rightbrain made the outputs consistent and easy to improve. That matters when it’s feeding emails automatically.”

 Joe McLaughlin, Rocket SaaS

Rightbrain Logotype

Log in

Book Demo

Fireworks Image Background

Ready to ship your next AI feature?

Book demo

Rocket SaaS is building an AI-powered outbound motion

Rocket SaaS runs marketing programs for B2B teams. One channel they use for new client acquisition is cold email. But email outbound has gotten harder. Everyone is sending more messages, so personalisation has to be genuinely relevant, not cosmetic.

The problem was simple:

  • Manual website audits do not scale
  • LLM only outputs can be inconsistent, with formatting changes and variable structure
  • Outbound workflows break when the output is not predictable


Rocket SaaS wanted a system that could run in the background, produce the same output structure every time, and be improved continuously without rebuilding the whole workflow.


Why Rocket SaaS built this in Rightbrain, not directly in Clay

Clay can already call models directly. Rocket SaaS chose Rightbrain because it made the workflow:


  • More consistent: structured outputs every time, so emails do not break when the response changes
  • Cheaper: more cost effective model usage, rather than burning Clay credits on expensive defaults
  • Easier to optimise: compare models quickly, swap to cheaper ones once the prompt is dialled in
  • More reliable: set a backup model so the workflow does not fail if one provider is down


“We could call an LLM directly, but Rightbrain made the outputs consistent and easy to improve. That matters when it’s feeding emails automatically.”
Joe McLaughlin, Rocket SaaS


Building a website audit Task with a scoring matrix

Rocket SaaS already had an internal landing page builder with a strong prompt based on best practice. They reverse engineered that thinking into a website audit scoring matrix.

The Rightbrain Task takes a prospect’s website and returns:


  • A score against the landing page best practice criteria
  • Three clear bullet point recommendations to improve the page
  • A structured output format that is identical every time


That predictable structure is what makes it safe to pipe directly into outbound messaging.


From Clay to Rightbrain to Instantly, fully automated

The workflow runs end to end without manual steps:


Clay creates or updates contacts from engagement

As Clay identifies new contacts engaging with Rocket SaaS ads, contacts are created in the workflow.


Clay calls Rightbrain via HTTP enrichment

Clay triggers the Rightbrain Task using an API call in the background.


Rightbrain returns a structured audit output

The Task outputs a consistent set of fields, like score plus three recommendations.


Clay populates fields and pushes to Instantly

Those fields are mapped into Clay columns and pushed into Instantly, where the audit insights become personalised lines inside outbound emails.


Model testing and prompt optimisation without breaking the system

Rocket SaaS started with a higher cost model to get quality right, then:

  • Tightened the prompt based on real outputs
  • Swapped to cheaper models to reduce cost and improve speed
  • Added fallback models so the enrichment does not fail if one model endpoint is unavailable


This makes the system scalable and keeps performance improving over time.

“We started expensive to get the output right, then used Rightbrain to test alternatives and bring cost down fast, without losing consistency.”
Joe McLaughlin, Rocket SaaS


Impact (so far)

This is early, but already proving the motion works:


  • Fully automated outbound flow that runs continuously
  • Good open rates, and first replies coming in
  • A repeatable framework for adding micro intelligence into outreach, without adding manual work


If you want, we can add metrics here like audits per day, average cost per audit, reply rate lift, and time saved.


What’s next: more Tasks across outbound and the agency

Rocket SaaS is using this as the first anchor use case, then expanding:


1) New outbound campaigns powered by Rightbrain Tasks

More variations of audit depth, angles, and triggers to level up automated outbound.


2) A brand and QA checker for client work

A Rightbrain Task trained on a client’s brand guidelines that can:


  • Grade assets before publishing
  • Check grammar and US or UK spelling
  • Validate required keywords and rules
  • Block anything below a threshold score for review


3) A full workflow audit to find the best places for automation

Rightbrain’s team helps identify where moments of micro intelligence will create the biggest leverage, then turn them into production ready Tasks.



“We could call an LLM directly, but Rightbrain made the outputs consistent and easy to improve. That matters when it’s feeding emails automatically.”

 Joe McLaughlin, Rocket SaaS

Rightbrain Logotype

Log in

Book Demo

Fireworks Image Background

Ready to ship your next AI feature?

Book demo

Rocket SaaS is building an AI-powered outbound motion

Rocket SaaS runs marketing programs for B2B teams. One channel they use for new client acquisition is cold email. But email outbound has gotten harder. Everyone is sending more messages, so personalisation has to be genuinely relevant, not cosmetic.

The problem was simple:

  • Manual website audits do not scale
  • LLM only outputs can be inconsistent, with formatting changes and variable structure
  • Outbound workflows break when the output is not predictable


Rocket SaaS wanted a system that could run in the background, produce the same output structure every time, and be improved continuously without rebuilding the whole workflow.


Why Rocket SaaS built this in Rightbrain, not directly in Clay

Clay can already call models directly. Rocket SaaS chose Rightbrain because it made the workflow:


  • More consistent: structured outputs every time, so emails do not break when the response changes
  • Cheaper: more cost effective model usage, rather than burning Clay credits on expensive defaults
  • Easier to optimise: compare models quickly, swap to cheaper ones once the prompt is dialled in
  • More reliable: set a backup model so the workflow does not fail if one provider is down


“We could call an LLM directly, but Rightbrain made the outputs consistent and easy to improve. That matters when it’s feeding emails automatically.”
Joe McLaughlin, Rocket SaaS


Building a website audit Task with a scoring matrix

Rocket SaaS already had an internal landing page builder with a strong prompt based on best practice. They reverse engineered that thinking into a website audit scoring matrix.

The Rightbrain Task takes a prospect’s website and returns:


  • A score against the landing page best practice criteria
  • Three clear bullet point recommendations to improve the page
  • A structured output format that is identical every time


That predictable structure is what makes it safe to pipe directly into outbound messaging.


From Clay to Rightbrain to Instantly, fully automated

The workflow runs end to end without manual steps:


Clay creates or updates contacts from engagement

As Clay identifies new contacts engaging with Rocket SaaS ads, contacts are created in the workflow.


Clay calls Rightbrain via HTTP enrichment

Clay triggers the Rightbrain Task using an API call in the background.


Rightbrain returns a structured audit output

The Task outputs a consistent set of fields, like score plus three recommendations.


Clay populates fields and pushes to Instantly

Those fields are mapped into Clay columns and pushed into Instantly, where the audit insights become personalised lines inside outbound emails.


Model testing and prompt optimisation without breaking the system

Rocket SaaS started with a higher cost model to get quality right, then:

  • Tightened the prompt based on real outputs
  • Swapped to cheaper models to reduce cost and improve speed
  • Added fallback models so the enrichment does not fail if one model endpoint is unavailable


This makes the system scalable and keeps performance improving over time.

“We started expensive to get the output right, then used Rightbrain to test alternatives and bring cost down fast, without losing consistency.”
Joe McLaughlin, Rocket SaaS


Impact (so far)

This is early, but already proving the motion works:


  • Fully automated outbound flow that runs continuously
  • Good open rates, and first replies coming in
  • A repeatable framework for adding micro intelligence into outreach, without adding manual work


If you want, we can add metrics here like audits per day, average cost per audit, reply rate lift, and time saved.


What’s next: more Tasks across outbound and the agency

Rocket SaaS is using this as the first anchor use case, then expanding:


1) New outbound campaigns powered by Rightbrain Tasks

More variations of audit depth, angles, and triggers to level up automated outbound.


2) A brand and QA checker for client work

A Rightbrain Task trained on a client’s brand guidelines that can:


  • Grade assets before publishing
  • Check grammar and US or UK spelling
  • Validate required keywords and rules
  • Block anything below a threshold score for review


3) A full workflow audit to find the best places for automation

Rightbrain’s team helps identify where moments of micro intelligence will create the biggest leverage, then turn them into production ready Tasks.



“We could call an LLM directly, but Rightbrain made the outputs consistent and easy to improve. That matters when it’s feeding emails automatically.”

 Joe McLaughlin, Rocket SaaS

Rightbrain Logotype

Log in

Book Demo

Fireworks Background Image

Ready to ship your next AI feature?

Book demo