the complete ai marketing team workflow
the ai native growth playbook for startups!! you no longer need a full marketing team to grow.
one person with good taste and the right ai systems can do a large part of this work.
you do need a gtm engineer....
who can build systems that find ideas, study customers, create content, launch campaigns, and learn from the results.
the goal is to stop doing the same manual work every day.
i have built systems for 2 ai startups in the last 30 days, they dont have anyone in their marketing team but has generated over 2m+ views on X. i am sharing the sauce for free. here is the ai native growth system i would build for an early startup :
a gtm engineer works between marketing, growth, sales, and engineering.
they understand content and customers, but they can also connect tools, build automations, use apis, and create ai agents.
you need to focus on : how can we build a system that finds good ideas, creates drafts, sends them to the right person, and learns from what performs?
create a system.
1. build a content engine
most teams start every week by asking what they should post.
you can automate a large part of this process.
first, find 20 creators in your niche.
study their best posts and save the formats that work.
look at:
then create custom skills for the best formats.
one skill can turn a product update into a short post.
another can turn a youtube video into an article.
another can rewrite the founder's rough notes in their own style. >do not ask ai to write something viral.
>give it the format, good examples, tone, rules, and mistakes to avoid.
>next, find five reddit communities where your users spend time.
>sort the posts by top posts from this week.
>look for repeated questions, complaints, and strong opinions.
an agent can collect these posts, group similar problems, and turn them into useful content ideas.
you can also send one personalized content idea to every teammate in slack each morning.
1.the founder gets a company story.
2.the engineer gets a technical idea.
3.the designer gets a visual breakdown.
4.the salesperson gets an idea based on a customer objection.
i analyzed all the open source ai humanizers then combined them into one claude skill. check it out it will be useful for you content stack :
2. turn slack messages into marketing tasks
some of the best marketing ideas are already inside slack.
someone says:
>we should write about this. >customers keep asking about this. >this launch needs a demo.
then everyone forgets about it.
connect an ai agent to your marketing channel.
when someone tags it, the agent can:
for all the automation. i made a video on how to do automation :you can connect devin to slack for this. the team can tag devin in a channel, start a task, and receive progress updates in the same thread. you can use@devinai
this is the easiest way to build ai automations:
the simple workflow is:
>team shares idea
>agent collects context
>agent researches and drafts
>human reviews
>content gets published
>results go back into the system
always keep a human review before publishing.
automation should remove repeated work. it should not remove taste.
3. build a customer research loop
customer feedback is usually spread across gmail, slack, support tickets, sales calls, and interview notes.
important information gets lost.
connect claude or another agent to the places where customers already talk.
you can also use listen labs to run customer interviews and understand the results.
the agent can group feedback by:
every week, it can send the team a simple report with:
this creates a strong feedback loop. customers give feedback.
then the agent finds patterns.
product fixes repeated problems.
the team speaks to customers again.
4. turn one recording into many videos
one video usually creates a lot of extra work.
you need to edit it, remove silence, add captions, find short clips, write hooks, and create different versions for every platform.
an ai video pipeline can do most of this.
start with one long recording.
the system can:
you can use remotion with claude to create videos from code and reuse the same templates.
to make ai ads, product videos, and ugc content faster. you can also use @invideoOfficial
record once and turn it into content for youtube, x, linkedin, and short video platforms.
5. find leads when they actually need you
ai has made it easy to send thousands of bad cold emails.
that is not good gtm engineering.
instead, build an agent that looks for buying signals.
for example:
the agent can watch the news, find the company, identify the right person, and prepare a short research note.
then you can reach out at the right time with a message that is actually relevant.
use ai to find the signal and collect the context.
do not use it to send the same message to everyone.
6. turn repeated tasks into skills
this is one of the best habits a gtm engineer can build.
every time you repeat a task, write down the steps.
at the end of the week, ask:
>will i do this again next week? >does it follow the same process? >can i explain what a good result looks like?
if the answer is yes, turn it into a skill.
you can create skills for:
a good skill should include the input, steps, examples, tone, rules, and final quality check.
use it every week and improve it whenever it fails.
over time, these skills become the marketing system for your company. this video is a must watch.