Mai-Lan Khong

Content Lead @Frontal_AI (275+ tech cos). AI-built, human-graded content for B2B founders. Grew my own account to Favikon #2 worldwide.

[Mai-Lan Khong @mailankhong]

By Prerona's account, Gojiberry went outbound, then Reddit, then content, then partnerships and X, then paid.

The post's lesson is one channel at a time.

I'd copy the order too, since I think channels are stages that run in sequence.

https://t.co/6Jt3Y2DY12

[Prerona Basu @HelloPrerona]

Hi, I’m Prerona 👋

38, from Bangalore, and this year I joined @gojiberryai as their Content Head.

One of the first things I wanted to understand after joining was how this team got to 4,000 customers in under a year.

I expected some secret distribution channel. There wasn’t one.

The team tested every channel hard, one at a time, based on where MRR was at each stage.

0 → 100 customers: pure outbound
Cold email and LinkedIn, run with an early version of our own product. No brand, no audience. Just finding people showing intent and starting conversations. Ugly, manual, and it worked.

100 → 250: Reddit
The first real breakthrough. Educational breakdowns in SaaS subreddits pulled 10M+ organic views. Traffic quality was mixed, but the volume kept the trial funnel full for months. Cost: close to zero.

250 → 750: content + free blueprints
LinkedIn, YouTube, motion-design videos, and giving away internal systems as free “blueprints.” Content brought the reach, blueprints built the trust, and a good share of readers converted. Founder-led content started compounding here.

750 → 1,500: partnerships + X
B2B influencers, sponsored newsletters, X, and a lifetime affiliate program that turned into a major lever. YC happened around this point too, and the pace went vertical.

1,500+: paid + hiring
Meta ads, Google ads, influencer agencies, and the first serious hiring across growth, sales, engineering and product.

Each stage stacked on the one before. Outbound still runs. Content still runs. The team only added the next lever once the current one was clearly working.

My biggest takeaway as someone who joined partway through: don’t chase five channels at once.

Work one until it works, then add the next.

[Mai-Lan Khong @mailankhong]

@ajinkyaaar Of the four things in your High tier, which one do the DMs usually point back to: the positioning, the contrarian insights, the client results or the content that sells without selling?

[Mai-Lan Khong @mailankhong]

@robimargz At step 4, when they do hire, do your prospects bring the plan Claude or GPT gave them to the first call, or start the brief from zero with you?

[Mai-Lan Khong @mailankhong]

@siddharthwv In your 500-follower example, how many sales calls does it usually take to close those 3 at $3k/month?

[Mai-Lan Khong @mailankhong]

@Anushkaa1407 On point 5, I'd keep meetings booked out of the outcomes an agent is judged on. Booked meetings are easy for any tool to pad, an agent included, so the business goal you hand it in point 1 has to be something that comes after the meeting.

[Mai-Lan Khong @mailankhong]

@itsalexvacca this is such a smart use of data people already have access to! reviews give you so much context for the first message

[Mai-Lan Khong @mailankhong]

Giving the agent real tools is the right half.

A growth org also decides which keyword, list and ad deserve the budget, and I'd keep that call with someone who knows the market, however many tools the agent has.

https://t.co/Lm434gSeC3

[Cody Schneider @codyschneider]

i am once pleading with you, i am begging you

you need to be giving your claude code all the tools it needs to do marketing

so that it can:

- make seed dance 2.5 ads with higgsfield

- do waterfall email enrichments with findymail, lead magic, apollo

- research the blog post it should write for x keyword using serper .dev

- have it scrape creators profiles with apify

- have it research links to build via data for SEO api

when you give it all of these tools

give it all your data

it is now it is your entire growth org

so get all these subscriptions today and give them to your agent

or get all of this here now - https://t.co/yvjwdS5i5N

[Mai-Lan Khong @mailankhong]

I'd underline the line about getting the team posting too.

A founder alone comes across as marketing, and a team posting without the founder won't say what the company itself thinks.

https://t.co/VKxkTZCQff

[Logan Gott @LoganTGott]

Founders Who Actually Get Clients From LinkedIn -

- Post consistently, even when a post flops
- Talk about their business live
- Sound genuinely like themselves
- Turn sales call objections into posts
- Share client results without feeling weird about it
- Have a profile that tells you exactly who they help
- DM the people engaging with their content
- Run outbound alongside content
- Get their team posting and commenting too
- Do launch posts for every new offer or feature
- Care more about pipeline than likes
- Play the long game (90 days minimum)
- Study trends and LinkedIn formats weekly

[Mai-Lan Khong @mailankhong]

When a model checks drafts against your brand rules, I'd have it return the flagged line and its reason along with approve or revise.

I think the verdict alone gives the writing agent very little to work with, and a reason is something you can fix.

https://t.co/fps6Bz5s2z

[Shann³ @shannholmberg]

how to add a decision model to your AI workflow

1,000 yes/no decisions can cost around:

~$10 on GPT-6 Astra
~$4 on Claude Opus 5.5
~$0.04 on JEV

use language models for planning, writing and coding, and decision models for decisions

a decision model can:

> check your agent’s output before it ships
> approve a draft or request revisions
> route a task to the right agent or tool
> classify leads, tickets or comments
> score fit, quality or priority
> check whether content follows your brand rules
> pick an option from a shortlist
> decide whether to continue, stop or request a review

at a much faster speed as well than any regular LLM

here’s how to add one

1. find the decisions in your workflow

look for repeated steps that need a choice, classification or score

start with one decision you can compare against examples you’ve already reviewed

2. define the question and allowed answers

give the model the information and references it needs

for a content workflow, that could include your draft, voice DNA file, anti slop rules and approved examples

ask it to return a choice such as approve, revise or human review

3. connect the answer to an action

your software uses that answer to decide what happens next

approved drafts move to your review queue, revision requests return to the writing agent, and uncertain cases go to you

4. test before letting it act

compare its answers with your reviewed examples

if it returns confidence scores, check whether those scores match how often it gets the decision right

start with actions you can easily reverse

5. put simple checks first

use code for fixed rules, such as required fields and character limits

then use the decision model for checks that need context, before starting more expensive work

6. track quality and cost

measure cost per completed task, incorrect decisions and how often another model or a person needs to review

use the failures to improve your questions, references and allowed answers

[Mai-Lan Khong @mailankhong]

@agamchaudhary_ Does that still hold when the outcome lands months after the work, like a long enterprise sale, or do you argue those raises on earlier signals?

[Mai-Lan Khong @mailankhong]

@pmitu I'd read luck on that list as the other channels arriving late. I don't think a B2B customer comes out of nowhere, so the lucky one has usually met you through something else on the list first.

[Mai-Lan Khong @mailankhong]

@sir4K_zen I'd pick the thread by who is replying in it more than by the size of the account. I think a reply is worth writing when it joins a conversation your buyers are already having.

[Mai-Lan Khong @mailankhong]

@samongaro_ In content I think AI takes over a step rather than a job, the one that turns a clear idea into clean copy, while deciding which idea is worth saying stays with the person.

[Mai-Lan Khong @mailankhong]

@theprasad_ AI can write a clean post on anything, and the post can still be empty when the writer handed over the thinking along with the typing. My read is that people notice the empty post and blame the tool because it's the easier thing to name.

[Mai-Lan Khong @mailankhong]

@jasonlk I'd hand the same agent the comments under your posts. A buyer who comments is showing interest too, just not through the form.

[Mai-Lan Khong @mailankhong]

@MauroJPelle On the Monday review, what makes a template a winner for you: saves, replies, or the DMs that come after it?

[Mai-Lan Khong @mailankhong]

@VukConfidential nice to see someone actually adapting instead of complaining about the algorithm!

[Mai-Lan Khong @mailankhong]

@itsalexvacca thank you for sharing the actual stack and not just the logos!

[Mai-Lan Khong @mailankhong]

Agree on counting relevant conversations before followers.

At 100 followers I'd look for most of them under other people's posts, where your buyers are already talking, more than under your own.

https://t.co/XO9c8EmJfo

[William Salim @WillStrategist]

Your first 1,000 followers on X mean nothing if none of them care about what you do.

When you're starting out, there's one metric I'd track before follower count:

Relevant conversations.

How many people in your target audience are replying to your posts, asking questions, or starting DMs?

100 followers who might buy from you are worth more than 10,000 who never will.

Build the right audience from day one.

[Mai-Lan Khong @mailankhong]

The weekly self-interview is the step I'd copy first.

Your answers about the last 30 days are a cleaner sample of how you write than your old posts, which already carry templates and some AI help, so a model fed those copies the templates too.

https://t.co/K8dys7sTXo

[Noah Kagan @noahkagan]

Had the 2nd best lunch of my year with @codyschneider

We met on X. I said let’s hang. He booked a flight from Idaho.🛬

He showed me how he built his 7-figure company in 8 months on X.

I already do #1 below. It works. Last week: +68,200 impressions. The 10 days before: +10,800.

Here’s his whole system:

1. Remix. He talks about only 3 topics and cuts each one a dozen ways. Same things, over and over.

2. Post 3x a day, every 8 hours. I didn’t want to bug people. Reality: very few of you see each post.

3. Drop the link 1 hour after posting. Drives him thousands of leads a month. I thought links got penalized. X’s own open-source ranking code has no link penalty.

4. Mine old content. Have AI dig through old podcasts and posts for anything worth reusing.

5. Weekly self-interview. AI interviews him about the last 30 days, pulling from his calendar, email and Slack. His answers become posts.

Copy #5. Paste this into Claude with your email and calendar connected:

“Interview me one question at a time about what I did, learned and got wrong in the last 30 days. Check my Slack, email and calendar first. After 10 questions, turn my best answers into 5 posts in my voice.”

Pretty, pretty good. Nice, Cody 👏