Hey Freelance Friends!
I started building ChatGPT bots for clients for one reason: I was tired of watching smart people run their businesses like everything had to live inside their head.
The same questions came up every week. The same decisions got re-decided. The same instructions had to be rewritten. And the founder stayed stuck in the loop, even when they were “delegating.”
Once you build a bot properly, it stops being a novelty and starts becoming what clients actually want: a reliable shortcut to decisions, answers, and execution.
This week, I’m going to show you how to build one that makes your freelance career easier, and how to turn it into something clients will happily pay for.
Sponsored
It’s the platform behind some of the world’s most-read newsletters, built for writers who want clean publishing, real growth tools, and monetisation that doesn’t feel bolted on as an afterthought. No duct tape. No fighting your software.
Freelance Forward runs on Beehiiv for a reason. It stays out of the way and lets the writing do the work.
If a newsletter is even a quiet idea at the back of your mind, this is the infrastructure part done right.
The real value of a bot isn’t the bot
Clients don’t wake up wanting “a chatbot”.
They want less admin. Less back-and-forth. Less confusion. Less time wasted repeating themselves. They want the business to run smoother without needing to be present for every little decision.
A good bot does three things:
It makes answers consistent
It makes delivery faster
It keeps work from falling through cracks
That’s why this is such a useful skill as a freelancer. You’re not just selling your time anymore, you’re selling something reusable.
And once you’ve built a few of these, you start seeing the pattern: most businesses are full of repeatable processes… they just don’t realise it until you point it out.
Step one: pick a problem that repeats
If you want this to work in the real world, don’t start with a “cool bot idea”.
Start with a problem that keeps showing up.
Here are the ones I see constantly when working with clients:
“Can you resend the info again?”
“What’s the process here?”
“How do I respond to this client?”
“What do we charge for this?”
“Can you summarise this brief?”
“Can you turn this into an email / proposal / invoice?”
“Where do I find that doc?”
“How do I explain this service clearly?”
If it’s something you’ve answered more than three times, it’s a bot candidate.
And if it’s something a founder answers every week, it’s a bot that saves them hours.
That’s the difference between a toy and a tool.
Step two: decide what the bot will do (and what it won’t)
This is where most people mess it up.
They try to build a bot that can do everything. It ends up doing nothing properly.
Your bot should have a job title, not a personality.
Here are examples that work:
“Proposal Writer for a Web Designer”
“Client Onboarding Assistant for a Marketing Agency”
“Rate Calculator for South African Freelancers”
“Internal Operations Assistant for a small retail business”
“Customer Support FAQ Helper for a service business”
“Content Repurposing Assistant for a newsletter creator”
The best bots are narrow. Clear inputs, clear outputs.
If you want clients to trust it, it needs boundaries.
Step three: write the instructions like you’re training a new hire
When I build bots for clients, I don’t write instructions like I’m talking to a robot.
I write them like I’m onboarding a very capable assistant who needs rules, context, and examples.
Your instruction should include:
1) The role
What this bot is responsible for.
2) The tone
How it should sound (professional, direct, friendly, etc.)
3) The rules
What it must always do and must never do.
4) The output format
What the final answer should look like.
Here’s a clean framework you can copy and paste:
Role:
You are a [specific assistant] helping [type of user] do [specific task].
Goal:
Your job is to produce [specific output] that is ready to use without rewriting.
Rules:
Ask 1–3 questions if information is missing.
If the user asks for something outside scope, tell them what you can do instead.
Keep responses structured and practical.
Never invent details, prices, or policies.
Output format:
Always provide:
Version 1 (short and direct)
Version 2 (more detailed, more persuasive)
Optional subject lines / CTA (if relevant)
That’s it. That’s the whole trick.
The bot becomes useful when it produces something you can actually send.
Step four: give it the right information (this is where it becomes “client-ready”)
A bot is only as good as the information it’s allowed to work with.
If you’re building it for yourself, you might feed it things like:
your services and pricing
your tone of voice
your FAQs
your process (how you onboard, how you deliver, how you invoice)
examples of your best writing (emails, proposals, messages)
If you’re building it for a client, you’ll usually need:
service descriptions
pricing packages
policies (refunds, turnaround times, boundaries)
common customer questions
brand voice guidelines
internal SOPs (even if messy)
This is where you become valuable: most clients have the information, but it’s scattered across WhatsApps, old Google Docs, and someone’s memory.
You’re not just building a bot. You’re organising their business into something usable.
That’s a real deliverable.
Step five: test it like a client would
Don’t test it with polite questions.
Test it with real situations.
If you’re building a bot that writes proposals, paste a messy client brief and see if it can:
extract the objective
propose a scope
suggest deliverables
create a timeline
write it in a client-facing tone
avoid overpromising
If it can’t, you refine the instructions.
This is also why clients pay you instead of doing it themselves.
Most people stop after the first draft.
The money is in the refining.
The part freelancers should care about: how this makes you money
There are two ways this skill pays off.
1) It makes you faster, so you earn more per hour without working more hours
If you’re constantly rewriting the same things:
proposals
intro emails
onboarding messages
“here’s how we work” explanations
follow-ups
scope breakdowns
A bot turns those into a system.
That means you respond faster, deliver faster, and you stop bleeding time on admin.
Even if you never sell bots as a service, this alone makes you more profitable.
2) You can sell the bot itself as a deliverable
This is where it gets interesting.
Clients will pay for bots because they feel like “something they own” that keeps working after you leave.
Here are offers you can sell right now:
FAQ Bot Setup (for service businesses)
Perfect for websites, DMs, and customer support.
Charge once for setup, then monthly for updates.
Sales Reply Bot (for busy founders)
Turns enquiries into ready-to-send responses.
Bonus if it handles objections and pricing questions.
Proposal + Quote Bot (for agencies and consultants)
Speeds up closing time. Clients love this because it saves them effort and improves consistency.
Internal Team Helper (for ops-heavy businesses)
Trained on their internal processes so staff stop asking the same questions.
Content Repurposing Bot (for creators)
Turns one newsletter into LinkedIn posts, IG captions, and short scripts.
If you want to start simple, sell one bot type, get one case study, then repeat.
How to price it (without undercharging)
Here’s a clean way to think about pricing:
You’re not charging for “building a bot”.
You’re charging for:
the strategy (choosing the right use case)
the structure (turning chaos into a system)
the setup (instructions, tone, outputs)
the testing (making it reliable)
the handover (so the client can actually use it)
A basic build might include:
1 bot
1 core function (example: FAQs or proposals)
1 round of testing and refinement
a short handover doc or Loom video
Then you upsell:
more bots
more workflows
integrations
monthly updates
If the bot saves a founder 5 hours a week, you’re not selling “a tool”.
You’re selling time back. That’s always worth money.
A simple way to get your first bot client
You don’t need a fancy sales page.
You need one message and one clear offer.
Here’s the approach I use:
Identify the repeated problem
Explain the cost of it
Offer the bot as the solution
Example:
“You’re answering the same questions from clients every week. I can build you a bot that gives consistent replies in your voice and reduces your admin. Once it’s set up, your team can use it daily.”
That’s it.
No long pitch. No hype. Just a practical outcome.
If you only take one thing from this
Don’t build bots because it’s trendy.
Build them because freelancing is full of repeat work, and repeat work is where your time disappears.
Once you know how to turn a messy process into a clean tool, you stop competing with freelancers who only sell labour.
You become the person who builds systems.
And clients keep those people.
— The Profreelance Crew

Tool of the week

If you want your bot to do more than “just chat”, this is the tool. With Zapier Interfaces, you can turn a custom bot into something clients actually use, like a lead intake form, a support helper, or a mini internal portal, and connect it straight to Google Sheets, Gmail, Notion, Slack, and more.
Best use cases for freelancers:
A Lead Qualification Bot that captures info and sends it to a spreadsheet/CRM
A Client Onboarding Assistant that collects requirements + auto-creates a project folder
A FAQ Support Bot that reduces inbox chaos without hiring more staff
It’s the difference between “cool demo” and “paid deliverable.”
Interesting Reads Recommended by Profreelance (If you’re looking for more reading)
Resource Archive

PROFREELANCE (Pty) Ltd
2023/279056/07
The content in this newsletter is for informational purposes only and does not constitute financial, legal, or professional advice. Pro Freelance and Freelance Forward are not affiliated with or endorsed by the platforms or tools mentioned (unless stated otherwise), and we are not liable for any losses, damages, or issues arising from your use of them. Always do your own research before making decisions related to your freelance business.





