A white label AI agent is built by configuring an existing agent platform with your own name, icon, URL, color, welcome header, default skills, and behavioural prompt, so the product presents as yours rather than the vendor's. The agent connects to the operator's working stack, including email, calendar, booking, payment, and messaging tools, and executes scheduled routines rather than only generating text. Reselling the branded agent to clients typically requires a specific plan tier from the underlying platform.

What Is a White Label AI Agent?

A white label AI agent is an AI system produced by one company and presented under another company's brand, with the reseller controlling the name, visual identity, default capabilities, and behavioural instructions while the underlying platform supplies the model, integrations, and execution layer. It is distinguished from a general assistant by acting on connected systems rather than only returning text.

Dubb Agent supports this through a branding configuration available on upgraded accounts, covering title, icon, URL, brand color, header, default skills, and a brand prompt. A workspace switch applies that branding, and reselling the co-branded experience to clients requires the scale plan.

Integration Is What Decides the Tool

There are a great many AI tools now, and choosing between them on model quality alone misses what actually matters for a sales or marketing operator. The question is not which one writes the best paragraph. It is which one can reach your stack.

Your stack is the set of systems your business already runs on. Email, whether Gmail or Outlook. Your calendar. Booking links such as Calendly. Payment and accounting on the back end, whether Stripe, PayPal, or QuickBooks. Email automation. SMS. The list keeps going, and it is different for every business.

An AI tool that cannot touch any of it is a writing assistant. A useful one, but you remain the integration layer, copying its output into the systems where work actually happens. The best AI tool you can build is the one that integrates with your whole stack, because that is the difference between advice and execution.

From Answering to Doing

This is the shift worth naming clearly, because a lot of people are still evaluating AI as though it were 2023.

The first generation produced copy, strategy, and text. You asked, it answered, and then you did the work. The current generation does the doing. It is the difference between a system that drafts your follow-up email and a system that sends it, logs it in the CRM, and books the reply into your calendar.

The industry calls this agentic AI. An agent is a system that takes actions on connected tools, not one that produces text for you to act on. Once you have understood that distinction, the integration question stops sounding technical and starts sounding like the whole point. Actions require somewhere to act.

The practical test I would apply to any AI tool: can it finish a task without me? If every output still lands in your lap for manual handling, you have bought a faster typewriter.

Routines Are the Team You Assemble

Routines are where the agentic idea becomes concrete. A routine is a standing job the agent performs without being asked, and assembling several is effectively hiring a team.

• Generate leads on a schedule rather than when you remember to.
• Send email follow-ups so the sequence continues while you are in meetings.
• Update the CRM, which is the task every seller postpones and every manager asks about.
• Update your calendar and tell you when meetings are coming.
• Scan your email for missed opportunities, the buried reply from three weeks ago that nobody answered.
• Prioritise deals in your CRM, so the list you open in the morning is ordered by something other than date created.

The compounding effect is the point. Any one of these saves minutes. Running together on a schedule, they change what your morning looks like, because the work happened overnight rather than waiting for you to start it.

What Custom Branding Actually Covers

Once the agent is doing useful work, you can put your own name on it. The configuration is available on upgraded accounts, reached by clicking upgrade in the left navigation or at the top, or through billing.

There is also a custom branding playbook in the skills section. Turn it on, tell the agent you want custom branding, and it walks you through the whole configuration as a guided questionnaire rather than leaving you to fill in fields cold.

Getting the Icon Right

The icon requirement trips up more people than anything else, and the fix is a design principle rather than a technical one.

The spec is 512 by 512 pixels, PNG, square, ideally with a transparent background. If you do not have anything in that format, ask the agent to generate one. A prompt as simple as asking for a 512 by 512 PNG icon with a transparent background featuring a particular letter, object, or shape will produce something usable. It can also process an existing logo into an icon.

The design rule matters more than the file spec. Think about how small this mark will eventually appear. Look at a branded pen: the logo is tiny and the icon smaller still. If your logo contains small text, that text will shrink into an illegible smudge. One bold shape or letter survives; a full lockup with a tagline does not.

Workshop it before you settle. This is the element that appears everywhere and is hardest to change later once clients have learned to recognise it.

Writing the Brand Prompt

The brand prompt is where your agent stops being a generic product with your logo on it and starts behaving like yours. You have up to 5,000 characters, which is more than it sounds.

Cover the things that make your approach distinctive:

• Role. What this agent is for and who it serves.
• Core belief. The conviction underneath your methodology.
• Voice. How you sound. Direct, warm, formal, funny.
• Principles. The rules you would give a new hire.
• What you do and how you work. The actual methodology.
• Guardrails. What it should never say or do.

If drafting that from scratch feels daunting, use another AI to write it. A free ChatGPT or Claude account is perfectly capable of interviewing you about your business and producing a brand prompt you then paste in. Using one AI to configure another is not cheating; it is the fastest route to a good first draft.

Personality is a legitimate instruction, not a frivolity. If you want it to crack jokes, tell it to crack jokes. If your brand is dry and sardonic, say so. Anyone who has seen a certain superhero adjust his assistant's humour setting has the right mental model: the dial exists, and you decide where it sits.

The guardrails deserve real attention. A branded agent speaks as you to your clients, so the things it must never claim, promise, or advise on are as much a part of your brand as the voice is.

A Worked Example

The clearest way to see this is a fully built one. Jeffrey Gitomer, the sales author behind The Little Red Book of Selling and The Sales Bible, has a branded agent built on this configuration.

Every element traces back to something recognisable about him. The name follows his own. The logo reflects the King of Sales identity his audience already associates with him. The color is the red that runs through his brand, down to the mechanic-style shirt he wears on camera. The brand prompt captures how he actually communicates: direct, specific, pragmatic, East Coast, with room for humour and a little sarcasm.

Underneath the branding, it is connected to his stack, including email, calendar, and CRM. Where a needed connection is not among the standard integrations, custom integrations and custom skills fill the gap, and custom routines run the work specific to his business: winning clients, booking stages, selling books.

What makes it a useful example is that none of the branding is decorative. The color, the voice, and the skills are all doing recognition work, so a person who knows his books opens the agent and finds it consistent with everything else they know about him.

Turning It Into a Revenue Line

The switch labelled apply my branding in my workspace is what extends your brand to the people you serve. On the scale plan, currently three hundred dollars per month, you can offer that co-branded agent to your own clients.

For a consultancy, coaching practice, or agency, this is a genuine addition to what you sell. Your clients already buy your judgment. An agent carrying your methodology, your voice, and your guardrails is that judgment operating between your sessions rather than only during them.

Now the honest part, because this deserves plain speaking rather than enthusiasm.

This is starting a business, not buying a passive income stream. The comparison sometimes made to a franchise is fair in one respect: you are licensing a system rather than building the technology. It is also fair in a less comfortable respect, which is that a franchise fails without an operator who does the work. Your outcomes will depend on your existing audience, your offer, your ability to sell and support software, and how much effort you put in. Some people build a meaningful revenue line. Others do not.

Anyone quoting you a monthly earnings figure for a business you have not started yet is guessing. We are not going to do that. What we can tell you is what the platform fee is, what you get for it, and that the plan tier exists partly as a filter, because a reselling model works considerably better with committed operators than with casual sign-ups.

Treat the three hundred dollars as the cost of your first month of a new business line and ask the questions you would ask of any business: who is my buyer, what do I charge, how do I support them, and how many do I need. If those answers come easily, the model is worth pursuing. If they do not, get them straight before you subscribe. Confirm current pricing and plan inclusions before committing, since both change.

Common Mistakes and Troubleshooting

Choosing an AI tool on output quality alone. If it cannot reach your email, calendar, and CRM, you are still doing the work. Evaluate integrations first.

Uploading a logo with small text as the icon. It will render illegibly. Use one bold shape or letter, and generate a proper icon if you do not have one.

Leaving the brand prompt thin. A logo change with a default personality is a skin, not a product. The prompt is where the differentiation lives.

Skipping guardrails. The agent speaks as you. Decide in advance what it must never claim or advise on.

Branding before the routines work. Get the agent doing something genuinely useful before you put your name on it. The name amplifies whatever is underneath.

Buying the reseller tier before you have a buyer. Know who you are selling this to first. The plan does not come with an audience.

Non-square or wrong-format icons. The requirement is 512 by 512 PNG and square. Anything else will not fit properly.

Best Practices Worth Adopting

Connect the stack before you configure anything else. A white label AI agent is only worth branding once the routines underneath it are doing real work.

Start with two or three routines rather than all of them. Get those genuinely reliable, then add more.

Use the branding playbook skill rather than filling in fields blind. A guided questionnaire produces a better configuration than a blank form does.

Write the brand prompt as though onboarding a new employee. Role, beliefs, voice, method, and limits are exactly what you would tell a person on day one.

Test the agent as a client would experience it before selling access. What I see most often is operators who configure carefully and never open their own product from the outside.

White Label Platforms Compared

Reselling software under your own brand is an established model, and the platforms differ in what the branded product actually does for the end client.

The distinction worth weighing is whether you are reselling tooling your clients operate themselves, or an agent that performs work on their behalf. Both are legitimate businesses. They attract different buyers and require different support commitments, so pick according to what you can actually service. Verify current pricing and terms with each platform directly.

Proof: Why This Actually Works

The branding pipeline described here was built because demand outran the manual process. Configurations were being done one at a time on live calls for individual customers, and the volume of requests made that unsustainable, so it was rebuilt as self-service. A feature that exists because a manual version could not keep up is a reasonable signal of real demand.

Two patterns show up across the operators who configure these. The first concerns sequencing: those who connect their stack and get routines running before touching the branding tend to end up with something they use daily, while those who start with the logo often end up with a well-branded agent doing very little. The branding amplifies whatever is underneath it, including nothing.

The second concerns the brand prompt. The configurations that feel genuinely distinct are the ones with a thoroughly written prompt covering voice, principles, and guardrails. Where the prompt is thin, the result reads as a familiar product wearing a new logo, which clients notice faster than operators expect.

Methodology note: these are directional observations drawn from the branded configurations we have helped build and from aggregated, anonymized patterns across Dubb users, not a controlled study. No figures are attached, and no representation is made about the revenue any individual operator will earn from reselling.

The unglamorous conclusion is that the branding is the easy half. The work is deciding what your agent should actually do, and writing down clearly enough how you think that a machine can act on it.

Frequently Asked Questions

What is a white label AI agent?

It is an AI system built by one company and sold under another company's brand. The reseller controls the name, icon, URL, color, welcome header, default capabilities, and behavioural instructions, while the underlying platform provides the model, the integrations, and the execution layer.

What separates it from a general assistant is that it acts on connected systems rather than only producing text. It can send the email, update the CRM, and book the meeting instead of drafting something for you to handle.

How do I put my own branding on an AI agent?

Go to ai.dubb.com and upgrade your account, using the upgrade button in the left navigation or at the top, or through billing. The branding options then become available, covering title, icon, URL, brand color, header, default skills, and a brand prompt of up to 5,000 characters.

There is also a custom branding playbook in the skills section. Turn it on and tell the agent you want custom branding, and it will guide you through the configuration as a questionnaire rather than leaving you to fill in fields cold.

What size does the AI agent icon need to be?

512 by 512 pixels, PNG format, square, and ideally with a transparent background. If you do not have a file in that format, ask the agent to generate one, or give it your existing logo to process into an icon.

Avoid logos containing small text, because that text becomes illegible at the size the icon is actually displayed. One bold letter, shape, or object works far better than a full lockup with a tagline.

What are AI routines?

Routines are standing jobs the agent runs without being asked, which together function like a team you have assembled. Common ones include generating leads, sending email follow-ups, updating the CRM, updating your calendar, notifying you about meetings, scanning your email for missed opportunities, and prioritising deals in your pipeline.

The value is cumulative rather than individual. Any single routine saves a few minutes, but several running on a schedule change what your morning looks like, because the work happened before you sat down.

Can I sell an AI agent under my own brand?

Yes, on the scale plan, currently three hundred dollars per month, which enables the switch that applies your branding for the people you serve. It is a natural addition for a consulting, coaching, or agency business, since your clients already buy your judgment and the agent carries your methodology between sessions.

Treat it as starting a business rather than as passive income. Your results will depend on your audience, your offer, your pricing, and your ability to support software clients. Confirm current pricing and plan inclusions before subscribing, as both change.

What should go in the brand prompt?

Your role and who you serve, your core belief, your voice, your principles, what you do and how you work, and your guardrails. Think of it as onboarding a new employee: everything you would tell someone on their first day belongs there.

If writing it from scratch is difficult, use a free ChatGPT or Claude account to interview you about your business and draft it, then paste the result in. Give the guardrails real attention, since the agent speaks as you to your clients.

Becoming the Operator of Your Own AI

Building a white label AI agent is a sequence, and the order matters more than any single step. Connect your stack first, because integrations are what let an agent act rather than advise. Build two or three routines and get them genuinely reliable. Then brand it: title, icon at 512 by 512, URL, color, header, default skills, and a brand prompt that captures how you actually think.

Everything in that sequence is available inside Dubb Agent on an upgraded account, with the branding playbook skill guiding the configuration. The scale plan adds the switch that lets you extend the branded experience to your own clients if you decide to build that business.

Start with the part that is free to get wrong. Turn on the branding playbook, answer its questions, and workshop your icon until it reads clearly at the size of a logo on a pen. If you get stuck, the daily sessions run at 10 a.m. Pacific and someone will work through it with you.