Growth agencies are sitting on a problem they did not expect to have: their clients want AI features, and they do not have the engineers to build them. Hiring an AI team is expensive and slow. Referring the work out risks the client relationship. White-labelling is the middle path -- and it is one we have built a formal programme around.

This post is unusually transparent. We are going to show you exactly how the programme works -- the tech, the contracts, and the margins -- so you can decide if it makes sense for your agency.

How the Programme Works

We act as the engineering arm. You own the client relationship. Your brand is on everything the client sees. Our name appears nowhere in client-facing materials unless you choose to disclose it.

The typical engagement flow:

  1. You bring us a scoped client brief.
  2. We produce a technical spec and fixed-price quote within 48 hours.
  3. You mark up our rate and present your price to the client.
  4. We build. You review. Client receives deliverables under your brand.
  5. Ongoing support is managed through your agency -- we sit behind your helpdesk.

The Tech Stack

Our white-label deliverables are production-grade, not prototypes. The standard stack we ship under partner programmes:

AI and LLM Layer

  • Models: Claude (Anthropic) for conversational and document tasks; GPT-4o for vision and structured extraction; Gemini 1.5 Pro for large-context document processing
  • Orchestration: LangChain LCEL for multi-step chains; raw API calls for latency-sensitive single-pass use cases
  • Vector stores: Supabase pgvector (under 500k vectors), Pinecone (scale)

Application Layer

  • Frontend: Next.js 15 (App Router) -- embeddable as a widget or standalone app
  • Backend: Supabase (auth, database, storage, edge functions) or PlanetScale + Railway
  • Integrations: Zapier, Make, HubSpot, Salesforce, Slack -- whatever the client's stack requires

Delivery Format

Deliverables are provided as: private GitHub repository (transferred to client or agency), Vercel/Railway deployments on client infrastructure, and documented API endpoints if the feature is headless.

Contracts and IP

This is what agencies ask about first. The answers are simple:

  • IP ownership: The client owns all code and deliverables outright upon final payment. We retain no licence or claim.
  • NDA: We sign a mutual NDA before any brief is shared. Your client's name and project details are confidential.
  • Non-solicitation: We will never contact your client directly. If a client approaches us independently (which happens rarely), we refer them back to you.
  • Liability: Our agreement is with your agency, not the end client. You manage the client contract on your standard terms.

The Margin Structure

We charge agencies a wholesale rate. The market rate for the same deliverable to an end client is typically 40-80% higher than our partner rate. The margin is yours to set.

Example (indicative, not contractual):

  • AI chatbot integration with RAG: our rate $8,000 -- market rate $12,000-15,000
  • Workflow automation (3 automations): our rate $5,000 -- market rate $8,000-10,000
  • SaaS MVP: our rate from $18,000 -- market rate from $28,000

Most agency partners mark up 30-50%. The client gets a competitive rate. You earn margin without engineering overhead.

Who the Programme Is For

The programme works best for:

  • Growth and digital marketing agencies whose clients are asking for AI automation and internal tooling
  • Strategy consultancies that identify the opportunity but need an engineering partner to execute
  • Boutique software agencies with a strong client base but no AI/LLM specialist on staff

It is not suited for agencies that want to build an in-house AI capability over time -- for that, we offer a different structure. But for agencies that want to deliver AI work profitably without the hiring risk, this is a direct path.