Dograh

Best White-Label Voice AI Platforms for Agencies

Best White-Label Voice AI Platforms for Agencies
ComparisonNovember 7, 2026·9 min read

Best White-Label Voice AI Platforms for Agencies

Vemu Sandeep
Vemu Sandeep·GTM Engineer, Dograh AI

GTM engineer at Dograh working across voice AI and enterprise AI. IIT (BHU) and IIM Visakhapatnam alumnus.

Judge a white-label voice AI platform on client separation, your own branding, the margin you keep per minute and model choice. Dograh, the open-source AI voice agent platform we build, lets an agency self-host it under its own name and logo or run one Dograh Cloud team per client. Autocalls and Trillet sell branded portals with minutes included.

Key Takeaways

  • Judge platforms on client separation, branding, margin per minute and model choice.
  • Finished white-label products bundle models and minutes, so your margin sits on their rate.
  • Retell and Vapi suit agencies that build their own client portal.

An agency resells a platform under its own name, so the platform's limits become the agency's promises. We judge a short list on four things agencies need, compare them in one table, cover each platform from its own pricing page, starting with Dograh, then work out one margin example.

Four things an agency needs from a platform

An agency resells calls to clients, so it needs more from a platform than a good agent. The first test is client separation. Each client's agents, numbers, call records and keys should sit apart, and a client login should show only that client. A platform that counts unlimited sub-accounts still needs checking on what each one actually isolates.

The second is your own branding. Callers hear the client's business, but the client logs in to a dashboard and receives invoices, and both should carry your name. The third is margin per minute, which is what you keep from every minute you resell after the platform, models and phone line are paid. The fourth is model choice, because the language model and voice decide both call quality and most of the cost per minute.

Each platform below is judged on those four only. For the questions to put to any platform before your first client signs, see our checklist for choosing a white-label voice AI platform.

The short list in one table

The table compares each platform on the four tests, with prices from each vendor's own page. Dograh appears twice, once for each way to run it.

PlatformClient separationYour brandingWhat you pay per minuteModel choice
Dograh, self-hostedAgents, phone setup and keys per client on your own copyYour name, logo and domain, licence notice keptNo platform fee, plus phone, models and serversAny provider by key, or open models you host
Dograh CloudOne team per client, switched from the sidebarCallers hear only your agent. The dashboard is Dograh's1¢-a-minute platform fee, plus phone and modelsAny provider by key, or Dograh's models
Autocalls White LabelUnlimited client sub-accountsCustom domain, rebilling through your Stripe$419 a month with 3,500 minutes, then $0.09Models included, no own keys
TrilletUnlimited client workspaces on AgencyYour brand and domainFrom $99 a month with 1,000 or 3,000 minutes, then $0.12 plus phoneNot stated on its pricing page
Stammer Pro White LabelClient Sub-Account WalletsYour domain, logo and colours$99 a month, voice $0.11 to $0.17 by modelOwn LLM keys apply to chat only
Assistable3, 10 or unlimited sub-accountsWhite labelling on every plan$225 to $975 a month, then usage by linePer-model rates, own SIP trunk
RetellOne workspace per clientDashboard not rebranded$0.07 to $0.31 a minuteIts model list
SynthflowA workspace per clientYour brand and domainAnnual enterprise agreementThrough the agreement
VapiAn organization per client, $20 a month eachNo white-label offer on its pricing page$0.05 hosting plus models at costModels passed through at cost

Dograh, self-hosted or one Cloud team per client

Dograh gives an agency two routes, and the brand question decides which one fits. Self-hosted Dograh is released under the BSD 2-Clause licence, so an agency can run its own copy on its own domain under its own name and logo, as long as the licence notice stays. Our logos stay ours, so your copy carries your logo in their place, and the rebrand is a change you make once in your copy.

Margin is cleanest on that route. Self-hosted Dograh has no per-minute platform fee, but servers, model keys and phone minutes still cost money. Dograh Cloud charges a 1¢-a-minute platform fee. Each client's model keys sit in your own database, so protect database access and backups as you would any client credential.

Model choice is wide on both routes. You can add several connections per provider and named model configurations, so one client runs on its own OpenAI key and another on a different voice provider, or on open models you host. Pre-recorded clips in the client's own voice, with a generated fallback, keep the agent sounding like the business. Full call traces and built-in QA let you show a client exactly how its calls went.

One client's calls are each traced and stamped by built-in QA. The agency then sits down with that client over the record of its own calls, while other clients sit on a shelf, kept apart, one team per client.
Each client's agents, numbers, call records and keys should sit apart, and a client login should show only that client. Everyone in a team has full access to the team, so invite a client only if they should see every setting and call.

Dograh Cloud is the faster start. Each client sits in its own team, which you switch between from the sidebar, and callers hear only the agent you built. Everyone in a team has full access to it, so invite a client only if they should see every setting and call. For billing, the Paygent integration lets you set rate cards per client. Our post on the white-label AI receptionist covers the receptionist version of this setup.

Dograh

Open Source Alternative to Vapi / Retell

Self-hosted voice agent platform — no per-minute fees

dograh-hq/dograh

Star on GitHub

Autocalls and Trillet, portals with minutes included

These two sell a finished white-label product with a bundle of minutes. Autocalls' pricing lists the Agency plan at $249 a month with 1,700 minutes, where clients sit inside their own Autocalls account, and Agency White Label at $419 a month with 3,500 minutes, unlimited client sub-accounts, a custom domain and rebilling through your own Stripe. Extra minutes are $0.09 on both. Models are included, and you never connect your own key.

That makes Autocalls simple to resell and narrow on model choice. Your margin is your price minus its rate, and you cannot move a client to a different model provider to cut cost or match a language.

Trillet's pricing lists white-label Studio and Agency plans from $99 a month, with 1,000 and 3,000 minutes included, then $0.12 a minute of AI usage covering platform, speech-to-text, language model and voice. Telephony is billed separately. Agency carries your brand and domain so clients never see Trillet, and it allows unlimited client workspaces. Its pricing page says nothing about choosing models.

Stammer and Assistable, sub-accounts with rebilling

These two focus on billing clients through the platform. Stammer's pricing lists Pro White Label at $99 a month, removing Stammer's branding for your domain, logo and colours. Clients get a Sub-Account Wallet that they preload, and that money goes to you. Voice is billed per minute by model, from $0.11 on GPT-4.1-nano to $0.17 on Claude 3.7 Sonnet. Its own-key option for language models applies to chat messaging only, so voice minutes always carry Stammer's model rate.

Assistable's pricing charges by client count, with Startup at $225 a month for up to 3 sub-accounts, Growth at $450 for 10 and Agency at $975 for unlimited, each with white labelling and usage rebilling. Usage is priced line by line, with $0.050 a minute for orchestration, separate voice and model lines, and $0.015 a minute for its own telephony, or nothing if you bring your own SIP trunk. Its own example comes to $0.090 a minute before add-ons on an imported number.

Assistable's line-by-line pricing gives you more control over margin than a bundled rate, and the sub-account count sets the plan.

Retell, Synthflow and Vapi

These three are platforms agencies build on, and none sells a rebranded dashboard on a self-serve plan. Retell offers a workspace per client, each billed for its own usage at $0.07 to $0.31 a minute, with 20 concurrent calls per pay-as-you-go workspace. Retell's own page says the dashboard itself is not rebranded, client members are read-only, and workspaces are not created through the API.

Synthflow sells white labelling only inside an annual enterprise agreement, with no per-minute rate on its pricing page. Under that agreement, each client gets its own workspace under your brand and domain.

Vapi charges $0.05 a minute for hosting, with models passed through at cost, and $20 a month for each extra organization, which agencies use to keep clients apart. Its pricing page carries no white-label offer, so the client-facing portal is yours to build. All three suit an agency with developers that wants to build its own client portal on top.

Margin per minute, as an example

Here is one worked example. An agency resells a receptionist agent to one client at $0.25 a minute, an assumed price, and the client uses 3,000 minutes a month. Each row uses only that platform's published rates.

PlatformCost of 3,000 minutesWhat the agency keeps of $750
Dograh Cloud, with Twilio, Deepgram and gpt-4.1-mini3,000 x $0.0475 = $142.50, plus $1.15 for the numberAbout $606
Autocalls White Label$419 plan, 3,500 minutes included$331, before any phone charges
Assistable Startup, its $0.090 example rate$225 plan + 3,000 x $0.090 = $495$255, before your own carrier
Stammer Pro White Label, GPT-4.1-mini voice$99 plan + 3,000 x $0.12 = $459$291, before phone charges

The Dograh Cloud cost per minute is the example build-up from our post on RingCentral AI Receptionist pricing, made of the $0.01 platform fee, Twilio minutes and streaming, Deepgram speech and voice, and gpt-4.1-mini. Self-hosted, the $0.01 line goes and your servers take its place, spread across every client you run.

The bundled plans look different as clients grow. Autocalls' and Assistable's plan fees cover many clients at once, so the per-client cost falls as you add clients, while Dograh's lines stay per minute. Phone minutes are the line you can buy in bulk, and our post on wholesale SIP trunks for voice AI agencies covers that. For the whole agency model, from stack to pricing clients, see our guide on how to start an AI voice agent agency.

Join the Dograh Community

Dograh is an OSS alternative to Vapi. Join our Slack community for queries, releases, best practices & community interactions.

Glossary

Sub-account
A separate space for one client inside an agency's platform account. Check what it isolates, such as agents, numbers, call records and keys, since a count of sub-accounts says nothing about that.
Rebilling
Charging clients for their usage through the platform, often at your own price through your own Stripe or a client wallet. Autocalls, Stammer and Assistable offer it, and Dograh supports per-client rate cards through Paygent.
Margin per minute
What an agency keeps from each resold minute after the platform, model and phone costs are paid. Bundled rates fix most of the cost, while line-by-line pricing lets you lower it by changing providers.
Licence notice
The copyright notice and disclaimer an open-source licence asks you to keep. Under BSD 2-Clause, an agency can rebrand its own copy of Dograh as long as that notice stays with the code.

Frequently Asked Questions

Get started with Dograh

Build, deploy, and scale AI agents with Dograh. Join the community of developers building the future.