Tavus launches no-code PAL Maker for face-to-face AI agents
Users describe an agent to Charlie, which configures its face, voice, memory, guardrails and deployment without an engineering build.
By Ryan Merket · Published
Why it matters
PAL Maker moves Tavus beyond developer APIs by letting operators build multimodal agents themselves, creating a broader funnel for its usage-priced video infrastructure.
Tavus co-founders Hassaan Raza (@hassaanraza97) and Quinn Favret (@qfav95) launched PAL Maker on July 29th, giving nontechnical users a conversational tool for building and deploying Tavus's video-based AI agents without writing code.
https://x.com/tavus/status/2082506151728685147?s=46
The San Francisco AI lab introduced PAL Maker in a thread on X, describing it as the first no-code builder for what Tavus calls PALs, or Personal Affective Links. The agents are designed to see and hear users during live conversations, retain context, display a generated face and voice, and take actions through connected tools.
Raza, Tavus's CEO, studied computer science at the University of Texas at Austin and held roles at Apple and Google before starting Tavus. Favret, Tavus's COO, runs operations, finance, sales and people operations. The pair took Tavus through Y Combinator's Summer 2021 batch, initially selling technology for generating personalized AI videos before expanding into real-time conversational agents.
Charlie turns a description into an agent
Inside PAL Maker, users speak with Charlie, an AI guide that asks questions about the agent they want to build. Tavus says Charlie then configures the system prompt, objectives, guardrails, knowledge, tools, face, voice and memory.
A user could ask for a physical therapist that watches exercise form, for example, or create an agent that screens job candidates, runs sales demonstrations, conducts workplace role-play training or checks patients in before a medical visit. Those are examples supplied by Tavus, rather than independently reported deployments of PAL Maker.
Finished PALs can be published as Tavus-hosted pages, embedded into websites and applications, or added as widgets. Tavus also says a PAL can independently join a Google Meet, while Presentation Mode lets it navigate a slide deck and Magic Canvas allows it to display forms, charts, calendars and other interactive elements during a video conversation.
That deployment layer is central to the launch. Tavus previously required developers to assemble its models and APIs into a working application. PAL Maker packages the same components behind a guided creation flow, widening Tavus's potential customer base from engineering organizations to sales, recruiting, healthcare and training operators who want to prototype an agent themselves.
Tavus puts its model stack behind the builder
PAL Maker runs on three Tavus models. Phoenix-4 renders faces and changes their expressions during a conversation. Raven-1 processes audio and visual signals, including gestures and expressions. Sparrow-1 manages conversational timing, deciding when an agent should listen, wait or speak. Tavus also allows developers to connect their own language models while retaining its perception, timing and rendering systems.
The architecture reflects Tavus's broader bet that AI agents will need an interface richer than a text box. Tavus is trying to own that interface layer by combining agent behavior with a generated human presence, rather than competing primarily as a general-purpose language-model provider.
Tavus says more than 100,000 developers and enterprises use its products and names Amazon and Mayo Clinic among its customers. The figure combines developers and organizations in one total, so it does not show how many are paying customers or how much usage comes from large deployments.
A $64 million push beyond AI video
PAL Maker follows Tavus's $40 million Series B announced in November 2025. CRV led the round, with Scale Venture Partners, Sequoia Capital, Y Combinator, HubSpot Ventures and Flex Capital participating. Tavus reports approximately $64 million in total funding and has not published a valuation.
The launch shows where that capital is going: Tavus is turning research models and developer infrastructure into an end-user creation product. That gives Tavus a route into departments that may have a use case for an interactive agent but lack the engineers needed to build one from APIs.
Tavus's current pricing also creates a direct conversion path. Consumer PAL plans include a free tier, a $20-a-month Plus plan and a $50-a-month Max plan. Developer subscriptions start with a free plan, followed by a $59-a-month Starter tier, a $397-a-month Growth tier and custom enterprise pricing. Limits are based partly on conversational-video minutes and concurrent sessions.
PAL Maker lowers the work required to produce a demo, while Tavus still charges for the infrastructure needed to run conversations at volume. The commercial test is whether those quickly created PALs survive beyond prototypes and become recurring parts of recruiting, sales, training and patient workflows.