Skip to content

Voice AI

Migrating Off Vapi or Retell: When It Is Worth It, and How to Not Break Things

The four reasons that justify a migration, the work managed platforms were quietly doing for you, and a cutover sequence that does not let customers discover the new stack is worse.

Vapi and Retell are good products and starting on one of them is usually the right call. But there is a point where the platform becomes the ceiling, and the question turns into whether to rebuild. Here is how I decide, and how I run it when the answer is yes.

The only four reasons that justify it

  • Cost. Your per-minute spend at real volume exceeds the cost of building and maintaining your own. Do the arithmetic properly, including maintenance.
  • Latency. You have measured, you need materially faster, and the platform gives you no layer to optimise.
  • Capability. A model, voice, language or integration you need and the platform will not add.
  • Data. A residency, retention or vendor requirement a shared platform cannot satisfy.
  • "We want more control" is not a reason. It is the mechanism. If you cannot name which of the four applies, stay where you are.

    What you are actually taking on

    The prompts, conversation design and tool definitions port over in days. What does not port is everything the platform was doing quietly: worker autoscaling, retry and failure handling, turn-detection tuning, call recording and storage, observability, and telephony configuration. Budget several weeks for that alone. A migration that skips it produces an agent that is cheaper per minute and worse to talk to, which is the worst possible outcome.

    The sequence that works

  • Measure the current agent first: per-turn latency, true cost per minute, and where calls drop off. Without this baseline you cannot tell whether the rebuild is better.
  • Rebuild on LiveKit or Pipecat, carrying prompts and tools across unchanged so the comparison is honest.
  • Run both against the same set of your own call recordings, and compare stage by stage rather than on overall impression.
  • Cut over by percentage of live traffic (5, then 20, then 50), with the old platform still running behind it.
  • Keep the rollback path working until the new stack has held a full billing cycle.
  • How long and how much

    Four to eight weeks and $10,000 to $28,000 for a typical single-agent migration including the comparison and gradual cutover. If your agent is simple and volume is low, I will usually tell you not to bother. The full scope is on the migration service page, or send me your current numbers.

    Direct Project Enquiry

    Tell me what you are building

    Share what you are building and where it is stuck. If a managed platform already solves it, I will say so. I review every enquiry personally and reply within about four hours.

    1. 1Send your project goal and the services you think you need.
    2. 2I review the context and whether I am the right fit.
    3. 3You get a direct reply with a practical next step.

    International projects welcome

    I work in English by email, video call or through Upwork, from GMT+7 with overlap hours for US and EU teams.

    Read the Privacy Policy, Terms of Service and Cookie Policy.

    Client project enquiries only

    This form is for businesses and teams looking to hire. Sales pitches, recruitment messages, guest posts and link-building outreach are not reviewed.

    Select Service *

    In a hurry? Message me on WhatsApp at nguyentienm@gmail.com

    Message me