The setup has been really easy to learn and the customization features of the agents have been really great. I love how easy it is to control the cadence, interrupt tolerances, tone, etc. There are also many free voices to use that are all high quality. The dictation also works really well for post-call transcripts and it can easily hook into other platforms like Make. Review collected by and hosted on G2.com.

I honestly haven't experienced any downsides of note yet Review collected by and hosted on G2.com.

The API-first approach makes integration incredibly fast — we had a voice agent live in production within days, not weeks. The real-time WebSocket architecture delivers noticeably low latency, which is critical for natural conversations. The ability to plug in any LLM (Claude, GPT) and any voice provider (ElevenLabs, Cartesia) gives us full control over the agent's personality without vendor lock-in. Custom tool calling during live calls is a game-changer for our use case — the agent can query databases, create tickets, and trigger workflows mid-conversation seamlessly. Review collected by and hosted on G2.com.

The documentation for post-call webhooks and custom analysis fields could be more detailed — we had to experiment to get post_call_analysis_data working with custom schemas. Debugging failed calls can be tricky since logs don't always surface the exact point of failure in the conversation flow. Also, voice cloning options are limited to specific providers per account, so testing different voices across environments requires workarounds. Review collected by and hosted on G2.com.

The biggest advantage of Retell AI is how fast it lets you go from idea to a production-ready AI calling agent. The voice quality, latency, call handling, and developer experience are all strong. It’s especially impressive for outbound and inbound automation where reliability and natural conversation flow matter. The API structure and integrations also make it much easier to build real business workflows instead of just demos. Review collected by and hosted on G2.com.

The platform can still feel limited when trying to build highly custom conversation logic or deeply complex workflows at scale. Debugging edge-case call behavior can sometimes take longer than expected, and pricing can become expensive for high-volume usage. More advanced analytics, monitoring, and native workflow tooling would also make the platform stronger for larger production deployments. Review collected by and hosted on G2.com.

I appreciate how Retell AI makes voice cloning and text-to-speech workflows accessible while still sounding natural and expressive. The platform’s interface is intuitive, and the ability to generate high-quality voice output quickly — without needing deep audio engineering knowledge or complicated setups — is very convenient. I also like how real-time iteration and previewing help speed up creative workflows without sacrificing sound quality. Review collected by and hosted on G2.com.

While Retell AI produces generally clear and intelligible output, I’ve noticed that the phonation layer can sometimes feel slightly uniform in longer passages. The prosody tends to flatten over extended dialogue, and there can be subtle inconsistencies in breath modeling and coarticulation between phrases. Greater control over micro-dynamics, pacing variation, and emotional contour would help the voices feel more human and less templated, especially in high-context conversational scenarios. Review collected by and hosted on G2.com.

I really like how, whether it's navigating the interface or customizing a voice agent, nothing breaks in the process. I've had experience with previous services where the website would fail to work; buttons would break, agents wouldn't answer. Retell works as you expect it to 100% of the time. Review collected by and hosted on G2.com.

I think the lack of a proper account settings/profile panel makes things a bit more painful, as there's no obvious way to setup 2FA, review signed-in sessions, etc. I would also expect a team/organization view for managing multiple workspaces. Review collected by and hosted on G2.com.

Retell AI makes it easy to build and test AI voice agents quickly. I like the clear dashboard, smooth call handling, good voice quality, and fast response time. It is very useful for creating AI receptionists and automating customer calls. Review collected by and hosted on G2.com.

There is not much to dislike, but some setup steps can feel confusing for beginners, especially connecting phone numbers, webhooks, and integrations. More simple guides and examples would make it easier for new users. Review collected by and hosted on G2.com.

Visual Interface for conversacional flows Review collected by and hosted on G2.com.

Right now, I don’t think I dislike anything. The only thing that I could probably add it’s some sort of wizard that actually asks you what your goal is and it lets and it tells you which agent should you use a prompt agent or a conversational flow?. Personally, I started with a prompt and had to migrate to a conversation flow Review collected by and hosted on G2.com.

I have tried tens of voice agent builders, and I think Retell AI is the only builder that offers true production ready voice agents. This platform has the least number of bugs as compared to others. It also offers workflows, something that others does not have. Everything is in this platform, be it observability, mobile numbers, analytics, quality assurance and more. Review collected by and hosted on G2.com.

I think the pricing is steep. The fast tiers can add up the costs very fast. Also, like any no-code builder, the choices are just too limited, like in terms of TTS, STT and LLM. Review collected by and hosted on G2.com.

What I like best about Retell AI is how easy it makes building a real human-like AI voice agent. The call quality sounds natural, the response time is fast, and it integrates smoothly with tools like calendars and workflows. I especially like that I can customize prompts deeply, connect it with automation tools like n8n, and create complete AI receptionists for businesses like salons. It saves time, reduces missed calls, and works 24/7 without breaks. For anyone building AI voice solutions, Retell AI makes the process simple and powerful. Review collected by and hosted on G2.com.

One thing I dislike about Retell AI is that it sometimes misunderstands certain words, especially business-specific service names, which can affect booking accuracy. It also requires careful prompt tuning to avoid issues like duplicate bookings or incorrect time handling. Additionally, pricing can feel slightly high when testing or scaling voice agents frequently. With improvements in speech recognition accuracy and more built-in safeguards for booking logic, it would be even better. Review collected by and hosted on G2.com.

Retell AI makes the programming of AI receptionists a breeze. They are incredibly realistic and mostly smooth, with the occasional delay/hiccup. Review collected by and hosted on G2.com.

As a telecommunications company, we have our own SIP trunks that we would like to set up with Retell AI, especially for use of SMS. However, our SIP trunks are currently not compatible for integration. This is incredibly inconvenient, as we have needed to go through the process of buying Twilio sip trunks, and have experienced repeated denials when submitting the SMS campaign. The rejection reason features a vague response, and when I requested to speak to an agent for clarification of what needs to be adjusted, I could not manage to get past their AI support agent to speak to a real person. I'm not impressed with their customer assistance aspect. Review collected by and hosted on G2.com.