Comparisons
The decisions before the build.
Build or buy. Hire or engage. Which platform, and whether you need one at all. Each of these is written to answer the question properly, including where the answer costs us the work.
How these are written
Every comparison here names the case where the other option wins, and says so in the same voice it uses to sell. A page that always concludes with its author is an advert, and both buyers and answer engines can tell the difference.
We compare what we have actually run. The voice platforms are judged on production builds we shipped on each of them, not on feature tables. The consultant comparison is priced against the salary we advertise ourselves. Where we have not used something, it is not on the page.
And each one carries something you can act on without talking to us: a cost model, a quiz, a decision rule. If a comparison cannot give you that, it did not need to exist.
Every comparison
AI consultant vs in-house hire
The three real options, what twelve months of each costs once ramp and management time are counted, and the cases where the honest answer is not to pay anybody.
A twelve-month cost model you put your own numbers intoCustom AI agent vs off-the-shelf
Off-the-shelf suits linear, low-volume workflows. Custom wins the moment an agent needs live data, deep branching or compliance-grade control.
A five-question build-versus-buy quizVoice AI platforms compared
Retell, ElevenLabs, Vapi and Bland judged on what we have actually shipped on each, including where each one stops being the right call.
Written from production builds on every platform namedWhat we will not compare
Comparison pages are easy to write badly, and the bad version is profitable enough that most of the internet writes it. Three things are off the table here, permanently.
Named competitors we have not used
Comparing a firm we have never worked alongside would be an opinion dressed as a review. Where a competitor genuinely belongs in an answer, we name the category and the trade-off rather than inventing a verdict on a company.
Feature tables as the argument
A row of ticks tells you which product has more features, which is almost never the question. What decides these choices is the shape of your work, the volume, who owns it afterwards, and what happens the day something breaks.
A conclusion decided in advance
Each page here has to be able to end with the option we do not sell. Two of the three currently do, in at least some circumstances, and if that ever stops being true this cluster has become marketing and should be read as such.
If you want the version of this conversation about your own business rather than in general, the free AI Opportunity Audit maps what you already run, and the glossary defines the vocabulary these pages use.
Or skip the reading
The free AI Opportunity Audit maps what your business already runs and what is worth building first. Two minutes, no call required, and the map is yours whichever way you decide.