Dehurdle

Dehurdle is an AI voice simulation platform for rehearsing conversations where the stakes are high enough that you'd rather not learn on the job. Sales pitches, price negotiations, difficult client calls, and leadership conversations are the core scenarios. You talk to an AI persona in real time, it pushes back the way a real counterpart would, and you get scored on how you handled it. It's voice first rather than text first, and the promise is real-time feedback on delivery, pacing, eye contact, and composure rather than a transcript to read afterwards.

The problem it targets is that practice for these conversations is scarce and awkward. Role-plays with a manager happen a few times a year, feel staged, and produce feedback that's mostly opinion. Meanwhile the first real version of the conversation is often with the customer who matters most, so people walk into their hardest calls with the least practice, and the coaching they do get arrives too late to change the outcome. Dehurdle turns the rehearsal into something you can do alone, on demand, and repeat until the scorecard moves.

The workflow has three phases. Prepare means picking or building the persona and scenario, and the library runs to 200 personas spanning accents from more than 125 countries, each with editable traits like pacing and price sensitivity. You can upload your own material, whether that's guidelines, a pitch deck, a call recording, or a playbook with question and answer sequences, pricing matrices, and competitor comparisons, and it's parsed and embedded into the scenario within about half a minute, after which the simulation engine reads those parameters to configure the persona. Practice is the live session, where you can present slides while the AI listens and responds, and where on-device camera analytics track composure and eye contact. Improve is the debrief, an objective scorecard covering delivery, pacing, and pauses, with suggested phrasing for the moments that went sideways and a private debrief with its Lumi AI coach on paid plans. Scorecards accumulate into a history on those plans, so you can see whether your pacing has actually changed over a month of sessions.

It's built for sales and revenue teams first, then consultants, managers, and HR or enablement staff who run training programs. The site names Deloitte, PwC, AB InBev, and Suzuki among its customers, and the Team and Enterprise tiers are shaped for exactly that kind of buyer, with shared minute pools, cohort analytics, role-based permissions, SSO and directory sync, and SOC 2 aligned data isolation. The shared minute ledger is the practical detail for managers, since a pool across a cohort is easier to budget than individual subscriptions, and cohort analytics show who's improving. Individuals can use it too, and the free tier exists for that.

What makes it stand out is the voice-first realism and the privacy posture around it. Text-based role-play tools miss the whole point of a tough conversation, which is what happens to your voice under pressure. Dehurdle listens to pace and pauses, watches composure through the camera without shipping the video anywhere, and holds a zero audio and video retention policy. The regional accent library is a practical detail for global sales teams who deal with counterparts who don't sound like the training videos, the personas aren't fixed, since accent, pacing, and price sensitivity can all be edited, and Enterprise customers can commission custom regional accent personas to match a territory.

The limits are the usual ones for a simulator. An AI persona is a good sparring partner, not a substitute for a live customer, and the scorecard measures delivery rather than whether the deal would actually have closed. Minutes are metered on every plan, so heavy practice gets expensive at the individual tier. The camera analytics need a webcam and a willing participant, and the most interesting extras, like custom voice clones and the REST API, sit in the top contract tier. The site doesn't publish a support email, and its customer claims are, as with any training product, easier to check in the free tier than on the marketing page.

Access is freemium. A free plan gives 15 minutes a month with one basic persona, real-time speech evaluation, vocal pace analysis, and an instant scorecard, and there's a free 60-second voice diagnostic before you sign up for anything. Pro is $66 a month or $799 a year, a 25 percent saving over paying monthly, for 120 minutes, the full persona library, pause analytics, Lumi debriefs, and scorecard history, and it can be paused for up to three months at no cost, so a rep between roles isn't paying for minutes they won't use. Team, Enterprise, and Scale are custom-priced, stepping up from 300 to 600 to unlimited minutes per user and adding the scenario builder, knowledge base uploads, custom accent personas, a dedicated customer success manager, webhooks, and uptime guarantees.