MoreYess — sales-call rehearsal, locally

What MoreYess is

MoreYess is a local-first sales-practice product. Sales teams use it to rehearse calls, refine their playbook, and run scenarios against a local LLM — with nothing leaving their machine. The workload that matters: sub-second first-token latency for natural conversation, structured playbook output, and zero cloud dependency.

Why it's on this site

MoreYess appears here only as a worked example of how benchmarks become a product. The case study connects the workload requirements to the benchmark evidence — specifically which latency targets, model choices, and quantizations made the product viable. see bm-001

The evidence-to-product chain

• Latency target: < 800ms TTFT for natural rehearsal cadence. Our Arc B60 Pro Vulkan result (740ms) meets it; CUDA on RTX 4090 (410ms) exceeds it.
• Model choice: 30B-class coder/general model at Q4_K_M — the sweet spot from our quantization sweep.
• Hardware floor: 24GB VRAM to fit the model with KV cache headroom. This is the "Arc Pro Frontier" build recipe.

The offer

These are the only authoritative facts, sourced from content/products/moreyess.yml:

• Current tier: Founding Member License — $97.
• Regular tier (after founding): Regular License — $197.
• Limit: first 50 customers at the founding price.
• Included: lifetime access and all future updates.
• No free trial. There is no trial, and we will not advertise one.

Editorial firewall

The benchmarks that justify MoreYess's hardware choices are determined only by what the hardware does. The fact that we own the product has zero input into whether Arc B60 Pro hits 740ms TTFT — it does, and we'd publish the same number if we didn't own MoreYess. See our full disclosure.

If MoreYess isn't for you

That's fine. The case study is still useful: it shows the decision chain from workload → latency target → model → quantization → hardware. Apply the same chain to your own product or use case using Model Fit and our benchmarks.