Anyone using both Profound and Spotlight? How do they compare on tone control?

I've been conducting a systematic evaluation of AI writing tools for technical marketing and documentation, with a specific focus on their ability to maintain a consistent, non-generic tone across multiple outputs. The marketing promise of "brand voice" is often just a wrapper around basic style instructions, so I ran a controlled test pitting **Profound** against **Spotlight** on this exact axis. My primary use case is generating a series of blog posts for a developer audience that must avoid the overly enthusiastic, fluff-laden tone common in AI output and instead adopt a measured, evidence-based, and slightly skeptical voice.

The test brief was as follows:

**Prompt:**
```
Write an introductory paragraph for a blog post titled "The Hidden Cost of Serverless Database Transactions." The tone must be: analytical, understated, and wary of vendor hype. Avoid exclamation points and overt salesmanship. Target audience: senior engineers and engineering managers.
```

**Profound Output:**
```
The promise of serverless databases often centers on operational simplicity and cost alignment with usage. However, a deeper examination of transaction execution in these environments reveals a more complex economic model. This analysis moves beyond the per-request pricing to consider the indirect costs imposed by connection lifecycles, cold start penalties on transactional logic, and the subtle trade-offs between consistency guarantees and latency. The aggregate financial impact of these factors can significantly alter the total cost of ownership calculations that many initial evaluations overlook.
```

**Spotlight Output:**
```
Serverless databases are revolutionizing how we think about backend infrastructure, offering incredible scalability and cost-efficiency! But have you considered the true cost of transactions in this new paradigm? While the pay-per-use model is attractive, hidden pitfalls like connection management and cold starts can quietly inflate your bill. This post will uncover these hidden expenses to help you make smarter, more budget-friendly architectural decisions for your applications.
```

**Analysis and Required Edits:**

* **Profound:** Required minimal editing. It adhered closely to the requested tone. The words "complex economic model," "indirect costs," "subtle trade-offs," and "alter the total cost of ownership calculations" fit the analytical and understated brief. I only removed "This analysis moves beyond" to make the paragraph flow more directly. The output was immediately usable.
* **Spotlight:** Required significant revision. It defaulted to the very hype-driven language ("revolutionizing," "incredible scalability," "smarter, more budget-friendly") the prompt warned against. The use of an exclamation point and a rhetorical question directly contradicted the instructions. The phrase "true cost" felt clichéd. To make this usable, I would need to rewrite most of the sentence structures and replace the evaluative adjectives with neutral, factual terms.

**Methodology & Configuration Notes:**

Both tools were configured with a custom "tone guide" I uploaded, which was a style document defining our target voice with examples. Profound allowed for granular, weighted parameters on a "Formality" and "Enthusiasm" slider, which I set to "High Formality, Low Enthusiasm." Spotlight uses a "Brand Voice" training module fed with sample text.

* **Profound's Approach:** Appears to treat tone control as a direct, weighted constraint in the generation process. The output suggests it probabilistically penalizes hype-laden language.
* **Spotlight's Approach:** Seems to rely more on semantic analysis of provided brand voice samples, which can be overridden by its underlying model's default tendencies, especially on short-form content.

**Preliminary Conclusion:**

For precise, prompt-driven tone control where consistency and avoidance of marketing language are paramount, **Profound** delivered a superior out-of-the-box result in this test. **Spotlight's** output, while engaging, failed to adhere to the specific constraints and required substantial editing to meet the brief. This suggests Spotlight's strength may lie in generating engaging first drafts where strict tone adherence is less critical, while Profound operates more like a constrained text generator tuned to specific parameters.

I am expanding this test to longer-form content (e.g., 1500-word articles) to see if Spotlight's brand voice training achieves better consistency over a larger token window. I'm also curious if others have performed similar A/B tests and what your parameters were. Specifically:
* What tone dimensions are you controlling for (e.g., formality, skepticism, conciseness)?
* Have you found the effectiveness of these tools' tone features varies significantly by output length?
* Are you using API access or the native UI, and does that impact consistency?

-ek