Software Should Work for You, Not the Vendor

Here's a question I find useful: whose interests does this software serve?

Not in a cynical way. Most software companies genuinely want to make their users happy. But there's a structural misalignment between vendor interests and user interests in the SaaS model that shows up in subtle ways in product decisions.

Understanding this misalignment helps explain a lot of product decisions that seem puzzling if you assume the vendor's goal is purely to make users happy.

The Engagement Trap#

SaaS companies with venture backing have an incentive to maximize engagement. Not because engagement is inherently good for users, but because engagement is a proxy metric that investors and acquirers use to assess value.

High engagement correlates with:

• Higher retention (engaged users don't churn)
• More data to train models and personalize recommendations
• More upsell opportunities (you're in the product more, you see more features)
• Higher acquisition multiples

The problem: maximizing engagement and maximizing user productivity are not the same thing. An app that makes you more productive might make you spend less time in it. That's bad for engagement metrics. An app that has more friction (more clicks, more features to navigate) might increase time-in-app while reducing your actual productivity.

This is why CRMs have so many features you never use. Feature-rich products have more surface area for engagement. More features = more reasons to be in the product = higher time-in-app metrics = better acquisition multiple. The features aren't there because users asked for them; they're there because they increase the product's valuation.

The Lock-In Feature#

Data lock-in is often presented as a security feature or a consequence of technical complexity. Sometimes it is. But it's also a deliberate product decision that serves the vendor, not the user.

Making it easy to export your data and take it to a competitor is obviously bad for vendor retention. So exports are often: limited in completeness (some fields don't export), export only in proprietary formats that require conversion, slow or throttled for large datasets, available only on higher subscription tiers.

None of these limitations serve the user. They serve the vendor's retention goals. The user's interest is: I should be able to take my data anywhere at any time in a useful format. The vendor's interest is: the harder it is to leave, the less I have to compete on product quality.

DenchClaw takes the opposite position by design. Your data is in a DuckDB file. You can read it with any DuckDB client without involving us at all. You can export it to CSV, Parquet, or JSON with a single query. The switching cost is intentionally zero, which means we have to compete on actual product quality rather than inertia.

The Pricing Ratchet#

SaaS pricing tends to go in one direction: up. Not because product quality increases proportionally, but because switching costs and inertia make price increases possible without corresponding product improvements.

Here's the dynamic: a SaaS company acquires users at a low price point (sometimes free). Users adopt the product, build their workflows around it, train their team on it. The switching cost grows as the product becomes more embedded in daily workflows. At some point, the vendor raises prices, betting that the switching cost exceeds the price increase. Often they're right.

This is the pricing ratchet. Every price increase in the SaaS model is the vendor extracting the switching cost they've built up. It's not predatory — it's the rational behavior of a company that built a good product and is now capturing the value of the lock-in they've created.

The user's interest is: I should pay for value delivered. The vendor's interest is: I should charge for switching cost. These are not the same thing.

DenchClaw is MIT-licensed and free. We don't have a pricing ratchet because we don't have a subscription. The commercial revenue comes from Dench Cloud (managed hosting), which you can leave any time with your data intact. The alignment is: you pay for convenience (managed hosting), not for permission to use your own data.

What Aligned Software Looks Like#

Software that's genuinely aligned with users has properties that are easy to state and hard to build:

Your data is portable by default. Not available on request, not exportable for a fee, not in a format designed to be annoying to migrate. Portable by default, in open formats, at any time.

The product is faster when you do less in it. Productivity software that's aligned with users should make tasks faster, not keep you in the app longer. The metric should be "how quickly can you accomplish what you came to do?" not "how much time did you spend in the app?"

Pricing is proportional to value delivered. Not to inertia, not to switching cost, not to whether the pricing page has enough opacity to hide the real cost. Simple, proportional, defensible.

You can audit what the software is doing. For AI-powered software especially: you should be able to see what data the AI used, what it did with it, and why it made the decisions it made. Opacity serves the vendor. Transparency serves the user.

The software improves from your use without using your data against you. Learning from usage is fine. Selling your behavioral data, training models on your confidential information, using your usage patterns to optimize against your interests — not fine.

DenchClaw tries to embody all of these. Open source, so auditable. Local-first, so your data is yours. MIT-licensed, so portable forever. Agent-designed to reduce time-in-UI, not increase it. No user data sold, no behavioral analytics beyond opt-in crash reporting.

The Honest Trade-Off#

I want to be honest about what you give up with software that's aligned with users rather than optimized for vendor metrics.

Polish. Enterprise SaaS products have enormous teams working on polish, onboarding, and the kind of fit-and-finish that comes from decades of product iteration. DenchClaw is newer and rougher. The trade-off is real.

Ecosystem. HubSpot has thousands of integrations. Salesforce has an entire AppExchange marketplace. DenchClaw has a growing skill ecosystem that's early. If you rely on specific integrations, you may find some not yet available.

Support. You can't call a DenchClaw support line. There's a Discord community and GitHub issues. For enterprise needs that require dedicated support, Dench Cloud is the answer.

These are real costs. I'm not pretending software that's aligned with users is strictly superior on every dimension. It's superior on the dimensions that matter most over a long time horizon: ownership, cost, privacy, and trust.

Frequently Asked Questions#

How do I know DenchClaw's agent isn't collecting data I don't know about?#

The source code is on GitHub under MIT license. You can read every line of code that handles your data. If you want to be certain, you can audit the code or disable all telemetry in the settings.

What happens if I have a problem and need help?#

GitHub issues for bugs and feature requests. The Discord community (shared with OpenClaw) for questions and community support. Dench Cloud customers get email support. There's no phone support.

Can DenchClaw be used by a non-technical person?#

The install requires a terminal command. After that, the primary interface is conversational. For genuinely non-technical users, the install step is a barrier. We're working on a one-click installer for macOS.

What data does Dench collect about my usage?#

Anonymous crash reports and aggregate performance telemetry (no personal data, no CRM content) if you opt in. This can be disabled. We never collect CRM records, contact data, deal information, or any of the content you store in DenchClaw.

Is local-first software always better than SaaS?#

No. For applications where centralization is inherent (social networks, payment processing, communication platforms), SaaS is the right architecture. For personal and team productivity tools where your data is primarily yours, local-first is often better.

Ready to try DenchClaw? Install in one command: npx denchclaw. Full setup guide →