For years I assumed self-serve analytics was the answer: build good dashboards, give everyone access, and the curious ones will look.

Turns out almost no one looked. We had dashboards with 3 lifetime views, and we're a data company (embarrassing to admit).

I've watched this happen at least a dozen times now with teams we talk to: customers start asking for analytics, the team says "how hard can charts be," and 2 quarters later they've shipped a worse version of a BI tool and their actual roadmap is on fire.

The problem is that customer-facing analytics looks like one feature but it's actually 5 products stacked on top of each other:

• A query layer against production data

• Charting that doesn't look terrible

• Permissions and multi-tenant isolation (get this wrong once and you're writing a security disclosure)

• Performance that holds up when a customer asks for 2 years of data

• And now, because every customer expects it, an AI analyst that actually gets answers right

We ship constantly as a small team, and we've been on a roll launching to Product Hunt every week for over 2 months now for different features (see our launch history). Each time I worry it reads as "this again?"

Our current strategy: launch features like they're products. Give the feature its own name, own story, own audience, and don't lean on the company's history to sell it.

Every AI product I've used starts the same way: an empty chat box and a blinking cursor. The tool has all the capability in the world, but it sits there waiting for me to figure out what to ask.

That's a design failure we've all just accepted. The blank page problem killed writing apps for years, and now we've rebuilt it into every AI interface.

Six years ago I got into Y Combinator and built the first version of @Basedash: AI data analyst. It was a simple tool that let you edit data in your database.
Now, six years later, we have a whole new product, and again we're shipping... database editing. But this time it's a look into the future of how companies are going to be run.
At the surface, this feature is kinda neat but nothing new. You can ask the Basedash AI to edit things in your database or call MCP servers to take action in other tools.
But, being a BI product, Basedash has a super deep understanding of your business. Probably better than you do yourself.
Soon, Basedash will proactively start suggesting ways to make your business better.
"If you extend this user's trial, they'll have a 50% higher chance of converting"
"If you tweak your ad sequence with this copy, it'll increase conversion rate by 20%"
One click to approve and run the experiment.
This is how we transition to AI-run companies.

Please check out our PH launch for actions and give us any support or feedback: https://www.producthunt.com/prod...

We just shipped SCIM provisioning (check it out!) and it got me reflecting on how much enterprise infrastructure we've built over the past year.

When we started selling upmarket, our answer to most security questionnaire items was "not yet." Today the checklist looks pretty different:

• SSO (SAML and OIDC): authenticate through the identity provider your company already uses

• SCIM provisioning: users, groups, and memberships sync automatically from Okta or Entra as people join, move teams, or leave

• RBAC: groups with fine-grained permissions on data sources, dashboards, chats, automations, and MCP servers

• Row-level security: control which rows each person can see, and the AI respects those policies in every query it runs

• SOC 2 Type II: audited annually, encryption in transit and at rest

• Audit logs: who accessed what data and when

• Self-hosting: deploy in your own network with Docker or Kubernetes, bring your own AI model keys, and align with HIPAA, ISO 27001, GDPR, or PCI-DSS requirements

For a long time ours was ARR, mostly because it's the number investors ask about. But it's a lagging indicator. By the time ARR moves, the decisions that caused it are months old.

So we shifted focus to activation, and it honestly changed how we build. Every feature debate started with "does this get someone to value faster?" and it worked, activation went from 5% to over 50% (we wrote a blog post about it here).