Short answers to the things people ask most.

A publishing platform for AI and ML practitioners. Native code rendering, real LaTeX, subfield-based discovery, full-article email delivery that keeps math and code intact, and Stripe-powered paid subscriptions — built specifically for people who write about machine learning.

Researchers, applied ML engineers, ML educators, and anyone who has tried to publish a post with a Python block on a generic blog and given up.

You connect a Stripe account in your dashboard. Subscribers pay you directly through Stripe — Empirq never holds your money. Stripe pays out to your bank on its normal schedule (typically 2–7 days depending on your country). The full revenue terms are in the Writer Agreement, §4 Monetization.

Empirq retains 10% of net subscription revenue; writers keep 90%. Founding writers (the first 20 accepted before public launch) keep 100% of net subscription revenue for as long as the account remains in good standing. Stripe’s processing fees (~2.9% + $0.30 per transaction in the US) are charged separately by Stripe. Full terms in the Writer Agreement, §4 Monetization.

Yes. Free publications can publish unlimited posts and accept free subscriptions. Paid subscriptions require Stripe Connect onboarding.

General-purpose newsletter platforms treat technical writing as an edge case — code arrives as gray boxes with no syntax highlighting, and math either renders as images or doesn't render at all. Blogging tools get closer but aren't built around AI/ML readers. Empirq is built specifically for technical writing: code blocks render with full server-side syntax highlighting, math renders cleanly inline and in blocks, and posts are discovered by curated subfield (interpretability, RL, vision, alignment, systems) rather than by an open tag cloud alone.

The curated list of AI/ML fields a post can belong to — NLP, computer vision, reinforcement learning, interpretability, MLOps and around twenty more. You pick up to five per post. Subfields are what readers browse and filter by, and what they choose interests from when they join, so they are the backbone of discovery on Empirq. The list is maintained by us rather than typed by writers, which is what keeps browsing coherent: one field means one thing, and a reader following it gets what they expected.

Subfields are where readers browse. Tags are what your article is about. Subfields come from our curated list; tags you type yourself, and they display with a leading # so the two are never mistaken for each other. Use tags for the specific ideas a subfield is too broad to express — concept drift, data drift, model-centric AI — where the subfield might only say MLOps. You can add up to five of each per post. Tags are searchable straight away, and every tag has its own page listing the posts that use it. Once five published posts share a tag, readers can follow it and new posts carrying it start appearing in their feed.

No, and that is deliberate — an open list of fields stops being a map. Tags exist for exactly this: anything more specific than the curated fields, or outside them, belongs in a tag. A post does not have to carry a subfield at all, so work that genuinely sits outside AI/ML can publish with tags alone and be found through them. If you keep reaching for the same tag and others do too, that is the evidence we use to decide whether it should become a subfield.

New posts in it start appearing in your feed. That is all it does — following a subfield or a tag never sends you email. Email comes from subscribing to a publication, which is a relationship with one writer who chose to send it. Interests are for browsing; subscriptions are for your inbox.

Yes. Export your Substack subscriber list as CSV (Substack has an export button under Settings → Import / Export), then upload it on /dashboard/subscribers/import. We deduplicate, validate emails, and create free subscriptions for everyone in the list. They keep getting your newsletters from you, now via Empirq.

No. Free posts are public. Paid posts require a paid subscription, but no account is required to read free posts or subscribe.

Yes. Every post is delivered to subscribers as a complete email — code blocks, math, headings, and images intact — not a “read more” teaser that bounces them back to the browser. Math and code render correctly in the inbox, so a technical article still reads like one. For paid posts, paid subscribers receive the full article and free followers receive a preview.

Newsletters are sent with proper SPF/DKIM/DMARC authentication on empirq.com. We track bounces and complaints and automatically disable subscribers that bounce too many times — which keeps your sender reputation clean.

Not at launch. Every publication lives under empirq.com/your-publication-slug, and that’s the URL your subscribers and SEO-indexed readers will see. Custom domains (e.g. posts.yourname.com) are on the post-launch roadmap — they need DNS, SSL, and deliverability work that isn’t justified before we have the writer base to support them.

Subscriptions are non-refundable by default but writers may issue discretionary refunds. The full policy — including chargebacks and cancellations — is in the Subscription & Refund Policy, §6.

Visit /settings/subscriptions and click Cancel — or click Manage to go through the Stripe Customer Portal. Cancellations take effect at the end of the current billing period; see the Subscription & Refund Policy, §4 Auto-Renewal.

You can export everything as Markdown at any time from your dashboard. If you delete your account, your posts are removed; if you keep your account but stop posting, your existing content stays accessible to your readers. See the Terms of Service, §12 Termination for the full terms.

No. Empirq does not license, sell, or share your content with third parties for training AI models. Your content may be processed by automated systems strictly to operate the Platform — for example, building search indexes, generating recommendations, or detecting spam — as described in the Writer Agreement, §3 Content Ownership and Licensing. If our use ever changes beyond that scope, we’ll update this answer and notify writers before any change takes effect.

Email [email protected].