All case studies
Zincan internal studyDeveloper tooling

Zincan

We test Mindy on zincan.com — here's what we learned in the first quarter.

Our newsletter draft turnaround dropped from a half-day project to a 20-minute review pass. Same week, same writer.

Nate, founder

Results

3.4×
newsletter draft turnaround
. Source: From ~4 hrs to ~70 mins per send, averaged across 6 weekly sends (Mar–Apr 2026)
−68%
time spent re-reading old posts
. Source: Self-reported, internal log over 4 weeks. Mindy surfaces the relevant excerpt with citation in seconds.
180hrs/mo
saved across the team
. Source: Conservative roll-up of time-tracking across 4 contributors. Excludes Slack / async wins that are harder to attribute.

Zincan is an AI venture studio. We write a lot — product updates, founder essays, customer stories — and the archive grew faster than our memory of it. Writers were spending half their drafting time re-reading old posts to make sure they weren't repeating themselves, contradicting themselves, or missing the obvious link back.

We tried the obvious things first: a custom GPT trained on the archive, then Notion AI's question-answering. Both were fine at generic prose and useless on our terminology — they'd invent stack names, mis-attribute quotes, and confidently hallucinate facts about our own roadmap. We needed something grounded in the actual archive, not adjacent to it.

Mindy is that thing. She reads the archive on every question, cites the originating post, and refuses to answer outside what she's actually seen. We've been running her against zincan.com since the beginning of the quarter. This is what changed.

Before & after

Concrete shifts the team can point at, not vibes.

Weekly newsletter draft
Before
~4 hours: re-read recent posts, hunt for the angle, draft from scratch
After
~20 minutes: Mindy drafts a candidate from the week's archive, human edits and ships
Finding the right old post to link to
Before
Open-tab archaeology — search the site, scroll, give up, paraphrase
After
Ask Mindy in Slack, get the post + the exact paragraph that's relevant
Monday content review
Before
30-min meeting, two people, one half-read shared doc
After
5-min async thread: Mindy posts the week's summary, humans react with 👍 or follow-ups

How they use Mindy

The day-to-day surfaces, not the marketing pitch.

Daily content review

Every morning Mindy posts a thread in #content with what shipped yesterday, what's still in draft, and what evergreen posts are getting unexpected traffic. The thread takes 90 seconds to scroll; the half-hour Monday meeting it replaced did not.

Newsletter drafting

We ask Mindy for a candidate draft each Friday afternoon. She pulls from the week's posts and last quarter's most-engaged archive, drafts in our voice, and includes inline citations. The human editor's job became review-and-trim instead of stare-at-a-blank-page.

Linking strategy

When a writer is mid-draft and wants to link back to a previous piece, they ask Mindy for the relevant prior posts. She returns the three best candidates with the exact paragraph each one would slot in alongside — no more open-tab archaeology, no more accidentally re-writing the same essay.

What we don't use her for

Honesty section: we don't use Mindy for original analysis, hot takes, or anything that needs a human point of view. She's the institutional memory, not the voice. We also don't use her on draft posts that aren't yet committed to the archive — she only answers from published material, so unpublished context is on us.

Try Mindy on your site.

Connect your WordPress site and your Slack workspace in under ten minutes. Mindy reads your archive on day one — no training, no tuning, no hallucinations on the names you already know.