LADLE for teams.

Team AI subscriptions have a predictable death pattern. Someone champions it. Seats get purchased. The champion demos it in one all-hands. Week 1, usage spikes. Week 3, it drops. Week 6, half the seats haven't logged in for 21 days. Renewal comes up and the CFO kills it.

The cause is almost always the same: the team didn't establish shared patterns for how they'd actually use it, so it stayed a novelty for the champion. This guide is about avoiding that pattern.

Start with one shared Voice Guide

Every team has a voice — the way you write PRDs, or the way you respond to customers, or the way your CEO writes internal memos. The single highest-leverage thing you can do on day one is convert that voice into a Voice Guide attached to a shared Project.

A Voice Guide is prose, not a template. Half a page describing register, forbidden phrases, structural conventions, and 3-5 examples of writing in your team's voice. Attached to a Project, every chat in that Project follows it. This is the difference between "LADLE writes generic AI prose" and "LADLE writes in a way that could plausibly go into your PR template".

Create 3-5 shared Projects, not 30

The temptation is to spin up a Project per person, per initiative, per anything. The failure mode: nobody knows which Project is the right one for a given task, so they default to a fresh chat, and the Projects go unused.

Better: 3-5 workspace Projects that correspond to real ongoing work streams. Examples: "Q3 planning", "Customer research synthesis", "Engineering RFCs". Each has its own Voice Guide (usually inheriting the team's default with additions), its own attached context files, and a clear "when do I use this Project" description at the top.

Run a shared prompt library

Every team develops 10-20 prompts they use over and over. Meeting-notes-to-action-items. Weekly-status-update. PRD-first-draft. Instead of everyone reinventing these individually, collect them in one place your team already reads (a Notion page, a shared doc, whatever).

LADLE's /prompts library is a good starting point but not enough — your team's prompts are more specific. A shared library that grows over time is one of the highest-return artifacts you'll create.

Model the pattern from the top

If leadership uses LADLE visibly — in Slack replies, in shared docs with a "drafted with LADLE, edited by me" attribution when it's true — the team adopts it. If leadership uses it privately and never mentions it, the team assumes it's a personal tool, not something the org endorses.

What not to do

- **Don't require training sessions.** Nobody remembers a 60-minute training. They remember watching a colleague use it well on a specific task. - **Don't audit usage.** LADLE deliberately doesn't give admins per-user chat visibility. Building a compliance-style rollout ("we're watching your usage") kills adoption. - **Don't standardize prompts too early.** The first 30 days are for discovery. Standardize what actually gets used, don't try to predict. - **Don't upgrade the plan tier until you have usage data.** Same principle as personal plans — Base is right until data says otherwise.

The 30-day check-in

At day 30, look at three signals: (1) how many seats have sent at least one message this week, (2) whether the shared Projects are being used, (3) whether any patterns of use have emerged that would benefit from becoming shared prompts. If seat activation is under 60% and shared Projects have <10 chats, the rollout is drifting toward the shelfware pattern — course-correct now, not at renewal.