The Chat Essentials kit is for people who work in a plain AI chat such as ChatGPT, Claude or Cowork. It gives your AI four workflows (research, decisions, follow-ups and writing) that need no agents and no repository. Every step saves to a task note, so you can close the chat and resume later.
4 templates · 21 steps · 6 approvals · 4 loops back
one chip per stored step · square chips wait for a person
Four workflow templates, installed together. Each one is a stored workflow your AI client receives one step at a time over MCP.
01 · 6 steps · 2 approvals
Answer a research question with a sourced report. The AI drives Scope, Gather, Synthesize, and a review checkpoint.
02 · 4 steps · 1 approval
Make a decision you can defend later. The AI drives Options and a recommendation checkpoint, then records the call.
03 · 5 steps · 1 approval
Turn a meeting into a sent follow-up. The AI drives through Extract, Draft, and a send-review checkpoint.
04 · 6 steps · 2 approvals
Take a blog post from brief to published draft. The AI drives through Outline, Draft, Edit, and a final read-through.
The four templates cover the work a chat user does most, and two of them chain by design: a research report's final note is the named input for a decision memo. Each saves its output as task notes, so the next session, or a different client, picks up where the last one stopped.
| From | To | How the work moves |
|---|---|---|
| Research Report | Decision Memo | Research Report Complete names its saved report as the input for a Decision Memo, and Options pulls from its findings. |
| Meeting Follow-Up | Blog Post | A customer conversation often becomes a post; the Blog Post brief starts from it. |
Each template runs on its own task. They connect through what a finished run leaves on record, not through a shared run.
Every approval step in the kit, read from the stored workflows. The run holds at each one until you answer.
| Template | Approval step | If you reject |
|---|---|---|
| Research Report | Scope step 2 | no rejection path; approve to continue |
| Final Review step 5 | back to Synthesize | |
| Decision Memo | Recommendation step 3 | back to Options |
| Meeting Follow-Up | Send Review step 4 | back to Draft Follow-Up |
| Blog Post | Outline step 2 | no rejection path; approve to continue |
| Final Review step 5 | back to Edit |
An illustrative run, labelled as an example. The step names are the templates' real steps.
You ask for research on EU expansion. The AI states the question and scopes it; you approve the scope.
Gather saved its findings with sources as a task note. Nothing lives only in the conversation.
You resume. The AI reads the notes, synthesizes the report, and you approve it at Final Review.
A Decision Memo starts from the report, lays out options and recommends. You make the call; it is recorded.
Connect your AI client first, then install the kit from the catalog in the ConvOps app. See how to connect a client
Installed templates are workflows in your workspace. Ask your AI to change one in plain words; the edit lands on the step that needs it.
new chat
more example prompts
No. The four templates save their work as task notes in ConvOps, not as files in a repository, so a plain chat client connected over MCP is enough.
Yes. The run state and every saved note live in ConvOps, not in the chat, so any connected client can resume the task from the step it stopped on.
The Free plan includes 3 workflow templates and this kit has 4, so the full kit needs Pro or above.
Any MCP client connected to ConvOps. Verified clients: Claude Code, Claude (desktop, web, Cowork), OpenAI Codex, ChatGPT, Cursor, OpenCode. None of its workflows needs a repository.