How to use PremiseArc
PremiseArc is an AI Workflow Coach. It helps you run structured playbooks, learn from each run, and improve over time — with AI execution available inside PremiseArc when you choose.
The coaching loop
CREATE a playbook · RUN in Guided mode · LEARN from outcomes and Run Insights · IMPROVE by editing your playbook · RUN AGAIN when you choose. Your Dashboard shows Continue your coaching work and Recent outcomes so you can pick up without hunting through menus.
Getting started
- Choose a guided starter workflow or create a playbook from scratch.
- Edit steps on the playbook editor. Use the optional Refine instruction wording control for this step only, or link reusable assets from your prompt library (optional).
- Run the playbook in Guided mode (recommended).
- On each step, use Generate with AI when you want PremiseArc to run this step with AI — you review the result before continuing.
- Complete the run to save an outcome you can review later.
Understanding playbooks
A playbook is a repeatable workflow: ordered steps you run again and again. New to workflows? Browse guided starter workflows to create your own copy, or link prompts from your library on the playbook editor. The Authoring Coach on the editor helps with structure — not prompt wording (use Prompt Coach on the Prompts page for that).
Running a playbook
From a playbook, choose Run this playbook. Guided (recommended) walks you step by step with coaching. Manual is optional if you prefer to copy prompts to an external AI tool or work entirely on your own — PremiseArc still tracks your progress when you save outputs.
Auto is a separate automation mode and is not offered as a live Production feature. Do not expect unattended multi-step execution in public Production.
How AI works in PremiseArc
PremiseArc keeps you in control. AI runs only when you ask — inside a structured workflow, not as an open-ended chat product.
- Generate with AI — execution: have AI create or fill in this step's answer when you choose. Review before you continue.
- Step Coach — coaching: get help understanding or improving this step or your answer when you ask. It does not run the step for you.
- What AI uses for this step — transparency: read-only preview of what PremiseArc may send for Generate with AI (this step plus eligible earlier steps in the same run).
- Automatic guidance — rule-based tips PremiseArc may show when it notices something useful. This is not Step Coach and not a chat you started.
You can still copy prompts to an external tool (see External AI workflows). PremiseArc does not use cross-run AI memory today.
AI and your data (staging beta)
- Generate with AI and Step Coach run only when you choose them.
- When you invoke AI, relevant current-step information and eligible same-run context may be sent to your configured AI provider.
- PremiseArc does not currently use cross-run AI memory.
- Use “What AI uses for this step” on a run step to preview context.
Send feedback
Closed-beta testers can share issues and ideas using the Send feedback link in the header, or set NEXT_PUBLIC_BETA_FEEDBACK_URL to your team's form.
Using Generate with AI
Generate with AI means you start AI for one step at a time inside PremiseArc. It is not automatic playbook execution. Review the response, edit notes if needed, then move to the next step when you are ready.
Run insights
During and after a run, Run insights are observations — things PremiseArc noticed about how the run went. They are not proof that you improved or that a problem was fixed.
Improvement proposals
Improvement proposals suggest changes you might make to your playbook. They are recommendations to consider — PremiseArc does not apply them automatically. Use Edit playbook when you decide to change structure; revision context on the editor can remind you what you were thinking (session-scoped).
Reviewing outcomes
An outcome is a historical snapshot from one completed run — captured when you completed that run. It shows what coaching and proposals were available at that moment, not a live recompute. Open outcomes from your dashboard or Recent outcomes.
Run progress and recurring patterns
Run progress on a playbook shows completion history and what recurred across multiple runs (same text or proposal showing up again). Recurrence means it appeared more than once — it does not mean PremiseArc proved a later run improved.
Recording your decision on proposals
On Run progress, recurring improvement proposals can have a disposition (Reviewed, Dismissed, Revisit later, Not applicable). That records your choice — not approval, not application, and not proof anything was fixed. Change disposition on Run progress; outcomes may show a read-only mirror.
External AI workflows (optional)
You can still copy a workflow prompt or individual step prompts to ChatGPT or another AI tool — especially in Manual mode. That remains supported, but it is not the primary PremiseArc experience. Prefer Guided mode and Generate with AI when you want to stay inside PremiseArc.
Prompt Coach and Authoring Coach
Prompt Coach (Prompts page) helps refine wording. Authoring Coach (playbook editor) helps workflow structure. Both support the same goal: clearer workflows and better runs.