Prompt System Workspace

One prompt system. Every client. Every model.

Stop rewriting the same prompts for every client or project. Build them once as reusable blocks with {{variables}} — then copy the finished system into ChatGPT, Claude, or Gemini.

Free during early access. Your prompts are plain text — copy them out anytime.

Get started See a live example → Log in

Project Map · Multi-Client Assistant Base

Assemble copy into: ChatGPT Claude Gemini

Watch one system serve three clients

The same base rules, three different clients. Switch one — the variables do the rest.

# Base Assistant Rules
You are the Northwind Logistics support assistant for a B2B freight platform.
Follow the escalation and refund rules and keep a concise, professional tone in every reply.

A miniature of PAL's brew step, running right on this page — the real builder does this across whole systems, for every client or your own project. More example systems →

How it works

Compose

Assemble text blocks and blend references into one prompt system — rules, tone, format, all in one place.

Parameterize

Mark the parts that change per client or per project with variables.

Copy & hand off

Copy the finished system into any model, or hand it off as a map your team can read and reuse.

Steal a working system

Three real prompt systems, fully readable, no signup required. Open one, copy the assembled system, and see the reuse for yourself.

Client Brand-Voice System

One tone ruleset, reused across every client's blog content.

Tone & Voice Rules
Formatting Rules
{{client}} Voice Profile
Blog Post Generator

See the system

SEO Content Pipeline

Brief, outline, draft, and meta — four blends, one keyword in.

Keyword Brief
Outline Generator
Draft Generator
Meta & Title

See the system

Multi-Client Assistant Base

One base ruleset underneath every client-specific assistant.

Base Assistant Rules
Output Contract
{{client}} Overrides

See the system

Why not just use ChatGPT, Claude, or Gemini projects?

Native project features are great for a single conversation. They weren't built for reusing one prompt system across clients, models, and a team.

Comparison of native model projects versus Prompt Alchemy Lab
Capability	ChatGPT · Claude · Gemini projects	PAL
Reuse across clients	Not supported	Supported
Modular blocks (DAG + variables)	Partial	Supported
Generation history you can search	Partial	Supported
Works across models	Not supported	Supported
Hand-off (Project Map)	Not supported	Supported

PAL doesn't run the model — it's where you build the prompt, then you copy it into the model you already use.

Questions you should ask

Does PAL run my prompts?

No — deliberately. PAL is where you build and reuse the prompt; you copy the result into whatever model you already pay for. No API keys, nothing routed through us.

What happens to my prompts if I leave?

They're plain text. Copy any blend or a whole Project Map out at any time — there's no proprietary format to escape from.

Is it really free?

Free during early access — every feature is unlocked. Prices shown in Pricing are founding-member rates: once billing goes live, early sign-ups keep them.

Why not just keep prompts in Notion or a doc?

Docs store text; they don't compose it. In PAL a block is referenced, not pasted — update the base once and every system that uses it follows.

Can my team use it?

Today PAL is a personal workspace: you hand systems off as copyable text and a readable Project Map. A shared team workspace is on the public roadmap — planned, not built.

Pricing

Simple plans for every team size.

Free during early access — every feature above is unlocked, including items marked “planned” (public roadmap, not yet built). Prices shown are founding-member rates: once billing goes live, early sign-ups keep them locked in.