Studio, Salon, Lab

Three rooms. Three tempos. One building.

Analytical work is not one activity. It is three activities that people confuse because they happen to use the same data. The confusion produces tools that are good at one thing and hostile to the other two. jinflow has all three rooms — and names them deliberately.

The Rooms

The Studio — where you build

Writing signals, composing theses, shaping the AFS. You are alone or with one other person. The work is slow, precise, and reversible. You try things, undo them, try again. The output is structure — an analytical framework that encodes your understanding of the domain. The Studio is a workshop. The smell is sawdust.

The Salon — where you meet

The CFO needs to understand what the data says. The consultant presents findings. The auditor verifies. The team debates whether a thesis is confirmed or just plausible. Nobody writes SQL here. The work is interpretive, social, and synchronous. Everyone is looking at the same screen — often literally. The output is a decision, or at least a shared understanding. The Salon is a drawing room. The smell is coffee.

The Lab — where you test

Something doesn’t look right. A signal fires too often, or not at all. A thesis score is suspiciously high. You open the hood: run queries, inspect edge cases, compare distributions, calibrate thresholds. The work is exploratory, iterative, and messy. You follow hunches. Most of them lead nowhere. The output is evidence — or the absence of it. The Lab is a workbench. The smell is solder.

---

The Tempos

Each room has its own clock.

	Studio	Salon	Lab
Tempo	Days to weeks	Minutes to hours	Hours to days
Mode	Asynchronous	Synchronous	Semi-synchronous
People	1-2 (author + reviewer)	3-10 (presenter + audience)	1-3 (investigator + domain expert)
Artefact	AFS commit	Shared understanding	Evidence (or its absence)
Tool	CLI + editor + git	Explorer + screen	REPL + playground + Explorer
Failure mode	Over-engineering	Miscommunication	Rabbit holes
Collaboration	Pull request	Shared URL	Shoulder-to-shoulder

The mistake most analytics platforms make is building for one tempo and forcing the other two into it. Dashboards are Salons that pretend to be Studios. Notebooks are Labs that pretend to be Salons. SQL editors are Studios that pretend to be Labs.

---

The Cycle

The three rooms form a cycle, not a pipeline.

Every rotation improves the analytical framework:

1. Studio: Author writes a signal based on a domain hypothesis
2. Lab: Signal is tested against data, calibrated, edge cases explored
3. Salon: Findings are presented, discussed, questioned
4. Studio: Feedback captured as SMEbits, signals adjusted, theses refined
5. Repeat.

The KLS is the artefact that moves between rooms. It is built in the Studio (jinflow make), examined in the Lab (Explorer + REPL), and presented in the Salon (Explorer + P2P2P). Each room adds to it — but only the Studio writes to it.

---

The Doors Between Rooms

The most interesting collaboration happens at the doors — when someone carries something from one room to another.

Transition	Carrier	What travels	What doesn’t
Studio → Lab	jinflow make	Compiled signals, theses, entities	Draft ideas, deleted attempts
Lab → Salon	Shared URL (P2P2P)	Findings, verdicts, evidence	Query history, false leads
Salon → Studio	SMEbit, notebook entry	Domain knowledge, corrections	Social context, politics
Lab → Studio	Signal adjustment, calibration	Threshold changes, scope fixes	The 47 queries that went nowhere
Salon → Lab	”Can you check…?”	A question	The answer (that comes later)
Studio → Salon	Release tag	A versioned, frozen KLS snapshot	The messy history behind it

The cycle is healthy when all three doors are open. Pathologies emerge when a door is blocked:

• Studio → Lab blocked — Signals are authored but never tested. They look correct in YAML but fire on nonsense data.
• Lab → Salon blocked — Evidence exists but never reaches decision makers. The analyst knows, the CFO doesn’t.
• Salon → Studio blocked — Feedback is given but never captured. The same questions come up every quarter. Institutional knowledge stays in people’s heads.

---

The Naming

The Explorer already names its spaces with these metaphors:

• Price Lab — a lab. You compare, you investigate, you test.
• Map Salon — a salon. You look together, you discuss what you see.
• Spend Studio — a studio. You compose views, you build analyses.

These names are not accidental. They encode the collaboration mode:

• A Lab invites experimentation. “Let me try something.”
• A Salon invites conversation. “What do you think?”
• A Studio invites creation. “Let me build this.”

Future Explorer pages should choose their suffix deliberately. The suffix tells the user not just what the page does, but how to use it and whom to bring along.

---

What jinflow Provides for Each Room

	Studio	Lab	Salon
Primary tool	CLI + text editor	Explorer + REPL	Explorer (P2P2P)
Build	jinflow make	—	—
Explore	—	jinflow evolve, Signal Builder, Calibration, Price Lab, Diff Loupe	Overview, Theses, Findings, Executive Summary
Capture	Signal/thesis/SMEbit YAML authoring	Query results, notebook entries	SMEbits (from conversation), notebook entries
Collaborate	git push/pull, AFS diff	Shared KLS (read-only)	P2P2P session URL
Version	Git commits, KLS snapshots	— (ephemeral by nature)	Snapshot tag (frozen for the record)
Present	—	—	Presentation mode, executive summary, reports

---

One Product, Three Modes

P2P2P is primarily a Salon technology — it connects people to data for shared understanding. But the cycle means P2P2P also serves the Lab (a remote expert reviewing findings) and indirectly the Studio (feedback from a Salon session becoming a git commit).

• Studio mode — offline-first, git-based, CLI-driven. No proxy needed.
• Lab mode — local Explorer, full interactivity, REPL access. Proxy optional.
• Salon mode — P2P2P, shared URL, read-only, presentation-optimised. Proxy essential.

One product. Three rooms. The user moves between them as the work demands.

---

Related

• P2P2P — Your data, your machine, their browser
• Operating Modes
• Collaboration Guide
• Explorer Guide