Use cases by role

flowss is one platform for very different kinds of work: a systematic review and its PRISMA flow, a verified pathway figure, a course students rebuild from memory, an architecture that lives in a repository, a strategy map, a process model you can simulate. This page is a practical guide to those workflows. It explains the use cases page, then takes each of its six roles in turn (researchers, scientists, educators and students, architects and engineers, product teams and founders, ops and process teams) with the studios involved, the plan each step needs, six starter templates, and a step-by-step workflow built from real controls. It ends with the role-based starter packs in the command palette and a quick "where should I start?" table.

At a glance

Across every role:

• The Studio and Sketch are free. The other seven studios (Study, Evidence, Weave, Science, Figure, Lab and BPMN) start on Starter. See Plans and what they include.
• AI is optional and plan-dependent. There is no AI on Free. Starter runs AI on your own key; Plus and Ultra add hosted AI points. Everything marked "no AI" below works without either. See AI in flowss.
• Work moves between studios. Every studio has Send to another studio…, so you can start in one and finish in another. See Moving work between studios.
• Templates are the fastest start. 5,012 of them, browsable without an account at /templates. See The template gallery.

The use cases page

The page at /use-cases ("Different disciplines. Shared possibilities.") helps you find the workflow closest to yours. Find your workflow scrolls to it.

1. Choose a role from the six tiles (Researchers, Scientists, Educators & students, Architects & engineers, Product & founders, Ops & process teams). Each tile shows the role's tagline. With a tile focused, the left and right arrow keys move between roles, and Home and End jump to the first and last.
2. The panel below shows the role's tagline as a heading, a paragraph on how flowss serves it, a Studios row (each studio is a link that opens it), and four short points on what you can do.
3. Open … (for example Open Lab) opens the role's front-door studio; Browse 5,012 templates opens the gallery.
4. Six starter templates shows six hand-picked templates for the role. Each card shows a preview, the engine, the title and a two-line description; Open opens the template in the studio that owns it. Engines that are only drawn inside their studio show a designed card instead of a preview: BPMN 2.0 (Lint · simulate · mine), Weave (Drag-and-drop canvas) and D2 (Diagram-as-code).

At the foot of the page, "Your next idea starts here." links to the full collection (Explore the collection) and to these guides (Read the guides).

Opening a studio or a template follows the usual rules: you are asked to sign in if you are not, and a studio your plan does not include sends you to the pricing page. See Opening a template in a studio.

Researchers

Publication figures, evidence attached.

Write figures as typed FlowScript in the Lab (PRISMA 2020 and CONSORT flows with the reporting-guideline checks built in) and hand the papers behind them to Evidence for synthesis. Compose journal-sized panels in Figure, and export TikZ for LaTeX submissions, SVG and PDF.

Starter templates

A typical workflow: from papers to a submitted figure

1. Frame the review in Evidence. Open /atlas, type your research question into the prompt, and add sources: paste abstracts, upload PDFs, search OpenAlex with Find papers, or import RIS, BibTeX or PubMed files with Import references…. Not sure yet? Load the example synthesis shows a complete run.
2. Screen and synthesise. If you imported a long list of records, screen them first in the Screen panel (with independent votes and agreement if you work in a team). Then press Synthesize evidence graph at the foot of the Sources flyout. Every claim Evidence keeps carries a verbatim quote it has checked against your source text.
3. Read the evidence. Select a link to see its GRADE certainty, the reasons for it, the backing quotes and, where effects are comparable, a pooled estimate. Switch to the Table view for the evidence table or the Summary of findings.
4. Export the synthesis. Choose Report (Word .docx), Report (HTML) or Report (Markdown) for a report with Methods, findings, GRADE and PRISMA, plus BibTeX (.bib), RIS (.ris), Evidence table (.csv), Claims (.csv) or Evidence graph (.json). See Screening, extraction and synthesis.
5. Draw the flow diagram in the Lab. Open /lab, choose Browse templates… and start from a PRISMA or CONSORT template, then type your own counts. The Lab checks that the numbers add up between blocks, so a figure whose arms do not sum to the randomised cohort is flagged before a reviewer sees it. Reporting standards… checks the figure against its reporting guideline. See Lab, Lab views and renderers and Reporting standards.
6. Compose the results figure in Figure. Open /figure, choose Import data file… (CSV, TSV or JSON), pick a recommended chart for each panel, then Page setup… to choose the journal and column fit. Run compliance check measures width, height, type size, resolution and colour against the journal; Generate figure legend drafts the legend; Export repro bundle packages the figure with its data, sources and a README. See Figure and Figure analysis and statistics.
7. Hand it to the manuscript. From the Lab, Publish this figure… and Export PNG at print size (300 dpi); from Figure, the LaTeX and Markdown copy options; from the Studio, Export as TikZ / LaTeX (⌘L), which converts a figure to a standalone LaTeX document you can copy, download or Open in Overleaf. See Publishing, provenance and compliance in Lab and Exporting your work.

Note: Evidence's claim extraction and the Studio's TikZ conversion use AI: your own key on Starter, or hosted points on Plus and above. With your own key, requests go to your provider on your own account. The Lab's arithmetic checks, Figure's compliance check and Evidence's GRADE, pooling and PRISMA counts use no AI.

Scientists

Figures where the science is checked.

The Science studio verifies as you draw: every edit runs deterministic checkers over the figure's text (equation balance, unit consistency, physical constants, DOI, CAS and PDB identifiers, species nomenclature), and each finding carries a one-click fix where one can be computed.

Starter templates

A typical workflow: a pathway figure that is correct

1. Pick a domain. Open /science and choose Biology, Chemistry, Physics or Mathematics. The domain decides the shape library, the templates and the AI's worked examples.
2. Start from a template or the shape library. Drop illustrated glyphs, connect them, and put the process on each arrow: in a pathway, the arrow's direction and label are the claim.
3. Let the checks run. Findings appear continuously in the status glyph. Choose Verify science (check every claim on the canvas) for the full list (reactions, dimensions, measurements, constants and identifiers), each with its fix. No AI is involved in any check.
4. Use the tools. Chemistry tools (balance, molar mass, pH, dilution), Units, dimensions & physical constants and Periodic table are in the command palette; Run model (simulate or analyse the models drawn on the canvas) runs a model you have drawn.
5. Tidy and caption. Auto-layout diagram (⇧⌘L), Distribute horizontally and Align left arrange the figure; Title and caption… adds the text that travels with it.
6. Export or compose. Export PNG, SVG, PDF, TIFF or JSON, or Send to another studio… to combine it with data panels in Figure.
7. Hand off when you need more fidelity. For a 3D molecule, a phylogeny or a free-body diagram drawn to scale, use the Studio's science engines (see Maths, science and academic engines).

Tip: The Science studio's Describe with AI (⌘E) drafts quickly; the checks are what make the figure trustworthy. Draft, then verify.

Educators and students

Courses that are diagrams, not slides.

Study turns any subject into a diagram-first course: lessons, an always-on tutor, and knowledge maps you rebuild from memory on a spaced-retrieval clock. The Classroom adds cohorts and assignments, and the Academy at /learn teaches flowss itself.

Starter templates

A typical workflow for a teacher

1. Pick or build a course. Open /study. Start from one of the eight built-in courses, or use New course (AI) with a topic, pasted notes or up to twelve files. To fork a built-in course into one you can edit, use Customize.
2. Turn a diagram into a course. From any studio, Send to another studio… → Study hands the diagram to Study's course creator as notes.
3. Set up a cohort. Open the Classroom (or Open Classroom (cohorts & assignments) in the Studio's ⌘K), start a cohort and share its join code.
4. Post an assignment. Give it a title, instructions and a starter diagram, written by hand or drafted with the AI lesson studio. Students work on it, save drafts and submit.
5. Review and grade. Read submissions, write feedback and enter a grade; a grade makes the work final until you clear it.

A typical workflow for a student

1. Join your cohort with the code under Join with a code in the Classroom, and open each assignment in the full Studio when you need more room.
2. In Study, orient yourself on the Learning Map, then work through Lessons: each one asks you to predict, build and explain before it tells you.
3. Rebuild models from memory on the BuildBench: every relationship gets its own verdict (reversed, wrong type, extra or missing) rather than a single score.
4. Press Start today's review each day: concepts and flashcards share one queue, scheduled so you see each idea just before you would forget it.
5. Ask the flowss Study Agent (Ask the flowss Study Agent) when stuck; it answers from the course you have open.
6. Check Progress and Calibration to see where you are confident and wrong.

Tip: New to flowss? The ten Academy tracks take about 75 minutes in all, are free, and each ends with a short exam. Passing every track unlocks a final exam and a certificate of completion.

Architects and engineers

Architecture that lives in the repo.

Write diagrams as code in Mermaid, PlantUML, Graphviz, D2, Structurizr, DBML and ERD (text your repository can diff and your pull requests can review), or drag and drop on the Weave canvas with containers, cylinders and swimlanes when the review needs a whiteboard.

Starter templates

A typical workflow: architecture as code

1. Sketch it first, if you like. In /sketch, draw the boxes and arrows on a whiteboard, then Convert sketch to diagram to get engine source. A deterministic read of clean shapes and arrows needs no AI; on Free you can convert to Auto-detect, Mermaid, Weave, D2 or Graphviz, the other convert engines come with Starter, and the AI reading of rough ink needs AI (your own key on Starter, or hosted points on Plus and above).
2. Write the source in the Studio. Pick the engine your team uses (Mermaid, PlantUML, Graphviz, D2, Structurizr DSL, DBML or ERD) or start from a template. Bring an existing schema or board in with Import SQL / DBML / draw.io (paste or file).
3. Keep the text canonical. Use Copy source or Download source and commit the file. Convert to another engine… moves a diagram to a different notation when you need one.
4. Review on a canvas when it helps. Send to another studio… → Weave lays the diagram out on a drag-and-drop canvas, and Weave's Copy as Mermaid, Copy as DOT, Copy as D2 and Copy as PlantUML bring it back to text.
5. Model the system around the boxes. Event storming, state machines, data-flow diagrams and fault trees are engines of their own; see Flow-systems analysis in Studio.
6. Publish where the team reads. Embed — copy iframe / Markdown snippet puts a live diagram in a wiki or README; the render API draws diagrams from CI; the REST API can verify that a diagram still matches the code; the MCP server lets AI assistants list engines, generate, convert, render and diff diagrams. See Embedding and the render API, Integrations and extensions and The flowss MCP server.

Tip: Version history… (⇧⌘H) in the Studio keeps earlier states of a figure, and a living diagram bound to a public repository flags drift when the code changes. See Version history and undo.

Product teams and founders

Strategy you can point at.

A native Wardley Maps engine (components plotted on visibility and evolution, with pipelines, evolve arrows and build, buy or outsource markers) sits beside the full strategy kit: business model and lean canvases, user story maps, funnels, quadrants and RACI matrices. Type it, drag it, or describe it and let the AI draft a first pass.

Starter templates

A typical workflow: a strategy workshop

1. Run the session in Sketch. Open a workshop board from Boards and canvases, and use Facilitate a session for a timer, reactions, dot voting and gathering ideas into columns. End session and write the outcome records what was decided.
2. Formalise the result in the Studio. Map the value chain with a Wardley Map, capture the model in a business model or lean canvas, and plan the release with a user story map. Each starts from a template in the gallery.
3. Prioritise. Use a quadrant (value against effort) for the backlog and a RACI matrix for ownership.
4. Present. In Sketch, Present frames as slides (⇧⌘E) or Export deck (PDF) / Export deck (.pptx); in the Studio, Present (fullscreen) or Export PDF; in Weave, Present and Export PowerPoint (.pptx).
5. Share. Copy share link creates a permalink with the diagram inside the link, so anyone can view it without an account. Because the diagram travels in the link, treat it as public. See Sharing and permissions.

Note: AI drafting (the flowss Studio Agent, and the flowss Sketch Agent's conversion and wireframe generation) needs your own key on Starter or hosted points on Plus and above. Everything else in this workflow works without AI.

Ops and process teams

Model it, lint it, run it.

A dedicated BPMN 2.0 studio with inline checking, one-click repairs and token-flow simulation that reports cycle times and bottlenecks, plus process mining that discovers the model from a CSV event log. Around it sits the flow-systems kit: value stream maps, control charts, Pareto charts and service blueprints.

Starter templates

A typical workflow: from event log to improved process

1. Discover the process as it really runs. In /bpmn, choose Mine process from event log and load a CSV of case, activity and timestamp rows. Move the noise filter to hide rare paths; the map is re-discovered as you change it, and opens as its own diagram tab. (From the Studio, Data → Chart (⌘D) can discover a process from the same kind of log.)
2. Or model it by hand. Start from Browse templates, a blank diagram, a .bpmn file, or Photo → BPMN from a whiteboard photo.
3. Fix what the checks find. The status badge merges the Problems (BPMN lint) and Compliance checks with a deadlock check; many findings have a one-click repair.
4. Simulate. Switch to Play to watch tokens move, see live cycle time and the bottleneck, set durations, branch weights and costs in Sim setup, and run a batch of cases in Analyze for cycle-time percentiles and the top bottlenecks.
5. Measure the improvement work. In the Studio, a value stream map computes lead time against work time, a p chart separates special causes from noise, and a Pareto chart shows the vital few causes. See Flow-systems analysis in Studio.
6. Hand it over. Export BPMN 2.0 XML for Camunda, Signavio or Bizagi; Export process handbook (HTML) or (Markdown) for roles, steps, decision tables and compliance with the diagram embedded; Save version to keep a snapshot. See BPMN.

Role starter packs in the command palette

You can tell flowss what kind of work you do in two places: the Studio's welcome tour (⌘K Open onboarding tour), which asks "What kind of work do you do?", and Studio Settings → Personalization & privacy. Selecting the same role again clears it. Your role tunes recommendations: the ⌘K palette's For you section (headed, for example, For you · Researcher) leads with the templates you use most and then your role's pack. In the welcome tour, picking a role also previews its accent colour; changing the role in Settings keeps the accent you have. The profile is optional and stored only in your browser.

Where should I start?

Tips

• Start in the free Studio even if your workflow ends in a paid studio: most diagrams you make there can be sent onward with Send to another studio….
• Search the gallery by what you are drawing ("prisma", "value stream", "wardley", "pedigree") rather than by engine.
• Set your role in Settings so ⌘K suggests templates for your field.
• Keep source canonical: copy or download it alongside every export, so the figure can always be regenerated.
• Templates contain illustrative values; replace every number with your own before you publish.
• For teaching, pair the Classroom (free) with Study (Starter) and the Academy (free).

Limits and known constraints

• Seven of the nine studios (Study, Evidence, Weave, Science, Figure, Lab and BPMN) need Starter or above. A paid plan may be on a waitlist until checkout opens for it; the pricing page always says which.
• There is no AI on the Free plan. AI features need your own key (Starter and above) or hosted AI points (Plus and above).
• Live sessions, comments and shared projects start on Plus; voice, screenshare and shared team libraries are on Ultra. See Live collaboration.
• Figure has no live sessions, comments or share link; hand its work on with exports or Send to another studio….
• A share permalink carries the diagram in the link itself; treat it as public.
• The six roles and their templates are suggestions. Any studio and any template is open to any role that has the plan for it.

Troubleshooting

Related pages

• What flowss is
• Quickstart: your first diagram in ten minutes
• The template gallery
• Learn tracks
• Moving work between studios
• Lab
• Evidence
• Figure
• Science
• Study
• Classrooms, cohorts and assignments
• Weave
• Sketch
• BPMN
• Flow-systems analysis in Studio
• Plans and what they include
• AI in flowss