Help centre

Signing in and your workspace

Sign in with your work email: we send you a link, no password needed. Your first sign-in creates a workspace; add colleagues from Administration > People (on the Team plan and above). See People, invitations and teams.

How to model: the process modelling guide

New to process modelling, or to Twiggy? Open Help > Process modelling guide (or the card at the top of the Help centre). It walks through modelling top down, a process map first, then a process chain for each area and a process flow for each process, with naming conventions, best practice and common mistakes.

Choosing what to model

Choose New model and what you are modelling:

• Process map: the enterprise processes and the process areas inside them. No flow.
• Process chain: a new process area on a map, ready for its processes. An area’s chain is simply the processes inside it, as blocks, and how they connect. No lanes.
• Process flow: one process, its steps, decisions, hand-offs and who does what. Placed under a step of another flow, it is that step’s sub-process.

You do not pick a level: how deep a model sits follows from where you place it.

The dashboard

The Dashboard lists sign-offs waiting for you, your maps and recent activity. To try things out first, choose Explore the demo.

Tabs of open models

Every model you open gets a tab. Middle-click a tab to close it, or right-click it to Close others, Close tabs to the right or Close all.

Using Twiggy across an organisation

The guide Twiggy in your organisation (Help > Guides) has playbooks for operational risk and controls, regulatory change, critical operations and resilience, process improvement, training and onboarding, and audit, plus a rollout plan for the first three months and who does what.

Opening Twiggy every day

Bookmark twiggymodeller.com/app. If you are signed in, it opens your workspace straight away; if not, it asks you to sign in and then takes you back. With a passkey, signing in is one touch.

Modelling real-world complexity

Process modelling isn’t about making complex work look simple. It’s about making the complexity understandable. Five principles:

• Model the work as it is done, not as it should be. Keep the exceptions and workarounds that matter; simplify by grouping steps, not by leaving things out.
• Keep the flow for the order of work. Rules, exceptions, controls and requirements go on the steps, so the diagram stays readable.
• Make the boundaries clear. Link each hand-off to the process it comes from or goes to.
• Tell what’s needed from what could go. Some complexity the business needs (an award, a law); some is left over from the past. Mark the second kind so it can be improved, not just copied.
• Write down what only people know. Note who knows how an awkward case is handled before they move on.

Go only as deep as each process needs: a process map for the big picture, flows for the work, and the detail on the steps where the risk or the change is.

Where each thing goes, on a step:

• Business rule: a criterion every case follows (“loans over $1m need two approvers”). Kept in the Business rules register and reused.
• Requirement: an obligation the step must meet (a law, a policy, a customer promise).
• Exception or workaround: a group of cases the usual way doesn’t fit, and what people actually do (“staff on the 1998 agreement are paid by hand”). Local to the step and often unwritten.
• A branch in the flow: only when the work genuinely goes somewhere else.

To record one, right-click a step › Properties… › Exceptions. Say what’s done instead and, if the knowledge sits with few people, who knows it. Tick Could go when the business doesn’t need it. Pin the section if you want it in the side panel. The AI asks about exceptions only once the main flow is in place.

Your name, photo and profile menu

Click your picture at the top right. The profile menu shows your workspace, plan and AI left this month. Upload a photo replaces your drawn face. Edit your name and titles sets how you’re greeted and shown to others. Admins can edit anyone’s name in Administration > People.

Importing Visio drawings

File > Import a package, .pml, BPMN or Visio…, or drop a .vsdx file on the dashboard. Each page with a flow on it becomes a process.

• Steps, sub-processes, decisions, start and end shapes, swimlanes and connectors come across. Connector text becomes the line’s label, and shape data becomes properties (owner, description, duration, cost). Systems are linked to your Applications register by name; ones it doesn’t have become notes and are listed. Anything else becomes a note.
• Off-page references become hand-offs, linked when the page they point to is in the same file.
• If the drawing uses stencil shapes Twiggy doesn’t know (a company stencil), it asks what each one is, with a guess. Tick Remember and the next drawing from that stencil imports straight away. Admins can change or forget these in Administration > Imports.
• A step drawn twice in a lane to avoid crossing lines becomes one step. Repeated lines are merged, and a loose connector end joins the nearest shape.
• Twiggy lays the flow out itself, so positions, colours and fonts aren’t kept. The report lists everything that was simplified or left out.

Save older .vsd files as .vsdx in Visio first. Free workspaces get 3 BPMN or Visio imports or exports in total; paid plans have no limit.

Interview, Feedback and Review

Open the AI assistant beside a model. Build drafts and changes the model, by interviewing you or a subject-matter expert one question at a time or from what you describe; Feedback applies comments you have pinned to elements; Review looks for gaps, edge cases and rules, and shows the model’s health against good modelling practice, with Fix with AI for what falls short.

Changes appear on the canvas as the AI makes them, and you can undo the last one. Details that belong elsewhere are parked for later rather than lost.

What to ask the AI

The guide Working with the AI (Help > Guides) has prompts to copy for each job, and when to use them:

• A first draft: “Let’s start with a high-level overview”, or paste a description or an existing procedure.
• Checking: compare with industry practice (such as BIAN or the APQC framework), find risk and control gaps, check against an obligation, work through what can go wrong, check the names.
• Improving: find waste, sketch a simpler future state as a variant, spot automation candidates.
• Using it: explain the process to a new starter, brief an approver, draft work instructions for a role.

Give the AI context (who the process is for, the product, the rules that matter), ask for one thing at a time, and confirm anything it marks with ?.

Adding and connecting

• Add with the palette or a key (T task, D decision, E event, S subprocess, L lane), after what is selected.
• Drag the + handle to connect; drag an element onto a line or into a lane to move it.

Renaming and the right-click menu

• Double-click (or Enter) to rename.
• Right-click anything for what you can do with it: Change type, Add after, and marking it unconfirmed (?). A whole lane can be unconfirmed too. Double-click the name at the top of the menu to rename.

Downloading a picture of the diagram

Download the diagram as an image (PNG) or SVG from the ⋯ menu, for slides and documents.

Arranging the panels

The Details panel shows the main properties of the selected element; the rest are chips under More. Right-click an element and choose Properties… (or press Alt+Enter) to see every section. Click the pin beside a section to show it in the panel for every element of that kind, and the pin on a panel heading to take it out again.

The Details panel shows the selected element’s properties, and Review & sign-off walks you through what is still open. On each panel’s title bar: ▲▼ folds it, « » folds the whole side away, and ⇄ moves it to the other side. Panels share the height with their neighbours: drag the gap between them. Drag a column’s inner edge to make it wider.

Naming checks

Twiggy checks names gently, in the issues list: a step or process area that reads as a name rather than work done (with a suggested rename, e.g. *Credit check* → *Check credit*), a decision that isn’t a question, and a flow with so many steps it is worth splitting into sub-processes. They stay reminders, even once a process is signed off.

How it’s done: a step’s instructions

Each step has a description (what it is) and How it’s done (how to do it, one line per instruction), in Details. Presentation mode and the procedure guide show it under the description.

• Modelling top down? Write a rough how on a high-level step. When you drill down, the AI starts the sub-process from it.
• Create its sub-process (under the box) makes the step a sub-process whose draft flow has a step for each line, every one marked unconfirmed (?), ready to model in detail. On an AI draft, the arrow beside Accept offers Accept and create its sub-process.
• Once a step opens a sub-process, its how is that flow: Details and Presentation mode show the sub-process’s steps instead, and you change them there.
• Draft with AI writes it from the step, its systems, controls and notes. A draft is marked *Drafted by AI* until you Accept it, or edit it.

Procedure guides

A procedure guide is a readable page built from your models: what starts each process, its steps in order with who does them, the systems, how each is done, the checks and why they matter, and where each decision goes. Change the model and the guide changes.

• For a process area: right-click it on the map (or in the explorer) > Procedure guide. You get every process inside it, in order, with where they fit and who does what. A level like *Originate SME loans* is the usual one.
• For one process: ⋯ > Procedure guide above its flow. In a process chain, Guide in its toolbar.
• Print at the top prints it or saves it as PDF.
• Web addresses in descriptions and instructions are links, so point to your knowledge articles there. A register item with a *url* field links too.
• The cover says how many steps have instructions yet, and warns when parts are not signed off.

Moving elements by hand

The layout is automatic, and you can fine-tune it:

• Drag an element within its lane and drop it away from any line. It snaps to line up with its neighbours (a blue guide) or back to its automatic place.
• Alt+Shift+arrow keys nudge the selected element a little at a time.
• Right-click > Position > Line up with the step before, or Back to its automatic place.
• Right-click the empty canvas > Reset layout clears every hand move.

Dropping an element on a line or in another lane still moves it there. Undo covers all of it, and exports keep your positions.

Getting around: up a level and back

• Up a level: left of a flow’s name is the level above it (the process area on the map, or the parent flow). Click it to go there; … lists every level up to the map. In a process chain, the arrow at the top left goes up to the chain above, or to the map.
• Back: the arrow at the start of the tabs returns to where you came from, such as the map’s chain you opened a flow from, or a guide.

Presenting, and high contrast

Present (above the diagram) shows a process or map full screen, one element at a time, with its details below: for workshops and training.

A start or end event that meets another process has a small arrow on it: click the arrow, or double-click the event, to present that process at the event where the two meet. Back (top left) returns.

For projectors or easier reading, turn on High contrast: in the View menu, or at the top of Presentation mode. Lines, borders and text go darker and bolder. It is one setting for both, kept in your browser.

Linking applications, risks and controls

Applications, risks, controls, business rules, requirements and roles live in registers, shared by every model in the workspace (Registers, under the explorer).

In a step’s Details, type in Risks, Controls or Applications to search the register. A name that matches nothing becomes a new item: proposed, usable straight away, and approved by the register’s owners (admins add items directly).

You can also open a register in the explorer and drag an item onto a step on the diagram, or onto the Details panel while a step is selected.

Open a register to see every item in a grid, with its fields and where it is used. Export it as CSV, or as a .pml library that imports into another workspace.

Find an item fast with the search box above the registers in the explorer (all registers, or pick one), then drag it onto a step.

Build a hierarchy on the register page: New › New group, then drag items onto the group (or pick them and choose Move or Group). An import with category columns builds groups for you.

Owners and admins can edit in the grid (double-click a cell), pick many rows and change them together, import CSV or Excel with a check first, and undo any change from Changes. Each item keeps its history.

Groups, editing many items, importing and undo

• Groups: drag items onto a group, in the hierarchy or in the grid, to put them in it; drop them on New group to make a group of them. Or pick them and choose Move or Group. Drop on All … to lift items back to the top.
• Many at once: tick rows (or Ctrl/⌘-click, Shift-click, or Pick all that match), then set a field, move, group, make active or retire them together.
• Edit in place: double-click a name or a field in the grid, or change it in the item’s panel. Changes save at once, without reloading the list.
• Right-click a row (or the item’s panel, or its ⋯) to add an item under it, move it, group or retire it, or approve a proposal.
• Where it is used: the processes that use the item, under their place on the process map, with the steps in each. Search it when the list is long.
• Columns: click a heading to sort by it (any field), drag its edge to make it wider or narrower, and use Columns (or the arrow on a heading) to hide one. Your columns are remembered in this browser.
• Import: Import a CSV or Excel file, map its columns (category columns build groups), check exactly what will change, then import.
• Import into registers (Administration › Imports): CSV files, or an Excel workbook with a sheet per table. A file named after a register (risks.csv) goes into it. A row can also hold linked items: name the columns after their register (Risk: Name, Risk: Rating, Control: Name…) and each row creates both and links them (a control mitigates the risk). Rows about the same item count once, and importing the file again updates rather than duplicates.
• Templates and mappings: Download template for the registers you pick gives a file the import reads by itself. For someone else’s file, Mapping… says which register and field each column goes to; admins save it under a name to use again, or from the API.
• Reuse only (Fields and owners): for a register kept in step with another system. The AI only picks existing items and tells you when nothing fits; a risk or control a model defines for itself is flagged, with Propose it to the register to send it to the owners.
• Owner fields take an email address, a name or a role title (as process owners do), so a file from a GRC tool can carry titles.
• Retire with a replacement: the models that used the retired item can move to the replacement in one go.
• Delete (right-click, or with rows ticked) removes items for good, with anything grouped under them, but only items no model uses: those you retire instead, so no model is left pointing at nothing. To clear a register, Pick all that match, then Delete.
• Undo: every change is listed under Changes, and each can be undone; each item’s History tab shows what changed.
• Fields and owners (admins): add your own fields, lists of options and the people who own the register.

Relations and the Linker

Items can be linked to each other as well as to steps: a control mitigates a risk, a control implements a requirement, a rule enforces one, an application depends on another, a role owns an item. Open an item and use Link… under Relations.

• The Linker links many at once. Open it from Linker beside Registers in the explorer, or from a register’s page. It starts empty.
• Drag into a pane: a process, a map, an area or a folder from the explorer (or several you picked) for their steps, or a register for its items. Steps always sit on the left. Or use Or choose… in an empty pane.
• Each pane shows the hierarchy (the processes above each step, the groups above each item) and searches it. A step is found by what it already uses too: search “credit risk” for the steps that carry one.
• Pick on both sides (a group’s box picks everything under it), then Link. A ✓ shows what is already linked to everything picked on the other side, ◐ to some of it. Undo session takes it all back. What’s in the panes is kept until you close the browser tab.
• Signed-off processes, and processes you can’t edit, are left alone and listed: start a new revision to change them.

The matrix

Open Matrix beside Registers in the explorer. Drag processes (a map, an area, a folder or a selection) or a register into Rows and Columns, or choose them.

• Processes against a register: how many steps of each process use each item, for example which systems or risks each process relies on.
• A register against a register: what is linked, for example controls against risks. ● in the register and on a step, ◐ on the same step only, ○ in the register only.
• Both sides keep their hierarchy. Fold a group to add up everything under it, or use Fold all, Open a level and Open all. Rows with nothing shows the gaps, such as risks no control covers.
• Save keeps the rows and columns for everyone in the workspace; pick one from Saved matrices. CSV exports the matrix as it is folded.

Insights: what needs attention

Open Insights from the dashboard’s Workspace tools for what needs attention in the processes you can see: reviews overdue or due soon, risks no control covers, steps with a risk but no control, retired items still in use, steps on systems due to retire, requirements nothing meets, items no process uses, and proposals waiting for approval.

Click a row to fix it: a step opens its process at that step, an item opens on its register. The checks in each model flag the same things: a retired item is a warning, a proposal or an old key a note.

Typing a process map as an outline

In a process map, type the processes as an outline: Enter adds the next one, Tab puts it inside the one above, Shift+Tab takes it out. Drag a box into another to nest it. Any process area can open one or more process flows; right-click it to create one, or to open its process chain. Use Show and Compact to see more at once.

Process chains: how processes connect

An area’s chain is the processes inside it: there is no separate chain to create. Right-click a box on the map › Add child process › As a child to start one, or › Open chain view for an area that has processes: its processes, as chevrons, and the connections between them. A box that is one process gets a flow instead (› Add child view › Flow view). Connections have no direction; tick Show direction for arrows from the hand-offs.

• Drawn (dashed): you connected them, from Connect or a process’s right-click menu. Modelling top down, draw these first.
• Confirmed (solid): drawn, and a hand-off between their process flows backs it up.
• From hand-offs only (faint): their flows hand over to each other but no connection is drawn. Click it and Draw it on the map.
• Drawn, not in the flows (amber): both have flows and none hand over to the other. It shows under to check, and in the map’s issues.

Below the chain are the processes outside the area that its processes hand over to or from. Double-click one for the chain it is in.

The explorer

The explorer on the left shows your models in their hierarchy. One click selects, a double-click opens, right-click adds or deletes. The folder-tree button also shows process areas on the maps that have no process flow yet.

Map health and the clean-up review

On a process map, open the AI panel’s Review tab. Map health scores the map from 0 to 100: processes with a flow, owners, descriptions, names as work done, even structure, no duplicates, and connections. Click a check to see the processes behind it, and one of them to find it on the map. The line shows how the score has moved.

Review the map with AI suggests fixes: duplicates, names, granularity, placement, gaps and missing descriptions, each with why. Accept applies one (undo takes it back); Dismiss leaves it. Suggestions that wouldn’t apply cleanly, or would cut a flow off the map, are left out.

Where a process is used, and its variants

With nothing selected, a process’s Details show Where used: the map areas and the steps of other flows that open it, the processes it comes from and hands over to, and its variants.

A variant is the same work done differently, for example per product or channel. Mark it with `variant_of=<process id>` in its header (in the model text, where your workspace allows it, or ask the AI); it then shows under its base process’s variants.

Hand-offs between processes

A start event can come from another process, and an end event can hand over to one. Select the event and use Comes from or Hands over to: the step before (or after) this one in the parent process is suggested first, and Find matches with AI looks across every model for the same moment under another name. Use ◂ ▸ in the toolbar to move between neighbouring processes.

Reviewing a model, then sign-off

Choose Help > How to review a model for a guided tour that sets the screen up as it goes:

• The AI switches to Review and checks the whole model for gaps; Fix with AI handles what falls short.
• The review list shows only the Unconfirmed ? items. Step through each with ◂ ▸: Confirm, Discuss with the AI, or tick several and Send to AI together.
• In Feedback, click an element, type what should change and press Enter; repeat for as many as you like, then Apply all with AI.

To ask people, click the status chip next to the model’s name (or ⋯ > Review and sign-off):

• Request a review: subject-matter experts give feedback, all at once or one after another, on the whole model or just some lanes. Each gets a private link, no account needed, when their turn comes. Their comments come back to the Review list; nothing is signed off.
• Send for sign-off: the model goes to the approvers of its folder (any one or all, as the folder says), or, outside an owned folder, to whoever you nominate (a team member or any email address). You can add stakeholders who must also approve just their lanes.

The Sign-off tab shows where the model stands and every round, step by step. A signed-off model is locked: a change starts a new revision, and readers keep the signed-off version until the revision is signed off.

Your processes, risks and reminders

Your processes: when the models name owners (*Head of Infrastructure*), pick the title you hold on the dashboard, or ask an admin to set it in People. The processes it owns are listed there, and the Owner view marks them *(you)*, with Show mine.

Risks (dashboard): every risk the processes carry, worst first, with the steps it sits on, their controls, and whether each process is signed off and due for review. Filter, then Download CSV.

Reminders: people who haven’t answered a sign-off are reminded after 3 days, then every 3 days (up to 3 times). Remind in the review dock sends one now.

Discuss, Review, Sign-off: which to use

The review dock at the bottom of a model has three tabs, used in this order (any of them can repeat):

• Discuss, any time and informal: questions pinned to elements, and Ask people… to send them to the right people. Nothing here changes the model.
• Review, a structured round: Request a review sends links to subject-matter experts. Their comments and the AI’s findings become items to accept, park or Work in with AI.
• Sign-off, the approval: Send for sign-off. Once approved, the model is locked and published to readers.

Between rounds, Review > Changes since last sign-off shows what you changed on the diagram.

Comments, @mentions and asking people

Right-click any element, lane or map process > Comment…. Threads appear in Details > Comments. With nothing selected, Details lists every thread on the model.

• Type @ to mention someone in the workspace. They get one email with a link to the element (at most one an hour per model).
• Reply, Resolve or Reopen a thread. A reply reopens a resolved thread. You can delete your own comments.
• Discuss > Ask people…: pick open questions and people (Twiggy suggests lane owners, the process owner and people who commented). Each person gets one email, and the questions wait under Questions for you on their dashboard until they reply.
• Send open comments to the AI turns answers into suggested changes for you to check.

Reviewers and up can comment; anyone who can see the model can read the comments.

Comparing versions: what changed

Review > Changes since last sign-off, or Compare with a version… (and Compare on any version in Version history), colours the diagram: added in green, changed in amber, everything else faded.

What changed lists every change in words (renamed, moved lane, property old → new, lines, lanes, scope card). Click a change to go to it. Removed items are only listed. Press Esc to stop comparing.

Sign-off links show the same view, so approvers see what changed since they last approved.

Time and cost per step

Give steps a Duration (hands-on time, e.g. 15m, 2h, 1d = 8 hours), a Wait before it and a Cost per run, under Performance in Details, or straight in the Time and cost panel.

View > Analysis > Time and cost colours each step by its time (red takes the most) and shows:

• the happy path (the main route through each decision), the longest path and the shortest path, each with elapsed time, hands-on time, waiting time and cost;
• how much of the elapsed time is hands-on, and which steps have no duration yet;
• what-ifs: out takes a step out, ½ wait halves its wait, and the totals update. What-ifs aren’t saved.

Modelling standards (house rules)

Admins set the standards every process model is held to in Administration > Modelling standards:

• each built-in check: as standard, off, info, warning or error;
• your own standards: every step names its system (or is manual), has an owner, a description or a duration; every process has an owner and a measure;
• the most steps in one flow before Twiggy suggests sub-processes.

The standards apply to each process’s Issues and its health score, and the AI follows them when it interviews and reviews. They don’t yet change the map health score.

People, invitations and teams

Administration > People:

• Add people: set someone up with a role, name and teams. Send the invitation now, or later with Send invite.
• Each person shows a status: Active, Invited (and when), or Not invited yet.
• Tick people to change roles, invite or remove several at once. The … on a row has the same actions for one person.
• Export CSV and Import CSV… move people, roles, teams and owner titles in bulk, with a preview before anything changes.

Administration > Teams groups people, so folders can be owned by a team rather than by named people.

Seats: who needs one, and how many you pay for

A seat is for someone who builds or changes models: modellers, admins and the owner. Approvers, reviewers and viewers never need one, and anyone can sign off from an emailed link without joining.

Administration > Plan, billing and backups > Seats shows seats in use, who holds one, and the places included for approvers, reviewers and viewers (10 for each seat on Team).

• On Team, admins choose how many seats to pay for with − and +, then Update. The change is prorated on the next invoice. You can’t pay for fewer seats than the number of people holding one.
• Add a seat automatically (on by default): making someone a modeller adds a seat. When it’s off, that’s refused until an admin adds a seat, so the bill stays fixed.
• On Corporate, seats are by agreement: contact us to change them.

Each seat also adds to the workspace’s monthly AI allowance.

What readers see: the published library

Viewers are readers. They see signed-off versions of processes, never a revision in progress, plus any drafts shared with them. Their home page is a search over what is published: steps, roles, systems, risks and controls.

• Workspace setting: Administration > People > What readers see.
• For a folder: right-click > Ownership and access > Drafts for readers.
• For one process: Review > Share with readers…, optionally until a date.

A shared draft is marked Draft, not signed off. Every sharing change is in the audit log.

Plans and the AI allowance

• Free: every feature, with a starter AI allowance.
• Personal: just you, with as many reviewers as you like, and a monthly AI allowance.
• Team: 2 to 20 modellers in a shared workspace, billed per modeller, with roles, two-step sign-in, unlimited reviewers and twice Personal’s AI allowance for each modeller, pooled.
• Corporate: a managed service with dedicated support and coaching, shared or dedicated hosting, single sign-on, IP allow-listing and your own encryption keys; talk to us.

Your picture at the top right shows how much of the allowance you have used, and admins get an email at 80% and when it runs out. When it runs out, everything except the AI keeps working; on a paid plan it renews on the 1st of each month, or an admin can buy an AI top-up (see below).

Roles, permissions and two-step sign-in

• Owner: everything, including the plan and billing; can hand ownership to someone else.
• Admin: models, plus inviting people, changing roles and workspace settings.
• Modeller: creates and changes models. Each modeller is a seat.
• Approver: signs off where named Accountable on a folder; can’t change models. Free, no seat.
• Reviewer: reads and comments; free, no seat. Viewer: reads only.

In a folder, teams are Responsible (build and send), Accountable (approve), Consulted or Informed: right-click the folder > Ownership and access. Manage people and teams in Administration > People and Teams. For two-step sign-in, see Passkeys and two-step sign-in below.

Cancelling your plan or deleting your data

Open Administration > Plan, billing and backups (or Settings > Plan and your data for your own account).

• Cancel subscription: your plan stays until the end of the period you have paid for, then the workspace moves to Free. Your models stay. Changed your mind? Keep my subscription.
• Card and invoices opens Stripe’s secure page.
• Download all models first if you want a copy.
• Delete this workspace (the owner only) deletes every model, interview, review and sign-off in it, and ends its subscription now.
• Delete my account deletes the workspaces where you are the only member, and takes you out of the ones you share.

The shared demo

The demo is one workspace anyone can open without signing up, so everyone using it sees the same models. Please don’t put personal or confidential information in it. It is cleared every day, and you can reset it from your profile menu > Reset the demo.

Reporting a bug or asking for a feature

Choose Help > Send feedback or report a bug. Only your team and the Twiggy team can see what you send. Your team’s requests shows everything your workspace has sent, where it is up to, and any reply from us.

Downloading, importing and moving sets of models

Every model is plain text you own. A package is a set of them in one .zip that imports back exactly: the models as .pml files in folders like yours, the register items they use, and your workspace properties.

• A map with its flows: right-click it in the explorer > Download > The map with its flows (.zip).
• A folder: right-click it > Download folder (.zip). Everything: right-click empty space in the explorer, or File > Export.
• Several models: Ctrl or ⌘ click them, then right-click > Download.
• Import: drop the .zip on the dashboard, or right-click a folder > Import into this folder…. You choose what happens to models already here: keep yours, or replace them (a new version; the history keeps yours). Imported models start as drafts.
• Moving: drag a map onto a folder and its flows go with it; drag a folder into another folder; or right-click > Move to folder… (also for several at once).

A package is a full backup of its registers too: groups, retired items, every field, aliases and relations come back as they were. A whole workspace takes its registers whole; a map or folder takes the items its models use.

Any zip of .pml files imports too: its directories become folders.

Importing and exporting BPMN

Choose Download BPMN from a process model’s ⋯ menu for a BPMN 2.0 file with its lanes and layout, or Open in bpmn.io to download it and open bpmn.io, then drop the file onto the page. File > Export > Every process as BPMN downloads every process at once.

Twiggy’s own details (applications, risks, controls, rules, ? marks) travel inside the file as extensions that other tools keep. BPMN is included in every paid plan; the free plan has 3 imports or exports to try.

To import, drop .bpmn files on the Dashboard (or File > Import a package, .pml or BPMN…). Each pool becomes a process, with its lanes, task types, decisions and parallel paths; message flows between pools become hand-offs. A short report lists anything simplified, such as a boundary timer (now a path from a decision after its step) or data objects (left out). Twiggy lays the model out its own way; files exported from Twiggy come back exactly as they were.

Passkeys and two-step sign-in

• Passkeys: your picture > Security > Add a passkey on this device. Then choose Sign in with a passkey: your fingerprint, face or device PIN, with no email and no code. A passkey counts as two-step sign-in on its own. Add one on each device you use.
• Authenticator app: on the same page, scan the QR code with an app such as Microsoft or Google Authenticator and enter the code it shows. Keep the recovery codes safe.
• Remember this device: after a two-step sign-in, tick it so that browser isn’t asked again for as long as your workspace allows (up to 90 days). Not on a shared computer.

Admins choose who needs two-step sign-in (nobody, admins and owners, or everyone) and how long devices are remembered, in Administration > Sign-in and security (Team and Corporate).

More AI: top-ups

On a paid plan, admins can buy an AI top-up from Plan and your data, or from the message that appears when the allowance runs out. A top-up is used only once the month’s allowance is spent, is shared by the workspace, and lasts 12 months. Plan and your data lists what is left of each.

Organisations, workspaces and shared registers

An organisation holds your workspaces and the registers they share, such as one group risk and control library used by every division. Models, folders and people belong to each workspace.

• Workspaces and admins (Administration > Workspaces and admins): see the workspaces, switch between the ones you belong to, and, on the Corporate plan, add more. The owners of its workspaces and anyone named there are organisation admins.
• Registers and sharing: which register each kind of reference (risks, controls, applications…) comes from in this workspace. Organisation admins decide, per workspace, whether each shared register can be used, only read, or not seen.
• A workspace’s own register: ⋯ > Give this workspace its own register, copied from the shared one (keys and all, so models keep working) or empty. Only that workspace sees and changes it; switch back any time, or share it with the organisation.
• Who changes what: shared registers are changed by their owners and the organisation’s admins; a workspace’s own register by its admins and owners. Everyone else can propose items.

Single sign-on (Entra ID, Okta, Google)

With single sign-on, people at your company sign in with their work account instead of an email link. An organisation admin sets it up under Administration > Organisation > Single sign-on.

• In your identity provider, register Twiggy as a web app with the redirect URI shown on the page, and make a client secret.
• Connect it: the issuer URL (Entra ID: https://login.microsoftonline.com/<tenant id>/v2.0), the client id and secret, and your email domains.
• Prove your domains: add the TXT record shown to each domain’s DNS, then Check DNS. Nobody can switch single sign-on on for a domain they can’t prove.
• Save and test signs in at the provider without changing who is signed in here. Then switch it on (Corporate plan), and choose whether to require it.
• Someone signing in for the first time joins the workspace you choose, with the role you choose.

Scheduled backups

Admins can make a backup token under Plan and your data > Scheduled backups. A script then downloads the whole workspace as a package every night into your own storage, with the command shown there. The package imports back exactly. Revoke a token to stop it.

No sign-in email?

Check spam, and that the address is the one you were invited with. Links expire after 15 minutes; request a new one.

“Couldn’t load your models”

The service is being updated; your models are safe. Try again in a minute.

Someone else changed the model

If someone else saved the model while you were editing, you will see a message and the latest version loads; re-apply your change.