Build the model. Then see what holds.

01

What scheem is

A canvas you can edit, a set of checks that read that same canvas, and a register that is the only place a model-level verdict can come from.

scheem is for people who have to defend a business model: founders, strategy leads, and advisors. The nine-block canvas is the authoring view. Workshops, checks, scenarios, and exports all read that model. A change in one block is visible to the tools that depend on it.

Three kinds of help sit side by side, and they are not interchangeable.

• You write the model. Blocks, workshops, and model settings are yours. AI drafts are proposals. You apply them, edit them, or leave them.
• Checks report what they can see. Some are language-model readings. Some are repeatable arithmetic on numbers you entered. A check can show a contradiction, a thin block, or a shortfall. It cannot say the model holds.
• The register keeps the verdict. Only the validation register may say the model holds, holds conditionally, does not yet hold, or is refuted. Completeness, coherence, confidence, maturity, pattern fit, and an AI opinion are findings.

The path through the product is Build, Test, Validate, Scale, then Communicate. You can export at any point. An export is a frozen account of the model at that moment, not the end of the work. The method behind the verdict is written up on the methodology page.

02

Getting started

Open a canvas, name the venture, and write the brief the rest of the tools will read.

1. 01Create an account and open the workspace. You land on the canvas. Explorer includes one canvas. Paid plans remove that cap. Prices and the full comparison are on Pricing.
2. 02Start a canvas. Use New in Projects for a blank model. On Strategist and above you can also start from a template. You can draft a first version from a two-sentence brief, or from a company URL, and then apply the lines you want to keep.
3. 03Name the canvas from the name in the top bar.
4. 04Open Model settings (the gear) and write the idea brief. Add the market, company status, goal, and any constraints you already know: funding stage, capital runway, monthly burn, time to market, technology readiness, and risk appetite. There is also a sustainability mode, a currency, and a flywheel cycle time. An Investor tab holds team facts and use of funds. Those two are facts you type. They are never generated for you.
5. 05Put a line in the blocks, yourself or by applying a suggestion. Then open a block header when you want the workshop for that block.

Technology readiness uses the published 1–9 scale, from basic principles up to a system proven in operation. Time to market is in months. If time to market is longer than the runway you entered, the workspace can flag a bridge phase: an interim path while the main offer is still out of reach.

The header ring is a progress cue. Its levels are Sketch, Structured, Elaborated, and Stress-Tested. They track how much of the workspace you have filled and exercised. They do not say whether the model holds.

03

Build

Shape one connected model: nine blocks, optional workshops, and suggestions you choose to keep.

The nine blocks

The canvas is the Business Model Canvas. The right side is the offer and the customer. The left side is how you deliver it. Cost and revenue sit along the bottom. Switch between the classic layout and a balanced 3×3 grid from the view controls. Each item you enter can be traced. Separate items with a new line or a semicolon.

Key partnerships

Alliances, suppliers, and joint work you depend on.

Key activities

The work that has to happen for the offer to be delivered.

Key resources

People, assets, intellectual property, and money the model needs.

Value propositions

The offer, tied to a job a specific customer is trying to get done.

Customer relationships

How you win, keep, and grow those customers.

Channels

How the offer is found, evaluated, bought, delivered, and supported.

Customer segments

Who it is for, and who it is not for.

Cost structure

Fixed, variable, and semi-variable costs, and what drives them.

Revenue streams

Who pays, how, and how that connects to the offer.

A block, up close

To edit quickly, type in the block. To go deeper, click the block header. That opens the workshop. The lock on a header keeps a block out of auto-fill. The distill control on the same header looks for duplicate and orphan lines.

Workshops

Nine block workshops (W1–W9) are optional depth. A valid model does not require them. Each one is a structured editor for a single block: segments and jobs, the offer, channels, relationships, revenue, resources, activities, partnerships, and cost. Segment sizing can estimate TAM, SAM, and SOM per segment. The resource workshop can walk a VRIO reading. The revenue workshop includes pricing and unit economics.

To ask the workshop for help, use the actions on that block. Typical actions are Draft (a first set of items when the block is empty), Add one, Check, Research (a check with web sources and citations), and Wildcard (one contrarian alternative). Research and Wildcard often sit in a secondary menu. A suggestion appears for you to apply. It does not replace what you already wrote.

Six pair workshops (P1–P6) relate two blocks: offer and segment, revenue and segment, channels and segment, relationships and segment, partnerships and resources, activities and cost. Use them when the question is the link between the two sides, not the contents of one block.

Distill and intake

Distill is propose-only. The deterministic pass, available on the lower paid plans, finds restatements and lines that do not connect. The language-model pass is Pioneer. It suggests merges by meaning. Nothing is deleted until you accept a merge.

Auto-fill from a brief, and fill-from-URL, are paid. Both return a draft. Apply the lines you want, block by block. Locked blocks are left alone.

04

Test

Pressure-test the model you have. A clean check means the tools found nothing they are able to flag. It does not mean the model holds.

To run a Check

On the right rail, open Check. It is a launcher, not a single test. It opens three tools:

• Consistency. Reads across blocks for contradictions, tensions, and vulnerabilities. A tension is a trade-off you can acknowledge. A contradiction offers Suggest a fix. That button drafts a change for you to review. The canvas stays as it is until you apply the draft. Dismiss hides a card. It does not erase the underlying issue. An empty list of logic cards does not mean every issue is resolved, and it does not mean the model holds. Explorer can run consistency once a day. Paid plans are uncapped.
• Canvas Completeness. Finds empty or thin blocks. A block with two or fewer items is treated as thin. The tool offers suggestions you accept. It measures how filled the canvas is. It does not read assumptions, evidence, or the register. Running it requires a paid plan.
• Assumptions. Surfaces bets the model is already making, so you can track them. Framing and testing those bets is the Validate step.

Calculated checks

Where you have entered structured quantities, scheem can compare them with repeatable arithmetic: delivery hours against demand, and unit-economics figures such as margin and payback. The result shows the calculation. It is kept apart from AI suggestions, and it is labelled Calculated from your inputs. In the Check view the same result appears as a capability finding or a declared-quantity shortfall. The arithmetic does not set the model verdict.

Structural audit, lenses, and compliance

Structural audit is a rule check of domain constraints. It is computed, not prompted, and it is on Strategist and above. Findings show up with tensions. A rule-based result is still a finding.

Lenses do two jobs. The rail’s Lenses control sets the analytical mode for other tools: Business, Platform, or ESG. Separate scans read the canvas through Technology, Sustainability, Regulatory, Competitive, and Financial Health. Scan findings are observations with a severity. Applying one adds a proposal to a block. You still accept it. Compliance checks obligations the model appears to trigger, by region and standard, and can propose a small set of items per block after a preview.

Live indicators on the canvas — a nudge when a known pattern seems to fit, a tension marker, a tests count — are signals. Pattern fit is a resemblance to a reference mechanism. It never raises the verdict. Novel mechanisms are allowed to look like nothing in the catalogue.

05

Validate

Say what has to be true, decide how you would notice if it were false, and let reviewed evidence move the register.

Assumptions and test cards

Reading the register is available on Explorer. Writing it — framing a proposition and moving it through the lifecycle — is Strategist and above.

To frame an assumption, open it and choose Frame as testable. Write one falsifiable sentence, how you will test it, and the threshold in your own words. Tick the deal-breaker box when a failed test would sink the model. You can also record how critical the bet is, and how sure you are today. Those numbers are your prior. They are not evidence.

Then open a test. The hypothesis starts from the proposition. Edit it before you run anything. Record the outcome yourself when you have one: passed or failed. The register’s words for what a test did to a proposition are supported, weakened, and refuted. A recommendation may say what the outcomes so far read as. You record the status. The product does not record it for you.

Statuses you will see include unframed, untested, testing, supported, weakened, refuted, superseded, and archived. An anchor marked drifted means the canvas line the assumption pointed at has changed. The assumption itself has not moved.

Evidence

The evidence library holds sources you attach to a proposition: files and links. An AI appraisal, on Pioneer, proposes how one source bears on one proposition, and in which direction. The row lands as proposed. It counts for nothing until you press Accept. An upload with no accepted appraisal is a file, not evidence. You can also link a source yourself, without the appraisal.

AI assumption decomposition, also Pioneer, drafts a split when one assumption is really several. The split stays unapplied until you confirm it. Multi-source demand research, Pioneer, runs from a framed market assumption (Research this assumption). You review the sources. They do not move the register until you accept them. Strategist can open a demand assessment on the same kind of assumption.

06

Scale

Explore other shapes of the same model, and plan a move only as far as the conditions you stated.

Scenarios

To generate a scenario, open Scenarios on the rail. The three types are Lean, Premium, and Partnership. Each is a full alternative model. Value proposition and customer segment are held constant by policy on a Lean rewrite. A scenario is a branch. It does not replace the canvas until you apply it. Explorer can generate one scenario a week. Paid plans are uncapped.

One-click pivot, on Strategist and above, recasts the canvas onto another template shape. Review the result before you keep it. Strategic options from the causal tools are proposals of direction, not a second verdict.

Strategy frameworks

The strategy suite is on Strategist and above. Each tool is a reading of the same saved model. Frameworks on the rail opens SWOT with Porter’s five forces, the risk matrix, the customer journey, and go-to-market. Strategy on the rail opens market sizing, unit economics, pricing intelligence, the growth model, and the executive brief. Scenarios adds a stress test that simulates lean, base, and premium variants of your numbers. Export Studio can run the suite’s five main readings for you when a document needs them.

Framework-shaped thinking also lives inside the workshops: the Fit Canvas in the Value Propositions workshop, jobs-to-be-done in Customer Segments, and VRIO scoring in Key Resources.

Advanced deterministic readings are Pioneer: flywheel loops in the causal story, lock-in, vulnerability, and multi-round causal chains. The causal story itself — how the blocks reinforce each other — is Strategist and above. It is a narrative of the model. It is not a verdict, and a document must not talk as if the chain “holds” unless the publication gate is open.

Trajectory

Trajectory is a Pioneer tool. It takes assumptions you state up front, runs conditional rounds, and returns a qualified roadmap: if this is true, then that follows. It is a conditional plan. It is not a forecast. The module is behind a kill switch, TRAJECTORY_ENABLED. When the switch is off, the rail omits Trajectory and the endpoints answer 404, including for Pioneer. Internal early-access accounts can still see it. When you do see it, it opens as a full-workspace overlay, and only while you are on the canvas rather than inside a scenario.

07

Communicate

Take a frozen account of the model into the room. The file names its version. It does not claim more than the register allows.

Export Studio

Open Export from the header, or from the command palette. It is not a button on the right rail. Export Studio composes eight document types from a catalogue of pieces: Strategic Dossier, Executive Pitch, Model Story (a web presentation), Evidence & Assumptions Report, Strategic One-Pager, Strategy Memo, Investor Pitch, and Workshop Pack. You choose the pieces. Some pieces need facts only you can enter, such as the team slide and use of funds.

PDF is on Strategist and above. PPTX, DOCX, ODP, and ODT are on Pioneer, for pieces that support that format. The model story is a web presentation. Share cards for a live canvas are SVG. A dossier can include the narrative, the canvas, risks, an assumption ledger, an evidence ledger, and a roadmap. Where the self-assessment crosses its threshold, the dossier can show declared fragility and a Red Blox warning. Those are warnings about the inputs. They are not a verdict.

Each export can carry a fingerprint tying it to the model version it was rendered from. Re-opening a stored output replays that frozen bundle. It does not blend in later edits.

Sharing, collaborators, and history

To share a live canvas, open Share and copy the read-only link. There is a grid view and a presentation view. Minting that link is Strategist and above. Email invites to collaborate on the canvas stay available on Explorer. Invited people work on the live model with you.

A frozen share is different. From a stored output you can mint a link of the form /shared/output/…. Readers see that snapshot, including a PDF replay, and later edits to the canvas do not change it.

Block history, with restore, is Strategist and above. Use it when you want a block back, not when you want a public snapshot.

Team seats, shared workspaces, roles, and single sign-on are not available yet. They are planned, not a plan you can buy.

08

Interface

The canvas is the centre. Tools hang off it, and the rail is a registry: what you see is the list below, not a custom set of buttons.

Around the canvas

• Header. Canvas name, save state, the progress ring, Model settings, Share, and Export.
• Left side. Projects (new, switch, rename, delete, restore), templates on paid plans, and account.
• Canvas. Classic or 3×3. Click a header for the workshop. Lock a block to keep auto-fill off it.
• Next step. A persistent bar suggests one next action from what is already on the canvas. It is a prompt, not a score of truth.
• Command palette. Press Ctrl+K (or ⌘+K on a Mac). Search jumps to tools, including ones that are not on the rail, such as Narrative and Export.

The right rail

The rail is two groups. Mode is Lenses. Analyze is Check, Scenarios, Strategy, Trajectory, Frameworks, and Compliance. Check opens Canvas Completeness, Consistency, and Assumptions. Frameworks opens SWOT, Risk, Journey, and GTM. Narrative (the causal story) opens from the palette. A tool can stay disabled until enough blocks have content. The tooltip says how many. Scenarios and Trajectory are canvas-scope only. Inside a scenario they stay disabled.

Opening a tool declares an effect, when that hint is switched on for your session: read-only, will propose changes for review, or creates a new scenario. Check and Frameworks are launchers, so they do not change the model by themselves. Some readings, including the causal story and Strategy, can open from a narrow panel into a wide reading. Close the reading to return to the narrow panel. Your canvas is still there.

10

Concepts and frameworks

Short definitions for words the workspace uses. If a word here and a verdict word ever seem to collide, the verdict ladder wins.

The model

Business model canvas

The nine-block description of how an organisation creates, delivers, and captures value. In scheem it is the authoring view of a larger model that also holds workshops, assumptions, and evidence.

Frontstage and backstage

Offer, segments, channels, relationships, and revenue face the customer. Resources, activities, partnerships, and cost are the operating side.

Jobs to be done

A customer hires an offer to make progress: a functional job, an emotional one, a social one. Pains and gains are the obstacles and the outcomes around that job.

Value proposition canvas

The fit between the offer and one segment’s jobs, pains, and gains. A framework reading. Not a verdict.

VRIO

Valuable, rare, inimitable, organised. A way to read a resource. A high reading is a hypothesis about advantage, not proof of it.

TAM, SAM, SOM

Total, serviceable, and obtainable market. Segment sizing can estimate them. The estimate is an input to a bet, not evidence the bet is won.

Checks and plans

Finding

A sourced observation: a contradiction, a thin block, a shortfall, a lens note, a pattern resemblance. It asks for attention or a test.

Proposal

A suggested change waiting for you. AI output, distill merges, compliance items, and scenario branches are proposals until you apply them.

Deal-breaker

An assumption you marked as fatal. If it fails, the model verdict can fall to refuted. With none declared, the ladder stops at holds conditionally.

Coverage

How much of the register the verdict is actually about. The rung “does not yet hold” is not used when coverage is absent.

Flywheel

A reinforcing loop across blocks. Detecting one describes a structure. It does not show the loop is working in the market.

Bridge phase

An interim path when time to market is longer than runway. A planning device, not a prediction.

Resolution posture

A way to answer a structural tension: Internalized, Externalized, Asset-Lease, or Civic-Hybrid. The block lines that follow are there for you to review.

Technology readiness

The published 1–9 scale. A 9 means the system is proven in the environment it is meant to work in. The scale’s authors adjudicate the rung. scheem does not.

SWOT and five forces

Strengths, weaknesses, opportunities, threats. Porter’s forces: rivalry, entrants, substitutes, supplier power, buyer power. Position readings.

Fingerprint

The stamp on a frozen export that ties it to the model version it was rendered from.

Progress ring

Sketch, Structured, Elaborated, Stress-Tested. How far the workspace has been worked. Never a substitute for the ladder.

For the rules that decide when a document may speak in verdict language, read the methodology. For what your plan includes, read Pricing.