What LEVR sells

LEVR sells a company its own strategy, encoded and enforced.

The product has two components. The Reference is the company's encoded strategy: the customers it answers to, the claim only it can make, its voice, and the rules that follow from those decisions. The Governance Layer makes the Reference enforceable — work produced in any connected AI tool is checked against the strategy before it goes out.

The commercial arc is one engagement delivered in three phases: a facilitator-led positioning workshop that produces the Reference, an agentic build that turns the Reference into finished go-to-market work, and a recurring activation subscription that keeps the strategy live and governs the work produced against it.

Delivery is managed by design. Every engagement is installed by an accredited facilitator and supported by an assigned LEVR technical integrator. There is no self-serve tier today.

Product architecture: one engagement, three phases

Phase 1 — Intelligence: the LEVR Positioning Workshop

A facilitator-led, single in-person day (09:00–15:40, seven sessions) with four to seven brand leaders. There is no client preparation burden. A scan of 100 or more competitors is pulled the same morning. The room is recorded continuously, and the strategy dashboard assembles from the client's own transcript during the day.

The day produces two decisions the client owns: the customers the strategy answers to, and the strategic claim only their brand can make. The strategy dashboard URL is delivered the same evening.

Status: proven in delivery. Currently moving from a terminal-driven build to the cloud platform.

Phase 2 — LEVR Flywheel: the go-to-market build

Within 48 hours of the workshop, a seven-to-nine-hour agentic build produces a go-to-market web experience: landing pages, ad copy, email sequences, avatar and voice-of-customer research, battle cards, sales one-pagers, lead magnets, video scripts, style guides, and press packages. The deliverable is two linked dashboards — Strategy (12 sections) and GTM Hub (22+ sections).

Material product change, 10 August 2026. The asset-dump model was replaced with instruction-first strategy cards: a visual, a single key sentence, playable audio, on-demand example generation, and a voice interface for interrogating the card. The reasoning was operational — a large pile of work that is 90% right creates review burden rather than removing it. The instruction is now the product; assets are generated on demand against it.

Status: proven in delivery; being decomposed into the platform.

Phase 3 — Activation: the recurring subscription

The subscription keeps the Reference current and puts it in front of everyone who produces work for the brand. It comprises the client dashboard, the free MCP connector, virtual employees, the agent harness, weekly market re-scanning, sub-accounts, the voice companion, and site-editor connectors. These are described individually in the next section.

Solution set and status

Status labels are used consistently: Live means running in production with at least one client; Shipping means in active client rollout; Building means under development and not yet client-facing.

Internal tooling not sold to clients: the facilitator console, the workshop dashboard, and the integrator console.

Value propositions and problems solved

The needs LEVR is bought against

• Executive alignment on differentiation. Ask five executives what makes the company different and you get five answers. The workshop forces two decisions and encodes them.
• Control of AI output that is already happening. Staff and vendors are already using AI on brand work. Buyers describe this purchase as buying guardrails.
• Finished, on-strategy go-to-market work without agency cycles.
• Per-location production that cannot be staffed. A 90- or 260-location brand cannot hire a marketer per site.
• Strategy that does not rot. Weekly re-research and quarterly re-calibration keep the Reference current.
• Reach every employee and vendor without tool migration. The free connector works inside the AI tools people already use.

Problems solved

Differentiation and competitive landscape

LEVR's stated position: Writer grounds AI in what a company has already written. Adobe, Frontify, and Bynder govern assets a company already has. LEVR generates what a company should say and has not yet decided.

Five structural positions

• Outside-in, not inside-out. Strategy is derived from the competitive field and the customer, not from an internal brand audit.
• Governed by design. The Reference pushes back on off-strategy work rather than passively storing brand assets.
• Installed by a trusted human facilitator. Distribution runs through accredited facilitators, not a self-serve funnel.
• A data company, not a software company. The compounding asset is the encoded strategy graph and the evaluation rubrics, not the UI.
• Positioning to production in one arc. No handoff between the strategy vendor and the execution vendor.

Competitive map by layer

Competitors by solution

The realistic near-term alternative for a target buyer is not a competing product. It is an agency retainer, or a consultant plus ChatGPT, or internal headcount.

ICP, buying unit and users by role

The facilitator is the structural customer

LEVR's go-to-market runs through accredited facilitators who sell, run the room, and own the client relationship. The facilitator archetype is a former CEO or agency principal with an existing network. Three sub-types are tracked, each with a different activation profile:

End-client ICP

Buying unit. The CEO is both champion and economic buyer. The executive team acts as the in-room jury. The facilitator carries the trusted relationship. Typical shape is a two-call close with workshop-to-live in days. Selling is organised around trigger events: roll-up integration, repositioning, a new executive team, a launch, agency consolidation, expansion, or visibly inconsistent messaging.

Disqualifiers. Single location; no budget authority; buyer wants a cheap point tool; organisation is AI-hostile; pre-brand startup.

Users by role

Roles as built in the platform today: facilitator, integrator, admin. Loops roles: admin, integrator, reviewer. Authorisation is deny-by-default and roles are never client-settable.

Packages and pricing structure

The pricing model below was set in August 2026 and supersedes all earlier structures.

Structure notes

• The usage layer is not a revenue lever. It exists to cap runaway model spend. Most customers are not expected to reach a limit.
• Payment terms. The workshop is billed 50% at signing and 50% at delivery.
• Seats. There are no per-tenant seat caps. Connectors are free and unlimited by policy. Base includes at least one strategy-admin seat; Enterprise adds more, with exact counts still being set. Harness seats are per-person and uncapped.
• Anti-lock-in. Clients can cancel in any month. The Reference and all assets are fully exportable, permanently.

Add-ons

Facilitator economics

The facilitator keeps the markup on the workshop. In the reference example the facilitator sells the workshop at $25,000 and LEVR retains $10,000. Facilitators take no markup on the platform fee; they receive a percentage of it plus a renewal bonus of roughly 10%.

Onboarding and engagement state machine

Onboarding flow

1. Engagement created. The facilitator lands the client and completes a three-step wizard: client details and logo → confirm the seven-session template, facilitator of record, and auto-assigned integrator → upload pre-work. This creates the engagement record and provisions the run container.
2. Prep brief auto-built. Research agents run against the client's website and supplied materials to produce the facilitator's preparation brief.
3. Client voice briefing. The client receives a voice-agent link and can ask anything. Every question enriches the briefing, and brand rules are captured up front.
4. Integrator notified at creation and has full visibility from that point.
5. In the room. The competitor scan runs the same morning. Recording is continuous. After roughly 90–120 minutes of transcript, orchestration begins building the strategy chapter by chapter.
6. After the day. End-of-day submission reprocesses the full transcript; the facilitator presents the strategy; the decisions taken are encoded as governance rules.

Engagement state machine

Internal handoff targets

These are internal operating targets, not contractual service levels. Integrator acknowledgement within two business hours of engagement creation; strategy and dashboard within 48 hours; joint QA before the reveal. Ongoing: acknowledgement within four business hours, P1 same day.

Provisioning a new knowledge-base client runs through a scripted routine that generates tokens, detects source formats, and produces a partner-ready handoff document. Adding a new virtual-employee tenant is one database row and one scheduled job entry.

Managed services and support model

Delivery is 100% managed

Every engagement includes an assigned technical integrator. There is no self-serve tier, so managed-service attachment is structurally 100% rather than a sold upsell. Services included or available:

• Technical integrator: build, deploy, rescue, activation, and facilitator accreditation assessment.
• The full agentic Flywheel build.
• Weekly Pulse market re-scanning.
• Virtual-employee production, targeting at least five approval-ready moves per week.
• Agentic services menu: web development, call-centre AI, paid-media management, analytics.
• Facilitator enablement: five-day training, a four-workshop accreditation ramp, and a co-delivered first build.
• Quarterly re-calibration, included in the subscription. This is the intended retention mechanism.
• 90-day source-of-truth handoff.

Support model

There is no help desk or ticketing system. Support runs on a two-tier human model: the facilitator owns the relationship and escalates but never fixes; the technical integrator resolves. Targets are four business hours to acknowledge and same-day for P1 — internal targets, not contractual commitments.

Triage is driven by console alerting rather than inbound tickets. Stale heartbeats and failed events raise de-duplicated, auto-clearing alerts into Discord, iMessage, and Slack. In-product paths include an explicit escalate-to-integrator action, a verbatim feedback log, the publish PIN gate, and site-editor approver routing. Governance questions can be escalated by voice.

Capacity target is 20 concurrent runs per integrator. No support-usage statistics exist yet.

Quality gates

Every build passes three tiers before it reaches a client. The same gate applies to each weekly Pulse diff.

Changes to the strategy, the strategic claim, the customer decision, or the knowledge base are always human-gated. There is no automatic path from a model to a canon change.

Product roadmap

v0 — first non-technical end-to-end run (now)

The definition of done for v0 is stated precisely in the platform specification: one non-technical facilitator runs one real workshop end to end without opening a terminal.

v1 — scale the execution plane and the channel

v2 — enterprise readiness and measurement

Known limitations and constraints

This section is deliberately complete rather than favourable. It is the same list LEVR maintains internally.

Commercial maturity

• No cohort has reached renewal. LEVR has been selling since Q2 2026. Net and gross revenue retention are therefore not measurable. An internal model assumes roughly 108% NRR from a 2%-per-month churn assumption plus expansion; that figure is explicitly unvalidated. GRR is not modelled.
• Historical ACV by vertical does not exist yet. Roughly three months of selling. What exists is list structure, shown in section 07.
• Subscription attach rate is unmeasured. Bear, base, and bull cases are modelled at 40%, 65%, and 80%; there is no real-world data behind them yet. This is the single most consequential unmeasured assumption in the model.
• Service levels are internal operating targets, not contractual commitments; none are published to clients yet.

Measurement

• Product analytics are not yet instrumented. There is no funnel, retention, or feature-usage reporting today.
• What does exist is comprehensive append-only audit and event data across every surface — audit logs, run events, feedback logs, decisions, publications, and AI-visibility time series — all queryable. The reporting layer over it is a named 12-month roadmap item.

Product limitations confirmed today

• The site-editor connector can edit existing files but cannot create new ones.
• One client's approval emails are routed through an interim gate while the transactional email domain is in test mode.
• Social publishing for one client currently runs through a white-label account; that client's own platform OAuth is pending.
• Vector search is deferred; retrieval is structured rather than embedding-based.
• The internal signal bus carries little beyond the Pulse routine today.
• One third-party messaging API is not capable enough to support the planned messaging agents.
• There is no bring-your-own-key model; execution runs on pooled LEVR credentials.
• Transcription is batched, never real-time. This is a locked design decision, and it sets a floor on voice-interface latency during the workshop.
• The MCP connector cannot force a host LLM to obey. The honest claim is an approval gate plus an audit trail, not universal enforcement inside third-party tools.

Compliance gaps

• No SOC 2, ISO certification, executed DPA template, or published sub-processor list yet. SOC 2 Type I is estimated at 6–9 months and is on the roadmap.
• No formal data retention TTLs published, though client-facing retention and export commitments are documented (see the technology document, section 07).
• Consent enforcement for end-customer contact data is deliberately left with the client.

Registered operational risks and their mitigations