Adoption & Governance

Lolly is an internal prototype in a closed pilot - a fast-moving behavioural experiment inside the enterprise, not a finished product. This page is the honest account of who it's for, how people are meant to adopt it, how we'll know if it's working, and who governs what it produces.

Most of this platform's documentation describes what Lolly can do. This page describes what Lolly is doing right now: running a pilot, gathering evidence, and trying to change a behaviour - how ordinary, non-designer colleagues get an on-brand file made.

Status

Lolly is a closed-pilot prototype. Treat it as one.

This framing is deliberate. Judged as a finished marvel, Lolly will disappoint. Judged as a pilot trying to prove a specific behavioural change - routine creative work, done safely, without a designer in the loop - it has a clear job and a clear way to measure whether it's doing it.

Who Lolly is for

Adoption succeeds or fails on the producer - the non-designer who has to make something on-brand and, today, either waits for a designer or does it off-brand in whatever tool they have. Everyone else in this table exists to make that person's path frictionless.

UserWho they areThe friction Lolly removesWhat they adopt
The producerMarketers, sales, events, ops, comms - non-designers who need finished, on-brand files"I need this now, I don't want to break the brand, and I don't want to wait for design"The app: pick a tool, fill in fields, get the asset
The brand owner / designerThe people who own how the brand looksRe-typing the same layout, policing off-brand output after the factAuthoring tools & the asset catalog - encoding the rules once
The developer / platform teamEngineers who automate and deployStoring binaries in Git, custom renderers, cloud image billsThe CLI, URL mode, MCP endpoints, self-hosting
The AI agentAutomated workflows that produce assetsToken-expensive, drifting, un-auditable image generationThe MCP tools - deterministic renders from parameters
IT & securityThe people accountable for data leaving the buildingColleagues uploading sensitive files to random web toolsOn-device utilities, air-gapped deployment, governance-as-data

The pilot's centre of gravity is the producer. If producers don't self-serve, nothing else matters - the developer integrations and the governance model are means to that end.

What onboarding looks like

The first 15 minutes (any producer)

  1. Open Lolly - the web app needs no install, no account, no sign-up. Nothing you type leaves your device.
  2. Pick a tool that matches what you need (an event tile, a quote card, a signature).
  3. Fill in the fields. No fonts, colours or spacing to decide - the tool already holds the brand rules.
  4. Get the file. Download it, copy a share link, or export a batch. Done.
  5. Save the session so you (or a teammate) can reopen and re-render it later.

Step 3 is the one that carries the pilot. A tool opens with the brand already on it, so the only thing left to decide is the words.

Deck Studio's first slide opened cold, already carrying the brand's type, colour and title layout before a single field is filled in

The measure of a good onboarding here is blunt: did they leave with a finished, on-brand file, without asking anyone?

The brand owner / designer

Onboarding is about handing over control of the rules, not the output:

  1. Define the asset catalog - logos, palettes, fonts as permanent IDs.
  2. Author the tools most in demand (start with the two or three asset types your team asks for weekly).
  3. Convert one high-visibility output and show the before/after side by side - this is the single most effective adoption lever.
  4. Set the guard-rails: lock what must never change, expose only what's meant to vary.

The Brand Studio's Catalogue step, where logos, images and fonts become permanent IDs a tool can call

The developer / platform team

  1. Run a tool from the CLI to see the same engine the app uses (lolly qr-code --url=… --output=…).
  2. Wire a render into a build step or an MCP endpoint.
  3. Decide the deployment shape: hosted PWA, self-hosted, or air-gapped runner pods.

The Dashboard's capability map is the inventory to scope that decision against: every part of the platform as its own card, grouped by what it makes and where it runs.

The Dashboard's What Lolly can do panel, the whole feature set laid out as grouped cards rather than a prose feature list

IT & security

  1. Confirm the data posture: no telemetry, no backend, nothing uploaded by default.
  2. Scope the pilot to a low-risk context while the security audit is still outstanding (see Status).
  3. Decide who owns governance - see Governance below.

Measuring adoption

We measure a behavioural change, not feature usage. The north-star is design-ticket deflection: routine creative requests that are now self-served and never reach the design queue at all.

SignalWhat it tells usType
ActivationShare of invited pilot users who render at least one real assetLeading
Self-serve rateAssets produced without a design ticketLeading
Time-to-assetBrief → finished file, in minutes not daysLeading
Tool coverageShare of routine asset types that have a matching toolLeading
Repeat useUsers who come back within a 30-day windowLeading
Design-ticket deflectionRoutine requests that never reach the design queueLagging / north-star
Story captureConcrete before/after cases collected from real usersQualitative

A leading signal moving without deflection following is a warning: people are trying Lolly but the work is still landing on a designer's desk. Deflection is the number that says the behaviour actually changed.

The 90-day pilot cycle

Adoption runs on a 90-day feedback loop. Each cycle:

  1. Weeks 1–2 - Onboard a cohort. Bring in one team, author the tools they most need, remove the obvious blockers.
  2. Weeks 3–10 - Use and observe. Watch the leading signals; collect stories; fix what's in the way.
  3. Weeks 11–12 - Review and re-aim. Read the deflection number, decide which tools and which next cohort come next.

The 90-day cycle is the cadence. It is not the goal - it's how often we re-check whether the goal is moving.

From cycle to deflection target

The goal the cycle serves is deflection, and it should ramp. The pilot targets 30% design-ticket deflection by month 6 - roughly one in three routine requests self-served away from the design queue.

MonthFocusTarget deflection
Month 1Onboard first cohort; author first toolsbaseline (~0%)
Month 2First self-serve wins~5%
Month 3End of first 90-day cycle; review~10%
Month 4Expand the tool catalog~18%
Month 5Onboard second cohort~25%
Month 6Target30%

30% is deliberately a pilot target, not an end state. It's the threshold that says the behaviour change is real and worth scaling - not a ceiling on what deflection could eventually reach.

Governance (when you want it)

Most people just make things - author their own tools in Layout Studio and ingest their own files into the catalogue, entirely in-app, with no git and no approval step. Governance is what you reach for when an organisation wants a shared, controlled catalog: an option that runs the rules the way engineering runs code - the rules are data, and changing them is a reviewable change.

Every part of Lolly as its own switch, so turning a whole category of tools off is one click, not a support ticket

The most common adoption set-back is not technical; it's framing and change management. Existing processes work today even when the output is off-brand, and "we can already make files" becomes the excuse not to migrate. The counter is to convert one high-visibility output well and let the before/after make the case. See the FAQ on adoption hurdles.

We need your story

Because Lolly collects no telemetry and phones nothing home, we genuinely do not know who runs it or how well it's working - and that's by design. The flip side is that the pilot depends on you telling us.

If you are piloting Lolly, the most valuable thing you can contribute is a concrete before/after: what you used to do, what you did with Lolly, how long it took, and where it fell short. That evidence - not more architecture - is what moves this from a promising prototype to something proven.

Honest limitations

To keep the framing straight, the things Lolly is not yet: