Manifesto

MANIFESTO V1

We're not here to make eLearning better. We're here to make it matter.

Somewhere along the way, corporate training became a thing that gets produced instead of a thing that teaches. A deliverable. A checkbox. A 47-slide module nobody finishes, designed by a committee that never met the learner, approved by stakeholders who'll never take it, deployed into an LMS where it goes to die.

That's a quiet catastrophe. Billions of dollars and millions of hours, every year, training people in ways that don't work. While the world demands they learn faster than ever.

We built Fabella because we got tired of waiting for the industry to admit it.

Here's what we believe.

Six-person production pipelines are over. They just don't know it yet.

For 30 years, training has been produced by relay race. The SME briefs the instructional designer. The ID briefs the writer. The writer briefs the developer. Stakeholders review. Brand reviews. Legal reviews (who insists on rewording every imperative verb). Six months later a course exists, and the thing it was teaching has already changed.

That model is collapsing. Structurally, not gradually.

The unit of production in every creative field is shrinking from team to individual, and training is next. Figma did it to design. AI tooling is doing it to code. Consumer cameras did it to film. Every time, the gatekeepers warned quality would collapse. Every time, the work got better, because the people who cared most were finally the people in control.

Instructional designers have felt this worse than anyone. You can spot a bad course in 30 seconds. You know what the intervention should be. And you spend your week formatting slides, chasing approvals, and defending decisions against people who've never sat with a learner.

That's the waste. The smartest judgment in the field, spent on the wrong tasks.

The prosumer-operator is what comes next. One person, ideally the one who already knows what good looks like, with the leverage of a team and none of the relay race. Real authorship over the work. The loop closed end to end. The committee no longer the chokepoint.

The industry's reflex will be to scale outward: more vendors, more outsourcing, more agencies producing more of the same. We're betting the other way. One operator with leverage beats six people with handoffs.

Training is an experience that changes what someone can do tomorrow.

The "content" framing is the original sin. It puts training in the same conceptual bucket as marketing collateral or a company newsletter. Something to be produced, distributed, consumed, and forgotten.

But training only matters if behaviour changes. And behaviour doesn't change because someone watched a video at 1.5x speed while answering Slack messages and refilling their water bottle.

Training is experience. Moments designed to do something to the learner. Surprise them. Challenge them. Make them decide. Let them feel the consequences of getting it wrong in a place where wrong is safe.

If your training could be replaced by a PDF, it was never training. It was filing.

Every learner is making a trade. You owe them the outcome.

A learner does not "consume" your training. They trade for it. They hand over an hour of their working life. An hour they're never getting back. An hour they could have spent doing the actual job, being with their kids, or learning something useful on their own. In exchange for an outcome you promised them.

Treat training like a product, because that's what it is. The learner is the counterparty. The deal is their attention for your transformation. Honor the deal or don't ask for the attention.

If you make them sit through a course they'll forget by lunch, you didn't fail to entertain them. You broke the trade. Quietly, politely, with a completion certificate to make it official.

We refuse to be polite about this anymore. Boring training is a breach of contract. Engagement isn't the cherry on top. It's the minimum acceptable threshold before you have the right to ask for someone's hour.

The LMS won.
The learner lost.

The training industry spent two decades optimising for the buyer. Procurement won. IT won. Compliance won. The learner, the one human being whose change in behaviour is the entire point, came dead last.

This is why the average corporate course feels like it was designed for a database rather than a person. Because it was. It was designed to be tracked, reported on, audited, and reconciled. Whether someone actually experienced the thing was an afterthought, somewhere between the export to xAPI and the quarterly compliance report.

A tool should start from the learner and work backward. If the audit trail has to come second, the audit trail comes second.

AI doesn't replace the creator. It removes the parts that were never the creator's job in the first place.

There's a panic in our industry that AI is coming for the instructional designer. It isn't. What's coming for the instructional designer is the busywork. The formatting. The asset wrangling. The SCORM packaging. The seventh round of small revisions. The conversion of one good idea into 9 slightly different file formats.

That work was never the value. The thinking was the value. The decision about what matters, who needs to learn it, what success looks like, where the friction points are, what story will land. That's the creator's job, and it always will be.

A tool should handle everything that isn't the thinking, so the creator gets to spend their time on the part that's actually theirs.

This is AI finally doing what was promised: removing the grind, returning the craft.

Blueprint-first authoring tools are losing. Conversation-first ones are winning. This is not reversible.

Real creators don't sit down and execute a blueprint. They have a feeling about what the thing should be. They start gesturing at it. They bump into the material. They discover what they meant as they go. This is how every good film, novel, product, and lesson plan has ever been made. The artefact emerges from the conversation between the creator and the medium.

Traditional authoring tools refuse to admit this. They demand you arrive with the entire structure pre-decided, then assemble it like flat-pack furniture. Predictably, what comes out the other end looks like flat-pack furniture.

Vibe Authoring is what happens when you build a tool that meets creators where they actually are. You talk. You try. You see. You adjust. You ship. The "vibe" isn't fuzzy. It's the actual texture of creative work, finally taken seriously by software.

Bolting AI onto a 2010s editor doesn't make it 2026.

Every legacy authoring tool is right now bolting an AI chatbot onto a 2010s editor and calling it the future. It isn't. It's a 2010s editor with a chatbot stapled to it.

AI-native authoring is structural. You don't get there by adding a button.

Domain-native. A general-purpose model that writes "training-shaped content" is not the same as a system that actually understands instructional design: branching logic, completion tracking, assessment integrity, the way a learning objective ladders into a scenario. The output has to work inside L&D. Looking like training isn't the same as being training.

A system of specialists. One model trying to be instructional designer, quality reviewer, editor, and format chooser all at once produces mush. Real craft comes from specialised intelligences, each excellent at its part, orchestrated together. That's an architecture decision, made on day one. You don't retrofit it.

Format intelligence. A roleplay. A branching scenario. A lightning round. An audio-first chapter. The tool should pick the right format for what you're trying to teach and build it in the editor. Not hand you a wall of text describing what such a thing might look like.

Closed loop. The conversation is the authoring session. Not a sidebar. Not a copilot panel. Not "generate, then paste, then edit." You say "add a roleplay to this chapter" and it appears in the chapter. The friction is gone.

You can feel this in the tools you're already using. The AI panel that appeared without warning. The pricing tier that went up to cover it. The chat sidebar that helpfully rewrites your scenario into something blander than what you started with. The "AI assistant" sitting next to the same slide-and-timeline editor you've been clicking through since 2014.

It isn't your imagination. You're being charged more to use a worse version of the tool you already had. That's what bolting AI on top of legacy architecture actually feels like from inside.

Adding a button doesn't change the foundations. Adding a panel doesn't either. You can't retrofit a product that was architected for a different decade — no matter how much you raise the subscription.

Tooling fragmentation is not a feature of the industry. It's an unpaid bill.

You author in one place. You manage assets in another. You publish to a third. You distribute through a fourth. You analyse in a fifth. Every handoff costs time, costs fidelity, costs the original idea a little bit of itself.

The industry calls this an "ecosystem." We call it a tax on creative momentum. Every tool you have to switch into is a moment where the thing you were trying to make slips a little further out of your hands.

A single environment, from intent to publication to insight, isn't a "nice integration story." It's a precondition for serious creative work. Anything less asks creators to do their best thinking in a wind tunnel.

"Engagement metrics" aren't engagement.

Completion rates measure compliance. Time-on-page measures patience. Quiz scores measure short-term retention of trivia. None of these measure whether the person who took the training can do something tomorrow they couldn't do yesterday.

Optimising dashboards that measure the wrong thing more precisely isn't progress. The metrics that matter are the ones the business actually cares about: does the work get done better? Do mistakes go down? Do new hires get up to speed faster?

If your training looks good in a quarterly report but doesn't change what happens on Monday morning, the quarterly report is lying to you.

This is the future we're building toward:

Training made by the people closest to the work, instead of three handoffs away.
‍
A great idea becoming a great learning experience in an afternoon, instead of a quarter.
‍
Engagement as the floor, not the ceiling.
‍
AI doing the grunt work. Humans doing the thinking. (Yes, in that order.)
‍
Corporate training that people actually want to take, because someone actually wanted to make it.

That future isn't science fiction. It's a product decision. And we're making it.

You're early. We like that.