There is a thesis behind Forge that we have not stated plainly enough. So let me state it now.

If you accept the premise that an AI Performance Officer generates coaching on the fly, adapted to each learner’s role, level, language, and moment, then you also have to accept the consequence: the way L&D teams are organised has to change with it. The org chart that worked when L&D was a content factory does not work when L&D is the governor of a content platform.

This is the bet Forge is making. It is the most uncomfortable part of our pitch and the most important one.

The factory model worked. It is also the ceiling.

For a decade, the most common L&D operating model has been a production line. The team is structured around producing courses.

A typical mid-size enterprise L&D function has:

  • Instructional designers who write courses
  • Learning architects who review them
  • Subject-matter experts who validate the content
  • Compliance and legal who sign off on regulated material
  • LMS administrators who deploy and assign
  • Learning ops who run reporting

This structure produces predictable output. A new compliance module ships in 6-8 weeks. A sales onboarding curriculum ships in a quarter. The team plans the year, executes the plan, measures completion, reports up.

That model is not broken. For a long time it was the only way to get learning to scale. We built Rise Up specifically to make this model faster, better authoring tools, better learner experience, better analytics. Cornerstone, Docebo, 360Learning, Workday Learning all do the same thing. Tens of millions of learners go through this stack every year.

But the production line is also the ceiling. The factory is what makes learning slow, generic, and disconnected from the moment when the skill actually applies.

What an AI Performance Officer changes

When coaching is generated on the fly from policy + playbook + role + moment + memory, four things change at once:

  1. Production cycles compress from months to seconds. A new pricing playbook reaches every rep the same morning the CRO publishes it. There is no 6-week production lag.

  2. Adaptation is per-learner, not per-cohort. The same coach delivers a different curriculum to a senior rep than to a new hire, because of the positioning test, the personal memory, the live context. No mass production of role-specific variants.

  3. Content variety becomes effectively free. A flashcard, a podcast, a quiz, a role play, a slide deck, a video, the coach picks the right format for the moment. The L&D team does not pre-author each format.

  4. The unit of work shifts from the course to the policy. What needs to be authoritative and curated is not the lesson, it is the source material the AI generates from.

These four changes break the existing operating model. Not in the sense that they make the existing team obsolete, but in the sense that the same team has to do a different job.

The new operating model: from factory to governance

We see the new L&D function organising around five new functions, not the old six:

1. Substrate curators. The new L&D priority is the policy library, the playbook library, the brand voice, the standards, the regulations. This is the substrate the AI generates from. The L&D team’s job is to make sure this substrate is current, accurate, organised, and tagged for the AI to use. This is curation work, not authoring work. It looks more like content strategy than instructional design.

2. Coach configurators. Each Forge coach, Sales, Compliance, Customer Service, Onboarding, Manager, has to be configured. Loaded with the right substrate. Aligned to the right KPIs. Connected to the right systems. This is product configuration work, closer to a sales operations role than to a course author role.

3. Quality samplers. When content is generated on the fly, you cannot review every output. You sample. You set quality thresholds. You spot-check. You build feedback loops. This is closer to model evaluation work than to content review. It requires data fluency.

4. Outcome owners. L&D shifts from reporting completion rates to reporting business outcomes, win rate uplift on coached deals, audit findings reduction, ramp time compression, NRR movement on coached accounts. This is a partnership role with the line of business owners. It looks more like RevOps or a strategy function than a traditional L&D analyst.

5. Platform operators. Someone has to keep the platform running, keep integrations connected, keep policies updated as regulations change, keep coach configurations aligned with org changes. This is L&D ops, but oriented around continuous platform operation rather than batch course production.

Five functions. Different from the old six. Different work, different skills, different rituals.

What this means concretely

Here is what we see, and predict, in the customer organisations adopting Forge:

  • The instructional design team shrinks. Some designers move into curation roles. Some move into coach configuration. Some leave the function. This is not a cost-cutting move; it is a re-orientation.

  • A “Head of AI Coaching” role appears. Reporting to the CHRO or the CLO. Owns the platform strategy, the coach catalogue, the KPIs the L&D function commits to.

  • Compliance and legal stay close, but their work changes. Less reviewing each course; more validating the policy substrate and the prompts. Less waterfall sign-off; more continuous policy governance.

  • The L&D budget mix shifts. Less spent on per-course production. More spent on platform operations and substrate curation. The total budget often stays flat, the productivity unlock comes from reach and outcomes, not from headcount cuts.

  • Reporting changes from completion to performance. The metric that matters is not “what % completed compliance training”, it is “what % of new hires hit ramp at month 3” or “what was the win rate uplift on coached deals last quarter”.

This is not a hypothetical. It is what we are watching emerge in the first cohort of organisations rolling out AI Performance Officers, including in our own design partner customers.

Why this is hard, and why we are honest about it

Most enterprise software pitches understate the change required to deploy. Vendors say “drop-in solution” because that is what buyers want to hear. We are choosing not to do that.

The L&D teams that ship Forge will look different in twelve months than they did when they signed the contract. Some of that change will feel uncomfortable. Roles will be reshaped. Skills will need to be acquired. New rituals will need to be built. Some people will not want to make the trip.

We do not pretend otherwise. The teams that pretend the operating model does not need to change will roll out an AI Performance Officer, fail to govern the substrate properly, see quality drift, blame the AI, and pull back to the factory model.

The teams that get this right, that lean into the new operating model, will compound. Their coaches get better every quarter. Their substrate gets richer. Their business outcomes show up in the numbers. They become the L&D function that the rest of the company actually wants to work with.

What L&D leaders should do this quarter

If you are a CLO, a Head of L&D, or a CHRO who owns this function, here is what we recommend:

  1. Map your current operating model. List the roles, the rituals, the cycles, the metrics. Be honest about what is actually working and what is theatre.

  2. Identify your substrate. What policies, playbooks, standards, brand voice documents would an AI Performance Officer need to generate good coaching from? Where do they live today? Who owns them? Who keeps them current?

  3. Pilot one coach with a real outcome attached. Not a generic “AI coaching pilot.” A specific coach (Sales, Compliance, Onboarding, Customer Service) with a specific business KPI it will move (win rate, audit findings, ramp time, CSAT) and a specific timeframe (90 days).

  4. Use the pilot to surface the org gaps. Which roles do you need to staff differently? Which rituals do you need to retire? Which metrics do you need to start reporting on?

  5. Plan the operating model change as a 12-month transformation. Not a 90-day pilot. The pilot is the trigger; the transformation is the work.

This is the bet. The L&D teams that take it will be the ones whose business actually performs.

, Arnaud