30/60/90-day onboarding plans that end in a readiness decision, not a checklist
Ascend onboarding turns each job role into one reviewed ramp template, starts a copy of it for every new hire or role change, and marks a person ready only when every gate has closed on evidence: a manager's sign-off, reviewed work, or a passed check.
Write the ramp once per job role, not once per hire.
Forty store associates should not get forty different onboarding plans. In Ascend, the plan belongs to the job role. An admin builds the role first (the skills it needs and the level each one is needed at), then its ramp template beside it on the same screen.
- An AI drafting agent can fill the first version from the job role's required skills, your published training programs and the documents retrieved for that role. A person publishes it.
- Every gate the agent proposes carries a sentence explaining why it is a gate, so a reviewer can disagree with a specific claim.
- 30/60/90 is the default shape, not a rule. Each template sets its own ramp length and phase boundaries.
- Items are tasks, training or skill focus, and each has an owner. Setting up system access is the manager's item, in the same plan as the new hire's.
- A template cannot be published without at least one gate, or with a phase that has nothing in it.
Day zero is the hire date. No per-hire drafting.
Starting a plan is a copy of the published template, not a fresh piece of AI writing. Each item's due date is the hire date plus its phase day, and training items become ordinary training assignments with those due dates.
- When your employee file is imported with a job role and a hire date within fourteen days of today, the ramp starts as part of the import.
- Changing someone's job role on the roster starts a role-change plan for the new role, dated from the day of the change, if that role has a published template.
- Editing a template later never rewrites a plan already running. Someone who started in January keeps the January plan; the next hire gets the new version.
- Due dates are stored, not recalculated. An overdue required item puts the plan at risk where a manager can see it, rather than quietly moving.
The AI assistant works onboarding through the same API as the person asking, with their permissions. Reads run at any autonomy setting; anything that changes a plan follows the dial for onboarding.
- “How is my onboarding going?”A new hire gets their own published plan and what is still open. Only their own: a plan still in draft is not visible to its subject.
- “Which sign-offs are waiting on me?”A manager gets the gates on their direct reports' plans still waiting on a person; an admin gets the whole organization. The assistant lists them. The manager signs off in the plan itself.
- “Start the Store Associate ramp for the people listed above.”Admins only, once the template and the people are already in the conversation. Starting a plan cannot be undone, because the new hire can read it at once, so it comes back as a proposal to confirm unless onboarding is on autopilot. For more than ten people it waits for confirmation at every setting.
- “Archive that onboarding plan.”Admins only, for a plan already in the conversation, and never unattended below autopilot: there is no unarchive, and the plan disappears from the employee's view. Training it already assigned stays assigned.
What “ready” means, and who gets to say it.
Most items close the way you would expect: the person marks it done, or the training assignment completes. A gate is different. A gate closes only on strong evidence, and the template names which kind:
- A manager sign-off, from someone who watched the person do it. Nobody can sign off a gate about themselves, whatever their role in the system.
- Reviewed work: the new hire submits a link to the work and a note on what it shows, and a manager accepts it or sends it back with a reason the new hire can read.
- A passed Understanding Check. An attempt that was only let through after repeated failures is recorded, but it can never close a gate.
- A passed certification conversation, where the person handles a simulated customer and a manager can read exactly why it was scored the way it was.
Self-attestation and finished training can never make a plan ready on their own. That rule is not a customer setting, and the database refuses a gate configured to close on a weak signal.
Managers see their direct reports' plans and work a single sign-off queue, with “Not yet” as a real action that needs a reason and moves the plan to at risk. Every plan has a printable evidence dossier: who signed what, when, how it was verified, and what is still outstanding.
| Question | Typical onboarding checklist | Ascend |
|---|---|---|
| What gets written | A task list per hire, often copied and edited by hand | One template per job role, reviewed once, copied exactly onto each hire |
| What closes an item | Someone ticks it | Ordinary items close on any accepted evidence; gates close only on a sign-off, reviewed work or a passed check |
| What finished means | Every box ticked | Every required item closed and every gate closed on strong evidence; no percentage decides it |
| Promotions and role changes | A new checklist, or the old one again | The new role's template, with skills the person has already proven marked as already demonstrated and cited |
| Who signs readiness | Often nobody, or the new hire | Whatever closed each gate is on record: a named signer who is not the new hire, or the check that was passed |
| What audit can see | A completion date | An evidence dossier per person, including what is still missing |
A promotion ramp covers the gap, not the whole job.
A new hire and a promotion run through the same code. Ascend compares the skills the target role requires against the skill levels the person has already been scored at. For a new hire the gap is the whole role. For someone moving up, it is only the difference.
Skill-focus items the person already meets are not deleted. They stay in the plan, marked already demonstrated, with the level held, the level required and the record that proved it. A gate is still a gate: if the template says a manager has to watch it, a manager still has to watch it.
When a role-change plan reaches ready, an admin makes the move effective, which updates the person's job role on record and closes the plan. A plan that is not ready cannot be made effective.
What buyers ask about onboarding.
- Does AI write a separate plan for every new hire?
- No. An AI drafting agent helps write the template for a job role, and a person reviews and publishes it. Each hire then gets an exact copy, so there is nothing new to review per person.
- Where does the start date come from?
- From the hire date on the employee record, which today comes from your employee file import. If there is no hire date, the plan starts on the day it is created.
- Can we let people become ready without a sign-off?
- No. Every template needs at least one gate, and a gate only closes on strong evidence. That is fixed for every customer.
Where onboarding connects.
Bring one job role. We'll run the loop on it.
Thirty minutes, your own documents, one job role. You'll see the company brain built, the drafts it produces, the sources behind them, and where the autonomy dial sits before anything publishes.