Real-time, human-like video and voice interactions built for scale. Our white-labeled conversational APIs integrate easily across products to power adaptive interviews, customer support, and training that feels personal.
Written by
Most onboarding training ends before the job actually begins. A new hire can finish every module and pass every quiz, then still freeze the first time a customer pushes back on a live call. The distance between "training complete" and "ready for real work" is where onboarding training either builds confidence or quietly fails.
SHRM's onboarding guidelines explain that employees get roughly 90 days to demonstrate they can do the job. Personified Application Layers (PALs), AI companions that see, hear, and respond face-to-face in real time, provide onboarding training with an interactive surface where new hires can rehearse the conversations that matter.
Onboarding training is the structured skill-building that turns a new hire into someone who can do the job independently. It provides guided practice on the actual tasks and conversations the role requires, with feedback and enough repetition to improve. It sits inside the broader onboarding process.
Orientation typically lasts a few days to a week and covers paperwork, policies, and introductions. Onboarding itself can run up to 12 months, but onboarding automation can shorten that path.
When onboarding training works, the new hire has already rehearsed the first real customer call, difficult meeting, or compliance-sensitive conversation before it happens. Completion certificates record module completion, but "onboarding complete" and "job ready" are separate milestones.
Gallup's onboarding research found that only 12% of U.S. employees strongly agree that their organization does a great job of onboarding, and that 20% of turnover occurs within the first 45 days.
Most programs measure completion without giving new hires enough rehearsal. Slide decks and learning management system (LMS) modules test recall on a posttest, but ar research by CCL (Center for Creative Leadership) attributes only about 10% of professional development to formal coursework, with 70% coming from challenging on-the-job experiences and 20% from developmental relationships.
New hires rarely get to practice a denied-claim call, a pricing objection, or a frustrated-customer escalation before facing one live, because human-led role play is hard to staff at the scale a cohort needs. ATD's training guidance recommends simulations and role-play that closely mimic day-to-day challenges, since retention from active practice averages around 75%, compared with roughly 5% for lecture-style learning.
Turning a module into a practice partner requires something that behaves like a person in conversation: an application that watches, listens, and responds with presence, the sense that someone is paying attention to what you actually do. Tavus is the human computing company, that builds human-like PALs (Personified Application Layers) for that job. This is how they can help during onboarding training.
Maya, a claims associate at a mid-size insurance carrier, takes a call from a PAL playing a policyholder whose water-damage claim was denied. She has already identified the escalation policy in a four-option quiz. Now she has to apply it while someone is upset.
The PAL responds like a person. When Maya rushes past the frustration, the policyholder pushes harder.
When she pauses to think, the PAL holds the space without cutting her off. When she finally names what the policyholder is feeling, the tone softens.
Maya can run this scenario at 8 pm on a Tuesday, before her first live escalation call the next morning. That's the point of practice: build the muscle before the stakes are real.
The PAL reacts during the conversation, so Maya sees the effect of her choices immediately. When the practice policyholder softens because she finally acknowledges her frustration, that reaction is feedback delivered at the exact moment the behavior occurs.
She doesn't have to wait for a manager to listen to a recording next week. She doesn't have to guess whether her tone landed. The reaction is right there on the screen, in the face and voice of the person she's talking to.
Afterward, the session transcript records where Maya hesitated and where she recovered. Her manager can jump straight to those moments during the debrief instead of scrubbing through a full call.
Coaching capacity is another constraint. SHRM research on AI coaching found that 61% of HR professionals say fewer than half of their managers effectively address underperformance among direct reports, and many new hires get too little coaching in their first three months.
PALs close the coaching gap without replacing managers. Three weeks in, Maya can rerun a denied-claim escalation five times in an evening without a human coach present. Cross-session memory means the PAL recalls she landed the empathy statement last time but stumbled on the coverage exclusion, and it starts there.
Teams can set rules for what routes to a human. Cases where a learner plateaus, or where nuance matters, go to live manager coaching. The rest run on-demand.
In Maya's practice session, the PAL plays a policyholder whose water-damage claim was just denied. Behind that conversation, four components operate in a closed loop with sub-second latency.
The stack:
That loop runs on the Conversational Video Interface (CVI), the API pillar that delivers PALs through real-time conversational video. No single component captures the human feel on its own; integration is what separates a demo from infrastructure that holds up in production.
Start with the situations managers say new hires mishandle. Then run scored cohorts and route the hardest cases to a human. The sequence fits the 30/60/90-day check-in cadence most programs already keep.
Talk to managers before designing scenarios. Ask where week-3 and week-7 hires stumble: the denied-claim call, the pricing objection, the handoff that goes sideways, the compliance conversation that trips a policy line. Build the scenario list from those reports.
Describe the scenario in plain language. The PAL Maker no-code builder handles configuration end-to-end: system prompt, Replica, Objectives, Guardrails, tools, and Knowledge Base. Program owners can talk through the scenario in the same session, adjusting tone and difficulty until it matches what new hires will actually face.
Upload the source material the PAL should draw on: policy documents, product specs, escalation trees, compliance guides. The Tavus Knowledge Base ingests PDF, CSV, PPTX, and TXT files and retrieves relevant context in ~30ms, so conversations don't stall while the PAL looks something up.
Apply the same rubric to every learner. Score dimensions like clear explanations, empathy, confidence, and answer accuracy. Send competency scores to your LMS through SCORM and xAPI, the course-tracking standards that pass scores and completion data across systems.
Readiness shows up in independent handling, scenario performance, and manager assessments. Use these measures to judge whether a new hire is ready.
Managers can review these measures alongside live-work results when deciding whether a new hire needs additional coaching.
Somewhere in week six, Maya takes her first live escalation call. Her manager has scenario results to inform the readiness decision. The policyholder is angrier than the practice version, but the conversation still feels familiar: she has already heard this objection, recovered from the stumble, and had a practice partner respond to what it saw.
Tavus builds PALs that see, hear, understand, remember, and respond face-to-face in real time. That presence turns onboarding training from mere module completion into genuine rehearsal, and it gives new hires like Maya a place to practice before the stakes are real.
See it for yourself. Book a demo.
The 30-60-90 rule structures a new hire's first three months around three check-in milestones. At 30 days, the focus is learning: absorbing the role, meeting the team, and understanding expectations. At 60 days, contribution begins, with the new hire owning smaller pieces of real work. By 90 days, they should be handling core responsibilities independently.
Ownership typically falls to People or Learning & Development, but the program only works when hiring managers remain involved. HR designs the structure, tracks completion, and maintains the content library. Managers define what "ready" looks like for each role, flag the scenarios new hires actually stumble on, and run the check-in conversations at 30, 60, and 90 days.
Yes. Remote teams already run most of their work through calls and shared documents, so practice scenarios that mimic real customer or internal conversations translate directly. PALs give distributed new hires a practice partner available across time zones, without waiting for a manager to schedule a role play.
