Your panel logs in. The résumé is polished, the LinkedIn profile is polished, and the candidate has had plenty of time to rehearse. You still need to decide whether this person can do the work, work with others, and justify the next step in your process.
That decision gets harder when hiring volume rises. Panels often waste time on broad prompts that reward polish over proof. The better approach is to ask questions that force detail, expose judgment, and produce answers each interviewer can score on the same rubric.
In practice, panel interview questions should do three jobs at once. They should test role fit, reduce bias, and create a record your team can defend if a hiring decision is challenged later. That matters in live panels, and it matters even more in asynchronous voice screens where consistency is the whole point.
I recommend treating the panel as an assessment system, not a free-form conversation. Use a small set of repeatable questions tied to defined competencies. Assign ownership so each panelist listens for specific evidence instead of forming a vague overall impression. Capture scores independently before discussion. That simple discipline cuts groupthink and makes debriefs faster.
The same structure works well in async voice screening. Every candidate gets the same prompt, the same response window, and the same scoring standard. You get cleaner comparisons, fewer off-script interviewer moments, and a process that scales without losing fairness.
If you are rebuilding the front end of your interview process, start with MyCulture.ai's HR hiring guide. Then use the eight questions below as a scoring framework designed for modern panel interviews and high-volume screening.
Table of Contents
- 1. Tell Me About Your Experience with Specific Role or Domain
- 2. Describe a Time You Failed or Made a Significant Mistake at Work
- 3. Why Are You Interested in This Role and Company
- 4. Walk Me Through Your Most Recent Project or Accomplishment
- 5. How Do You Handle Disagreement or Conflict with Team Members
- 6. What Are Your Salary Expectations and Career Goals for This Role
- 7. Describe a Time You Had to Learn Something New Quickly in a Role
- 8. Tell Me About a Time You Influenced a Decision or Led Without Authority
- Panel Interview Questions, 8-Point Comparison
- Turn Your Panel from a Conversation into a System
1. Tell Me About Your Experience with Specific Role or Domain
Three panelists listen to the same opening answer and leave with three different impressions. One hears confidence. One hears vague buzzwords. One hears a candidate who might fit if the team fills in the blanks. That is a process failure, not a candidate problem.
This question sets the scoring baseline for the rest of the interview. In a high-volume process, it also tells you whether the candidate can translate experience into evidence quickly enough to justify more interview time.
The mistake is asking it too broadly. “Tell me about your experience” gets polished summaries and résumé narration. A better version anchors the answer to work the panel can score: “Tell me about your experience managing enterprise renewals over $250K,” or “Tell me about your experience building backend services that handled high request volume.”
That level of specificity matters even more in an asynchronous voice screen. Candidates cannot rely on slides, chemistry with the room, or a strong panelist rescuing a weak answer with follow-up questions. They have to organize the answer on their own, which makes signal easier to compare across a large applicant pool.
What to listen for
Strong answers usually cover four separate things. Score them separately.
- Match to scope: Does the example fit the level of the role, or are they stretching a junior project to sound senior?
- Domain detail: Do they name the systems, customers, constraints, tools, or workflows involved?
- Decision quality: Can they explain trade-offs, not just activities?
- Communication control: Can another interviewer listen to the same recording and reach a similar score without guessing?
This split matters. I have seen candidates with real domain depth undersell themselves because their answer structure was weak. I have also seen polished candidates speak fluently about work they barely owned. One overall score hides that difference. A rubric exposes it.
Use role-specific prompts and define the proof points before interviews start. For a data analyst, that might mean stakeholder intake, data cleaning choices, and how they turned analysis into action. For a sales manager, it might mean territory design, forecast accuracy, deal inspection, and team coaching. For a product manager, it should include prioritization logic and the trade-offs behind launch decisions.
How to make it scorable in a panel or async screen
Keep the prompt narrow enough to produce a two to three minute answer. That is usually enough time for depth without creating rambling responses that panelists score inconsistently.
A practical format looks like this:
- Ask for one relevant example, not a career summary.
- Require context, ownership, and outcome.
- Add one constraint tied to the role, such as scale, customer type, regulatory complexity, or cross-functional coordination.
- Score content and delivery on separate lines.
For example, instead of asking a software engineer, “Tell me about your experience with distributed systems,” ask, “Describe one service you built or significantly improved. What problem did it solve, what trade-offs did you make, and what part did you personally own?” Instead of asking a salesperson, “Tell me about your enterprise experience,” ask, “Walk through the largest complex deal you led. How did you advance it, who was involved, and where did you have to change strategy?”
The difference is repeatability. Generic prompts create interpretation risk. Specific prompts create evidence.
Red flags that usually hold up under review
Once the rubric is clear, weak answers become easier to spot.
- Vague verbs like “supported,” “helped,” or “was involved” without ownership
- Heavy jargon with no explanation of how the work got done
- No numbers, scope, or operational constraints when the role clearly involves them
- Inability to explain why one option was chosen over another
- Contradictions between the résumé and the spoken example
These are common signs of inflated experience, but there is a trade-off. Some strong candidates are concise to a fault, especially in technical or operational roles. Do not confuse brevity with weak experience. If the structure is thin but the details are credible, score the content for what is there and note that follow-up may be needed later in the process.
The goal of this question is simple. Get enough role-specific evidence, early, to decide who merits deeper evaluation and who does not. Done well, it turns a familiar opener into a consistent filter the whole panel can use.
2. Describe a Time You Failed or Made a Significant Mistake at Work
A panel asks this question because résumé signal breaks down fast under scale. In a high-volume funnel, plenty of candidates can recite a polished failure story. Fewer can explain the decision they made, the consequence it caused, and the control they put in place so it did not happen again.
That is the standard to score against.
The mistake matters, but the operating pattern matters more. Good answers show judgment under pressure, clear ownership, and a prevention mechanism that would still make sense in your environment. In customer success, that might be mishandling a renewal risk and then building a tighter escalation path. In finance, it might be a forecasting miss followed by a better review cadence. In engineering, it could be a production bug that led to stronger test coverage or approval rules.
For panels and async voice screens, this question works best when you score four things consistently:
- Clarity of the failure: Did they state what went wrong in plain language?
- Ownership: Did they explain their part without hiding behind team wording?
- Correction: Did they change a behavior, process, or decision rule?
- Prevention: Did the fix reduce the chance of repeat failure?
A strong answer is specific enough that different interviewers would score it similarly. That is what makes this question useful at volume. You are not rewarding self-awareness in the abstract. You are looking for evidence that the candidate can absorb a miss, correct course, and lower future risk.
Owning a mistake without building a safeguard is not growth. It is just a story.
STAR can still help candidates organize the answer, but the scoring focus should stay on substance. Context should be brief. Ownership should be obvious. The strongest responses spend more time on what changed after the mistake than on defending why it happened.
This is also one of the better fraud filters in a panel format. If the story is borrowed, inflated, or heavily rehearsed, follow-up questions usually expose it. Ask what signal they missed, who was affected, what they changed first, and how they knew the fix worked. Weak answers go vague fast.
Watch for three patterns that create downstream management risk:
- They pick a fake failure that is really a strength pitch
- They describe the problem well but never name their own contribution to it
- They talk about learning, but cannot point to any process, habit, or control that changed
One caution. Some credible candidates answer this slowly, especially in asynchronous voice interviews where they cannot read the room. Do not confuse a measured delivery with weak accountability. Score the content. If the example is real, bounded, and operationally clear, the panel can trust the signal.
Used well, this question turns a familiar panel prompt into a repeatable screen for coachability, judgment, and risk reduction. That is far more useful than a generic test of humility.
3. Why Are You Interested in This Role and Company
Mass application behavior shows up fast. In a high-volume funnel, plenty of candidates are qualified enough on paper. Fewer can explain why your role, your company, and your problem set make sense for them specifically.
A good answer connects three things. The company's work, the role's scope, and the candidate's own trajectory. For a fintech startup, that might mean interest in embedded payments because the candidate previously worked on API-based products. For healthcare tech, it might come from direct frustration with the workflow the product is trying to fix. For B2B SaaS, it might reflect familiarity with the customer pain points your buyers already have.
This is one of the most useful panel interviews questions for async voice screens because enthusiasm is hard to fake for long. Candidates who care usually reference specifics naturally. Candidates who sprayed applications across fifty openings tend to stay generic.
Separate interest from performance
Don't confuse charisma with motivation. Some excellent candidates sound measured, not animated. What matters is whether they've done enough thinking to give a credible reason for wanting this exact job.
A practical threshold helps. Strong responses usually include at least two concrete references to the company, role, market, product, or customer problem that go beyond generic job description language. You don't need a rigid numeric rubric to see the difference. It's obvious when a candidate says, “I'm excited about growth” versus when they explain why your product direction fits work they've already done.
Use follow-up prompts in your scorecard notes, even if you don't ask them live:
- Space commitment: Would this person still want to work in this category if we rejected them?
- Role alignment: Are they drawn to the actual work, or to brand prestige?
- Retention logic: Does their reason suggest they'll stay long enough to matter?
Field note: The candidate who says “I need a job” may be honest, but honesty alone doesn't create retention, discretionary effort, or long-term fit.
Weak answers usually reveal themselves in broad statements about culture, compensation, or “growth opportunities” with no evidence that the person understands what the team does. That doesn't mean reject everyone who sounds practical. It means don't mistake availability for commitment.
4. Walk Me Through Your Most Recent Project or Accomplishment
A panel asks this question. One interviewer wants evidence of ownership. Another wants proof the candidate can explain complex work clearly. A third is listening for judgment under constraints. This prompt gives all three a usable answer if your process is set up well.
In high-volume hiring, this is more than a conversation starter. It is one of the cleanest ways to test whether a candidate can turn a résumé claim into a scorable work example. “Led launch,” “improved reliability,” and “owned campaign performance” all sound strong on paper. The answer tells you whether the person can explain the goal, constraints, decisions, their role, and the result without hiding behind team language.
The best answers follow a simple sequence. What was the problem. What did the candidate own. What trade-offs shaped the decision. What happened in the end. That structure works well in live panels and in asynchronous voice screens because every reviewer can score the same evidence instead of reacting to confidence or storytelling style.
A product manager might explain how they reduced onboarding friction across engineering, design, and marketing. A developer might walk through a migration that cut incidents but increased short-term delivery risk. A marketer might explain how they chose one audience segment over another and why. The function changes. The signal does not.
What to score separately
Do not collapse this into one impression score. Split it into components your panel can compare.
- Clarity of explanation: Could a cross-functional interviewer follow the sequence without extra translation?
- Ownership: Did the candidate separate their decisions and actions from the broader team effort?
- Judgment: Did they explain constraints, trade-offs, and why one path was chosen over another?
- Execution detail: Did they know the mechanics of the work well enough to answer pressure-tested follow-ups?
- Outcome credibility: Did the result sound specific, proportionate, and connected to the decisions they described?
This question is especially useful in asynchronous voice screening because the recording and transcript create an audit trail. Reviewers can revisit the same answer, compare it against role criteria, and spot inflation early. If someone claims they led a major initiative, the answer should show operational detail, not just a polished headline.
I also like this question because it exposes a common hiring risk. Candidates who were close to the work usually remember the messy parts. They can explain what nearly failed, what changed midstream, and what they would do differently. Candidates borrowing a team win often stay vague once you ask who made the hard call or what constraint shaped the final decision.
Use one follow-up across every interview to keep scoring consistent: “What was the hardest trade-off you had to make?” That single prompt tends to surface judgment faster than three generic follow-ups.
Here's a useful demo of stronger answer structure in practice:
Do not reward polish alone. Reward explainability, ownership, and decision quality. That is how this question shifts from a familiar panel prompt into a repeatable screening tool.
5. How Do You Handle Disagreement or Conflict with Team Members
Conflict answers reveal more than teamwork. They expose judgment under pressure, emotional control, and whether someone can disagree without becoming destructive.
The best version of this question asks for one real event, not a philosophy statement. An engineer might describe benchmarking two technical approaches and accepting the better option. A sales rep might explain an account ownership conflict and how it was resolved. A cross-functional operator might discuss roadmap friction and a phased compromise.
When candidates answer this well, they usually show that they understood the other person's position even if they disagreed with it. That doesn't mean they were passive. It means they could separate issue resolution from ego protection.
When panel dynamics get messy
This question has gained importance because panel formats themselves can create pressure. One under-discussed problem is competing panelist agendas. Candidates may get overlapping or contradictory follow-ups in real time, and many aren't prepared for that. One published discussion of panel interview dynamics notes that candidates often freeze in multi-voter conflict situations, and that larger panels create more interruption pressure than smaller ones, as discussed in this panel interview video analysis.
That matters for your process design. If your panel interrupts itself, you're no longer evaluating conflict handling at work. You're evaluating improvisation under interviewer disorder.
“I want to answer both points because they connect” is a strong candidate move. It acknowledges tension without taking sides.
Use this question to score the candidate's conflict style, not your panel's chaos tolerance. Strong responses usually include a specific incident, the other party's perspective, the resolution path, and what changed afterward. Weak responses often sound like courtroom summaries. “I was right, they were wrong, and eventually they understood” is not collaboration.
A practical rubric helps:
- Avoidant or aggressive: They withdrew, escalated too fast, or stayed combative.
- Passive compromise: They preserved peace but didn't solve the issue well.
- Collaborative problem-solving: They surfaced facts, clarified goals, and moved toward resolution.
- Growth from conflict: They adapted a future behavior because of the experience.
This is also one of the easiest answers to compare across panelists because tone carries so much signal. Contempt, dismissiveness, and low empathy come through fast in voice.
6. What Are Your Salary Expectations and Career Goals for This Role
Some hiring teams avoid this question in panel interviews because they think it breaks the flow. That's a mistake. Compensation mismatch and career mismatch create avoidable churn, and it's better to surface both before the offer stage.
The key is to ask it cleanly and score the answer for alignment, not negotiation style. A strong candidate usually knows the range they're targeting, can explain why it fits their experience, and can connect the role to a plausible next step in their career. A weaker answer is often evasive, unrealistic, or focused only on immediate pay with no thought to role scope.
In high-volume hiring, it's even better to publish the compensation band up front so candidates can self-select. Then this question becomes about fit and judgment rather than information asymmetry.
What good alignment sounds like
You're listening for coherence. Does the candidate's compensation expectation match the seniority of the role they're pursuing? Do their career goals fit the actual path your team can offer?
Useful follow-ups include:
- Retention risk: What would make them leave this role earlier than expected?
- Growth logic: Does the role move them toward a discipline they want?
- Level calibration: Are they stepping down strategically, or because they can't land at prior level?
A candidate who can explain both compensation logic and career direction is usually easier to close and easier to retain.
This is one area where interviewers often overreact. Someone asking for more money than you can pay isn't necessarily unreasonable. They may just be wrong for the role. Treat it as an alignment issue, not a character issue.
For panel interviews questions, this one works best when one person owns it rather than letting the whole panel pile on. Keep the discussion respectful, short, and anchored to the job you have. If the role can't support the candidate's compensation or trajectory, forcing optimism into the process wastes everyone's time.
7. Describe a Time You Had to Learn Something New Quickly in a Role
A team rolls out a new system in the middle of a hiring surge. Training is thin, deadlines stay the same, and the person you hire either gets productive fast or slows everyone else down. That is the scenario behind this question.
Use it to test ramp speed under pressure, not curiosity in the abstract. Good candidates can explain what changed, why the learning curve mattered, and how they got competent without creating delivery risk for the team. In high-volume hiring, that distinction matters because “fast learner” is easy to claim and hard to verify unless the question is scored the same way every time.
The strongest answers are specific and operational. A candidate might describe learning a new CRM before a territory launch, picking up a reporting tool after a process change, or taking over a workflow in an unfamiliar tech stack after a teammate left. The details matter less than the method.
Score the method, not the confidence
This question works well in asynchronous voice screens because it surfaces process. Candidates have to explain how they approached the ramp, what they used to get up to speed, where they got stuck, and how they checked their own progress. That gives panels a cleaner signal than polished personality alone.
For a scalable process, score this answer against a few fixed criteria:
- Context: Was the learning need real, urgent, and relevant to work?
- Approach: Did they use a clear plan, sequence, or practice loop?
- Support model: Did they rely fully on others, or use support well while driving their own ramp?
- Proof of competence: How did they know they were ready? Output, accuracy, speed, certification, stakeholder trust, or independent ownership all count.
- Reflection: Can they explain what they would repeat or change next time?
This also helps a panel avoid a common mistake. Strong communicators often sound adaptive even when the example is thin. A scorecard keeps the group focused on evidence.
Here's a practical calibration framework:
- Low initiative: They waited for instructions and can't explain their own learning process.
- Guided learner: They improved with close support and followed a structure someone else created.
- Self-directed learner: They built a plan, used the right resources, and reached competence on schedule.
- Force multiplier: They learned quickly, documented the path, and helped others ramp faster after them.
Weak answers usually break down in one of three places. The challenge was minor. The timeline does not hold up. Or the candidate never explains how they verified they were competent.
Listen carefully for proof. “I picked it up quickly” is not enough. Better answers include signals like reduced error rates, independent execution, positive customer outcomes, faster turnaround, or becoming the go-to person for that tool or process.
For panel interview questions, this one is especially useful when adaptability is a real job requirement and not just a nice phrase in the job description. In a structured panel, one interviewer can score learning agility while another scores clarity and judgment. That makes debriefs faster and more defensible, especially if you are reviewing large candidate volumes and need consistent pass-fail reasoning.
The best candidates make learning visible. They show how they shorten ramp time, protect output while learning, and turn ambiguity into a repeatable process. That is the signal worth hiring for.
8. Tell Me About a Time You Influenced a Decision or Led Without Authority
This question separates passive executors from people who can move work forward across functions. It matters for managers, but it matters just as much for strong individual contributors.
The best answers involve a real decision, real stakeholders, and some form of resistance. A designer might use user research to change a product direction. An operations lead might build a case for a process change. An IC might identify a recurring gap, socialize a solution, and gain adoption without formal control over the team.
What you're listening for isn't just persuasion. It's judgment. How did they build credibility? How did they handle pushback? Did they respect roles and constraints, or did they push harder until someone gave in?
Influence that actually scales
Candidates who've led without authority usually understand that influence is contextual. Sometimes they use data. Sometimes they build coalitions. Sometimes they earn trust by doing the hard prep work no one else wanted to do.
Good responses often include:
- Problem definition: They saw something specific that needed to change.
- Case building: They gathered evidence, examples, or stakeholder input.
- Resistance handling: They addressed concerns rather than dismissing them.
- Outcome and reflection: They can explain what happened and what they'd repeat or change.
Hiring judgment: “I convinced them I was right” is weaker than “I understood the objections and adjusted the proposal.”
This question also helps calibrate maturity. A candidate who says they went around their manager may be showing initiative, or they may be showing poor judgment. The difference is whether they understood the organizational context and handled escalation responsibly.
In async voice screening, this prompt is especially effective because influence stories tend to expose the difference between title-based authority and earned credibility. Candidates who've persuaded peers or stakeholders can walk you through the mechanics. Candidates who haven't usually default to vague claims about being proactive or collaborative.
Panel Interview Questions, 8-Point Comparison
| Question | Implementation Complexity 🔄 | Resource Requirements ⚡ | Expected Outcomes ⭐📊 | Ideal Use Cases 💡 | Key Advantages ⭐ |
|---|---|---|---|---|---|
| Tell Me About Your Experience with [Specific Role/Domain] | 🔄 Low–Moderate: open-ended prompt; needs scoring rubric | ⚡ Low: brief voice answer + optional transcription search | ⭐⭐⭐⭐ 📊 Reveals domain depth and communicative clarity | 💡 Screening for domain expertise and role fit (technical/customer-facing) | ⭐ Filters inflated resumes; transcription enables keyword matches |
| Describe a Time You Failed or Made a Significant Mistake at Work | 🔄 Moderate: requires culturally sensitive prompts and STAR scoring | ⚡ Moderate: trained reviewers to assess ownership and emotion | ⭐⭐⭐⭐ 📊 Surface resilience, accountability, and learning orientation | 💡 Evaluating emotional intelligence and growth mindset | ⭐ Exposes accountability vs. blame; tone reveals authenticity |
| Why Are You Interested in This Role and Company? | 🔄 Low: straightforward prompt but needs specificity threshold | ⚡ Low: minimal time; benefits from company-savvy reviewers | ⭐⭐⭐ 📊 Identifies motivation, research effort, and retention risk | 💡 Filtering for motivated applicants and cultural alignment | ⭐ Distinguishes genuine interest from generic applications |
| Walk Me Through Your Most Recent Project or Accomplishment | 🔄 Moderate–High: may require technical probes and context setting | ⚡ Moderate: domain-expert review and transcript verification | ⭐⭐⭐⭐ 📊 Provides verifiable evidence, technical depth, and metrics | 💡 Validating skills for senior or technical roles; reference checks | ⭐ Concrete proof of capability; easy to cross-check claims |
| How Do You Handle Disagreement or Conflict with Team Members? | 🔄 Moderate: needs trained panelists to interpret tone and nuance | ⚡ Moderate: requires listening for examples and follow-ups | ⭐⭐⭐⭐ 📊 Assesses collaboration style and interpersonal risk | 💡 Hiring for team-based roles, managers, and cross-functional work | ⭐ Reveals conflict behavior and emotional regulation via voice |
| What Are Your Salary Expectations and Career Goals for This Role? | 🔄 Low: direct qualification question; sensitive framing required | ⚡ Low: quick to ask; benefits from published salary bands | ⭐⭐⭐ 📊 Aligns compensation expectations and long-term fit early | 💡 Early-stage funnel qualification to avoid mismatches | ⭐ Saves time by eliminating budget misalignment before panels |
| Describe a Time You Had to Learn Something New Quickly in a Role | 🔄 Low–Moderate: prompt is simple; scoring depends on role needs | ⚡ Low: short answer; transcription captures tools/methods | ⭐⭐⭐⭐ 📊 Predicts ramp speed, adaptability, and initiative | 💡 Startups, fast-changing tech stacks, and high-growth roles | ⭐ Signals self-directed learning and resourcefulness |
| Tell Me About a Time You Influenced a Decision or Led Without Authority | 🔄 Moderate: requires probing to confirm influence vs. formal authority | ⚡ Moderate: reviewers should evaluate evidence and outcomes | ⭐⭐⭐⭐ 📊 Identifies leadership potential and persuasion skills | 💡 Matrix organizations, IC leadership, cross-functional change | ⭐ Surfaces persuasion methods and ability to drive change without title |
Turn Your Panel from a Conversation into a System
A candidate clears the recruiter screen, shows up polished, and leaves half the panel feeling confident. Two days later, the debrief falls apart. One interviewer liked their presence. Another heard vague answers. A third never got evidence on the skill they were assigned to assess. That is not a panel problem. It is a system problem.
Strong panel interviews questions create a repeatable decision process. The goal is simple: each interviewer tests a defined competency, scores against the same rubric, and documents evidence that another reviewer could inspect later. That structure improves fairness, cuts duplicate questioning, and gives the hiring team a record it can defend if a candidate challenges the process.
This matters even more in high-volume hiring.
By the time someone reaches a live panel, the team should already have proof of communication quality, relevant examples, and basic role fit. If the panel is the first time anyone hears the candidate explain their work, the company is using senior interviewer time to do early screening. That is expensive and hard to scale.
The eight questions in this guide work well because they force candidates to give specifics: what happened, what they owned, what trade-offs they made, and what result followed. That makes them useful in a live panel and just as useful in an asynchronous voice screen. A structured recorded answer gives every reviewer the same raw input instead of a summary filtered through recruiter notes or first impressions.
The failure point is usually not the question set. It is the missing scoring layer. Teams ask decent questions, then collect feedback like “strong communicator,” “seems sharp,” or “not sure on leadership.” Those comments do not hold up in calibration. They also do not help a TA leader spot whether the issue is candidate quality, interviewer inconsistency, or a weak prompt.
Build the process with clear handoffs:
- Assign one competency to each question
- Define what a strong, mixed, and weak answer sounds like
- Separate scoring for ownership, judgment, communication, and role alignment
- Require private score submission before debrief
- Use the same rubric in the async screen and the live panel
- Track disqualifiers and red flags in writing, not from memory
That last point matters. Written criteria reduce drift across interviewers and across hiring cycles. They also make panel training easier because new interviewers can review sample answers and see how the team scores them.
I have seen the best results when the async screen does the first-pass evidence collection and the panel handles validation, depth, and edge cases. That is the right trade-off for speed and quality. Recruiters stop pushing borderline profiles into expensive interviews. Hiring managers get a tighter slate. Panels spend their time testing judgment and nuance instead of re-covering basics.
Platforms like WorkSignal fit that workflow well. Candidates complete a short voice screen on their own schedule. The hiring team reviews transcripts and recorded responses against role-specific criteria before the ATS fills up with weak matches. That gives TA leaders a compliance-aware screening layer, cleaner calibration data, and a panel process that behaves like an operating system instead of a loosely structured conversation.
