Research in AI, human-robot interaction, autonomous vehicles, and machine learning at MIT.

Lex Fridman

I stand with the people of Iran ❤️

I have a lot of Iranian friends, including my childhood best friend. Amazing people.

It's one of the great cultures and peoples in the history of the world.

Stay strong ❤️

Here's my conversation all about AI in 2026, including technical breakthroughs, scaling laws, closed & open LLMs, programming & dev tooling (Claude Code, Cursor, etc), China vs US competition, training pipeline details (pre-, mid-, post-training), rapid evolution of LLMs, work culture, diffusion, robotics, tool use, compute (GPUs, TPUs, clusters), continual learning, long context, AGI timelines (including how stuff might go wrong), advice for beginners, education, a LOT of discussion about the future, and other topics.

It's a great honor and pleasure for me to be able to do this kind of episode with two of my favorite people in the AI community: Sebastian Raschka and Nathan Lambert. They are both widely-respected machine learning researchers & engineers who also happen to be great communicators, educators, writers, and X posters.

This was a whirlwind conversation: everything from the super-technical to the super-fun.

https://lnkd.in/gViznQAt

Open video: State of AI in 2026: LLMs, Coding, Scaling Laws, China, Agents, GPUs, AGI | Lex Fridman Podcast #490 on youtube.com

State of AI in 2026: LLMs, Coding, Scaling Laws, China, Agents, GPUs, AGI | Lex Fridman Podcast #490

"We are all in the gutter, but some of us are looking at the stars." - Oscar Wilde

Michael Pedersen

Revenue Growth | Emotion-Based Brand Messaging To 3x Your Revenues | Bridging the gap between sales and marketing

High-pressure phone sales teams lose 30%-40% when recovery takes longer than the next dial.

DecideFast™ Between-Calls Performance Stability Software 
"Tough calls don’t crash the day.”

"DecideFast™ runs alongside your existing tech. 
No integration required.
No coaching.
No training.
No workflow changes.
No IT involvement.

Teams do not fade because of skill or effort. They fade because timing breaks under pressure.

Training happens before the shift.
Coaching happens after the shift.
Dashboards explain the problem once the day is already lost.

Nothing supports recovery in the seconds after a call, when performance actually drops.

That is the gap DecideFast™ fills.

DecideFast™ compresses after-call recovery so each interaction starts clean.

Used in seconds.

No management overhead.

See fast after-call recovery live. 
No slides. No pitch.

CEO & Founder

Does your website convert? When was the last time you gave it a refresh? How does it compare to your competitors?

Your company website is your brand on the Internet. An under-performing and outdated website will cost you revenues. 

Let’s jump on a discovery call to see if we are good fit to work together 

Visit: www.AZNetmarketing.com or 
Email: mike@aznetmarketing.com

How is this even possible?

30–45% annual attrition has become normal in phone sales.

That range shows up consistently across industry benchmarks from Contact Center Pipeline and ICMI, especially in high-volume and BPO environments.

Now put a dollar frame on it.

Research from SQM Group and Gartner estimates the cost to replace a frontline agent typically lands between $10,000 and $20,000 once you account for recruiting, training, ramp time, and lost productivity.

That means a 300-seat operation turning over a third of its workforce each year is quietly absorbing millions in replacement cost alone.

What’s important here.

This didn’t become normal because leaders stopped caring.

Managers are under real pressure. They run WFM, QA, coaching, engagement tools, training programs. They are actively trying to stabilize performance with the systems available to them.

The issue is timing.

Most of those programs work before the day starts or after the damage shows up in reports.

Performance doesn’t usually break in meetings or dashboards.

➡️ It breaks in the moments between interactions.
➡️ Right after a bad call.
➡️ Right before the next one.
➡️ When pressure carries forward and nobody sees it.

So teams adapt.
🚫 They manage averages.
🚫 They absorb churn.
🚫 They normalize something that would feel unacceptable in almost any other function.

This isn’t about blame.

It’s about noticing when something abnormal quietly became accepted as “just how it works.”

At what point did “this is brutal” turn into “this is normal”?

Mike Pedersen

Phone Sales Teams Lose 30%-40% Output After Lunch | Between-Calls Performance Software That Keeps Afternoons From Breaking | No Training. No IT. No Workflow Changes

Can a phone sales rep reset in 10 seconds?

Most leaders have never tested this.
Not because it sounds impossible.
Because nothing has ever been built for that window.

Sales performance is designed around calls.
Measured on calls.
Coached on calls.

But floors don’t drift during calls.
They drift after them.

That small gap between one call ending and the next starting quietly determines the rest of the day.

🚫 No system reaches it.
🚫 No process operates there.
🚫 No tool moves fast enough.

Leaders feel the impact anyway.

⬇️ Energy drops by mid-day.
⬇️ Confidence slips in real time.
⬇️ One bad call spills into the next.

By the afternoon, the floor is still staffed, but performance is uneven.

Most solutions arrive too late.

🚫 Dashboards explain yesterday.
🚫 Coaching fixes tomorrow.
🚫 Training assumes calm and recall.

None of those help in the moment that actually bends the day.

Sales performance doesn’t break on calls.
It breaks in the seconds right after them.

When recovery is immediate, performance holds.
When recovery drags, the day compounds in the wrong direction.

That gap has always been there.
Leaders have felt it.
They just haven’t had anything fast enough to stabilize it.

The real question is simple.

What happens in the 10 seconds after a call ends?

That moment decides whether the next call starts clean, or carries residue.

DecideFast™ operates in that gap.
Between calls.
In real time.

We’re running a contained test with sales leaders who want steadier floors without adding training, process, or disruption.

If this matches what you see on your floor, message me and we’ll see if it fits.

Every sales leader I talk to shows the same curve.

Mornings prove the team can execute.
Afternoons make it look like they can’t.

✅ Same reps.
✅ Same scripts.
✅ Same leads.

Most explanations miss the point because they focus on effort, motivation, or skill. None of those change at noon.

What actually changes is the operating condition.

🚫 Calls stack.
🚫 Difficult interactions cluster.
🚫 Nothing clears between one call ending and the next beginning.

So by mid-afternoon, reps aren’t performing worse.
They’re performing under load.

That’s why dashboards lag reality.
By the time the numbers dip, the damage already spread across dozens of calls.

If morning performance exists, the capability exists.
The question isn’t how to train harder.
It’s how to stop degradation from compounding during the day.

Most teams don’t have a way to manage what happens after the call and before the next one.

That gap quietly decides how the day finishes.

If you run a floor, you already know this pattern.

Most phone sales teams don’t struggle because of tools, talent, or effort.

They struggle because daily output isn’t predictable.

Same team.
Same dialer.
Same leads.

One day hits.
The next day misses.

That inconsistency creates real downstream problems.

• Managers spend their afternoons intervening instead of managing
• Clients feel the instability before reports ever show it
• Forecasts become guesses, even with solid activity data

The dialer does its job.
Activity is optimized.

What changes is execution as the day wears on.

Pressure accumulates.
Judgment slips.
Tone changes.
Small drops compound across dozens of calls.

Most systems measure results after this happens.
They were never designed to stabilize performance while it’s forming.

That gap is why output feels harder to control than it should.

Sales performance still drops later in the day.

Even in teams using Gong or similar tools.
That confuses a lot of sales leaders.

Because the calls are recorded.
➡️ The data is there.
➡️ Coaching is happening.
➡️ Yet consistency still breaks.

Here’s why.
Most sales tools explain what happened on the call.
They do not support what happens after the call.

➡️➡️ Insight explains what happened.
What carries forward after a call determines what happens next.

Between calls is where problems stack:
➡️ One bad conversation carries forward
➡️ Confidence dips before the next dial
➡️ Hesitation shows up quietly
➡️ Call blocks lose stability

By the time coaching or reviews happen, the day is already off track.

This isn’t a tooling failure.
It’s a missing layer.

Insight explains performance.
Recovery protects it in real time.

If your team starts strong and fades by the afternoon, what actually happens between calls today?

That’s the question worth looking at.

After Call Work looks fine on dashboards.

It’s killing output in reality.

Reps finish a hard call.
✅ They log notes.
✅ They click dispositions.
✅ They pause.

🚫 That pause turns into hesitation.
🚫 Hesitation turns into slower dials.
🚫 Slower dials turn into crushed afternoons.

Leaders see:
 • ACW implemented
 • Systems configured
 • Activity targets set

But still spend the day reacting.
ACW tracks tasks.
It doesn’t stabilize execution.

If nobody owns what happens between calls, consistency erodes even when everything looks “set up correctly.”

That gap is where performance quietly breaks.

Solution providers sell “performance software.”

Here's my problem with it.

Most of it measures. Almost none of it improves execution/outcomes.
Here’s the line that gets blurred.

What most companies call performance:
➡️ Dashboards
➡️ Reports
➡️ Visibility
➡️ Retrospective analysis

Trends after the outcome is locked.
That is measurement.
Measurement explains what already happened (lagging indicators).

What performance actually is
✅ Execution under pressure
✅ Consistency as conditions deteriorate
✅ Output holding when rejection stacks up
✅ Results changing in real time

I come from an athletic background (a full-ride D1 college scholarship).
Track and field. Javelin, then Decathlon. Attempted to make an Olympic team.

➡️ Performance was never the quality of the data.
➡️ Performance was the clock. The distance. The placing.
➡️ You didn’t win because practice was tracked better.
➡️ You won because execution improved.
➡️ Coaching works, but not in the moment of pressure between calls.

Phone sales work the same way.
If execution doesn’t change between calls, performance doesn’t change.
If nothing stabilizes execution during the day, afternoons break.

If tools only explain losses after the fact, they are not performance tools.
Measurement tells you what happened.
Performance determines what happens next.

This is why DecideFast™ sits in performance stability.
We operate where execution actually breaks.
Between calls. Under pressure. In real time.
✅ It does not measure outcomes after they fall.
✅ It stabilizes execution while work is happening.
That distinction is the whole point.

It's time companies focus on outcomes, not just measurement.

If you’re an inside sales leader and output is still off, you fix it. You don’t keep buying dashboards.

Phone-based sales operations lose 30% of daily output after lunch.

The pattern appears across teams, industries, and tech stacks.

What operations leaders see:
✅ Strong morning dial rates.
🚫 Declining afternoon production.
➡️ Same reps. Same leads. Same hour count.

What the metrics show:
✅ First-half call volume: 12-13 per hour
🚫 Second-half call volume: 8-9 per hour
🚫 Conversion rates: drop proportionally
✅ Manager intervention: increases 3 PM onward

What this represents operationally:
⬇️ 30% of capacity unused every afternoon.
⬇️ Full payroll expense. Reduced output delivery.
⬇️ Forecast misses despite effort increases.

Standard fixes don't address it:
🚫 Better list quality doesn't prevent the decline.
🚫 Script improvements don't maintain afternoon pace.
🚫 Incentive adjustments don't change the pattern.
🚫 Additional coaching doesn't stabilize output.

Because the constraint isn't in those areas:
➡️ Training builds capability over weeks.
➡️ Coaching refines execution over days.
➡️ Dialers optimize connection rates continuously.
➡️ CRMs track activity after it occurs.

None operate in the moments between calls when one difficult interaction affects the next four attempts.

That's where the 30% disappears.

That's the operational gap costing seven figures annually.

It's time to stop normalizing 30% drops and fix it.

Inside Sales Leaders:

Your dashboard is red. Again.

12 calls/hour this morning. 8 calls/hour right now.
Same reps. Same leads. Different output.

You know exactly when it started.

That brutal rejection at 1:47 PM. The one where the prospect hung up mid-sentence. Your rep stared at the screen for 12 seconds. Dialer queued the next call. Rep wasn't ready.

That call tanked the next four.

You watched it happen. You couldn't stop it.

Every. Single. Day.

DecideFast™ operates in those 8 seconds.

Between one call ending and the next connecting. Where your rep either clears what happened or carries it forward.

Pilot teams: Afternoons now match mornings.

If you're done watching 30% disappear daily, let's talk.

DM me. 10 minutes.

The One Moment No Phone Sales Tech Stack Covers

Between-Calls.

These moments during the phone sales day explain why performance breaks even when everything appears to be in order.

It is the only moment no system owns.
Dialers move the next call. CRMs log the last one. Nothing is responsible for what carries from one interaction into the next.

It is the only moment managers cannot scale into.
A manager can step in after the damage is visible. They cannot be present in the seconds between every call across dozens of reps.

It is the only moment dashboards cannot see.
Dashboards show outcomes after execution already changed. They never show the hesitation, tone shift, or pace decay forming between calls.

It is the only moment where one interaction affects the next.
A harsh rejection does not stay isolated. Without clearing, it alters the next conversation, then the next, then the next.

It is the only moment that explains same-day instability without blaming reps.
The capability is proven in the morning. The breakdown happens later because carryover compounds, not because reps forgot how to sell.

That moment sits between calls.

DecideFast™ is a between-calls stability performance software for phone-based sales teams, designed to prevent output decline throughout the day. 
Not training. Not coaching. No IT lift. No workflow change. 

Used in the seconds after a tough call to clear the carryover fast and maintain dial pace and next call success.

We are executing low-risk, 14-day, 25-rep contained tests to stabilize afternoons for phone sales teams. 

Let's talk if you want to hit your afternoon numbers and have your managers intervene less.

taplio

The Battle

Lex Fridman

Michael Pedersen

Followers

Posts

Engagements

Likes

Performance ScoreLex Fridman's LinkedIn Performance

Performance ScoreMichael Pedersen's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Lex Fridman

Michael Pedersen

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Lex Fridman's LinkedIn Performance

Performance ScoreMichael Pedersen's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreLex Fridman's LinkedIn Performance