LLenny's Podcast
← All episodes
Paul Adams (CPO of Intercom)26 October 2023

What AI means for your product strategy

5Frameworks
15Insights

Frameworks in this episode

Strategy3 steps

Differentiation vs Table Stakes

Balance the roadmap between what attracts new customers and what they require to switch at all

Read →
Strategy4 steps

Map Your Product Against What AI Can Do

Start from your product's core job, then classify each part as AI-replace, augment, partial, or not-yet

Read →
Marketing3 steps

Product-Market-Story Fit

Great product in a great market still fails if the story is wrong or missing — add the third leg

Read →
Leadership3 steps

Swinging the Pendulum

Spot an undesirable state, correct it — but know you'll overcorrect, so aim for the middle

Read →
Sales4 steps

The Four Forces of Switching

Understand any adoption decision as a tug-of-war between attraction, anxiety, habit, and inertia

Read →

Insights & moments

The myth-busts, hot takes, explainers, and tools worth keeping.

Hot Take· 4

Hot Take15:30

AI Is a Meteor Coming Toward You — Bigger Than Mobile

Adams describes people splitting into two camps on AI: those who are all-in and see it as a technology shift bigger than mobile and maybe bigger than the internet, and skeptics who dismiss it as hype like crypto, web3 or the metaverse. He places himself firmly in the all-in camp and says even new releases like ChatGPT Vision blow his mind while we're still scratching the surface.

  • He wakes up thinking about AI every day since ChatGPT launched (Nov 29)
  • One camp is all-in, viewing AI as bigger than mobile — maybe bigger than the internet
  • The other camp sees hype-cycle repetition of crypto, web3 and the metaverse
  • He references the hype-then-trough-of-disillusionment curve as where many skeptics sit

“This is a like meteor coming towards you like this is you know bigger than mobile as a kind of Technology shift as big as…”

Paul Adams · 16:00
#ai#hype-cycle#technology-shift#opinion
Hot Take31:00

AI Is Already Better Than a Radiologist — and That Could Be a Good Thing

Adams points out that becoming a radiologist in Ireland takes seven years of training, yet AI already appears better at reading X-rays because it can ingest and synthesize every X-ray ever made — something no human can. He argues the job doesn't disappear but fundamentally changes to checking AI's output and bedside manner, and could help clear Ireland's long hospital waiting lists.

  • Radiology takes ~7 years of training in Ireland
  • AI can ingest and synthesize every X-ray ever made, which no human can
  • The job shifts from analyzing scans to verifying AI and patient bedside manner
  • This could be positive given long hospital and X-ray waiting lists

“It's seven years training to like learn that learn that skill so seven you be radiologist and then you kind of just into the job…”

Paul Adams · 31:30
#ai#healthcare#jobs#opinion
Hot Take32:00

AI Voice and Face Cloning Could Create a Digital Afterlife

Adams calls voice cloning the craziest capability he's seen — AI can copy your voice almost indistinguishably, and combined with 3D face modeling (as in Zuckerberg's metaverse demo with Lex Fridman) it can replicate a real person. He speculates AI could ingest everything someone said and did so that after they die you could still talk to them — an unsettling 'afterlife' that echoes a Black Mirror episode.

  • AI can copy a voice almost indistinguishably from the real person
  • Face-modeling tech (per the Zuckerberg/Lex Fridman metaverse demo) is close to photorealistic
  • Combined, voice and face are nearly indistinguishable from a real person
  • AI could replicate a deceased person so relatives can still talk to them — he notes a Black Mirror episode with the same premise that didn't end well

“AI will be able to ingest all the things people say and do and when people die it'll be able to replicate that person you…”

Paul Adams · 33:00
#ai#voice-cloning#ethics#future
Hot Take49:30

Don't Panic About AI Job Losses — It Plays Out Through Attrition

Adams cautions leaders against fear-mongering about headcount (e.g. announcing 'we only need two engineers per team'). He cites studies showing people don't tend to lose jobs so much as have roles moved around, and notes customer support is high-attrition — so 'a bot will take everyone's job' more realistically plays out as roles simply not being backfilled.

  • Don't walk around declaring you need fewer engineers per team — it's not how it plays out
  • Studies show jobs get moved around rather than eliminated wholesale
  • Customer support is a high-attrition role
  • AI-driven reduction more likely happens via attrition (people quit and aren't backfilled) than mass layoffs

“loads of great studies over the years about how people don't end up losing jobs um the jobs get moved around”

Paul Adams · 50:00
#ai#jobs#customer-support#leadership

Story· 5

Story04:30

Freezing On Stage During a Cannes Keynote — and Surviving It

Paul Adams recounts giving Facebook's keynote at Cannes, the world's biggest advertising festival, a few minutes into a talk he'd rehearsed word-for-word. He froze, had a panic attack, and walked off stage still miked up while the crowd laughed. He managed to walk back out and finish, and the takeaway is that even the worst public-speaking nightmare can work out and isn't as big a deal as it feels.

  • He was asked to do Facebook's Cannes keynote the year after Zuckerberg got a hard time on privacy
  • It was the first talk he'd ever rehearsed word-for-word rather than ad-libbing from talking points
  • He froze, walked off stage miked up, then recovered and finished the talk successfully
  • Ever since, he carries a tiny doubt before every talk but believes you have to adapt when life throws curveballs

“and um I just froze I I couldn't remember what I was supposed to say uh it was the first ever time in my life…”

Paul Adams · 05:30

“I had some version of a panic attack I walked off stage I was still miked up”

Paul Adams · 05:30
#public-speaking#failure#career#resilience
Story08:00

Why Google's Social Projects Failed: They Came From Fear

Reflecting on his four years at Google working on Buzz and Google Plus, Adams argues the projects failed because they were motivated by competitive fear of Facebook rather than a genuine desire to solve people's problems. He contrasts this with the great products Google was building at the same time, like Maps and Android, and says the experience taught him how not to lead.

  • Google feared Facebook as an existential threat; Facebook thought it could build a social ad unit to eat Google's revenue
  • Google Buzz and Google Plus came from a place of fear, not from trying to make a great product for people
  • Meanwhile Google was building Maps and Android — products Adams considers among the best ever made
  • Fear-driven projects, in his view, don't lead to good things

“motivation for those projects came from a place of fear you know it didn't come from a place of Let's Make a great product for…”

Paul Adams · 08:30
#google#product-strategy#leadership#failure
Story09:30

The Research at Google That Basically Predicted WhatsApp

As a UX researcher at Google, Adams smuggled formative research into tactical usability studies, mapping out dozens of people's entire social networks over roughly 18 months. The same pattern emerged every single time: people needed far better ways to communicate with small groups of family and friends — obvious in hindsight (WhatsApp, iMessage) but not at the time.

  • He'd add 20 minutes of open-ended research onto assigned usability studies
  • He mapped people's family, friends and communication channels with dozens of people over ~18 months
  • The consistent finding: people need better ways to communicate with small groups
  • The insight was made public via a conference talk, which is what got Facebook's attention

“the same pattern emerged every single time which was people need way better ways to communicate with small groups of family and friends”

Paul Adams · 10:00
#user-research#google#product-insight#communication
Story11:30

Treated Like a Spy: Leaving Google Plus Mid-Build for Facebook

Adams left Google halfway through the confidential Google Plus project to join Facebook, walking out with all the plans in his head. Some colleagues saw him as a traitor; when he announced his departure Google quarantined him and forensically analyzed his laptop, assuming he was a spy. He traces the internal antagonism back to the project being run from a place of competitive fear.

  • Google Plus was highly confidential — staff worked in a separate building with a different key card
  • Leaving mid-project meant walking to 'the enemy' with the plans in his head
  • Google quarantined him and forensically analyzed his laptop, assuming he was a spy
  • He again attributes the intensity to the project being driven by competitive fear

“when I left uh Google kind of assumed that I was one of the spies you know I was quarantined I told them I was…”

Paul Adams · 12:00
#google#facebook#career#corporate-culture
Story23:30

Intercom Ripped Up Its Strategy From First Principles for AI

Adams describes ChatGPT's launch as a 'before/after moment' where Intercom tore up its strategy almost entirely and started again from first principles around why people use it — customer support. Sam Altman soon named customer service as one of the first industries to be disrupted, validating a 'bet the farm' pivot that produced Intercom's AI product, Finn.

  • Intercom's head of ML flagged ChatGPT's launch as the before/after moment
  • They rebuilt strategy from first principles around their core: customer support
  • Sam Altman publicly named customer service as an early industry to be disrupted
  • The pivot was a 'bet the farm' mindset that led to building Finn

“there was before and that is after and like everything has changed so we literally ripped up our strategy almost entirely and started again like…”

Paul Adams · 24:00

“one of the first Industries that's going to be disrupted as customer service”

Paul Adams · 24:30
#intercom#ai#product-strategy#customer-support

Tool· 1

Tool74:00

The Reference-Call Question a Candidate Can't Dodge

Adams shares a hiring technique he was given for reference calls: since referees are always the candidate's favorite people and best colleagues, ask 'what feedback will I be giving this person in their first performance review?' It's enlightening because there's always a real answer and the referee can't dodge it.

  • Referees are inevitably the candidate's best past colleagues and favorite managers
  • The question: what feedback will I be giving this person in their first performance review?
  • It works because the person can't dodge it — there's always a genuine answer
  • He credits the question to a former colleague, Alyssa

“so this question is what feedback will I be giving this person in their first performance review it's an amazing question because the person can't…”

Paul Adams · 74:30
#hiring#interviewing#management#tool

Takeaway· 5

Takeaway17:30

You Just Have to Make the Time to Try AI

For busy people who feel they don't have time to go deep on AI, Adams says there's no shortcut — it's about priorities, not crazy hours. Set aside dedicated time to read and actually play with the tools (like ChatGPT and GPT-4 Vision), or it'll pass you by, the same way the mobile wave left slower companies behind.

  • It's about priorities, not working crazy hours (he thinks beyond ~50 hours/week you make bad decisions)
  • Reading and hands-on play are non-negotiable to stay up to date
  • Upgrade to access newer capabilities like GPT-4 Vision and actually experiment (e.g. photographing food to assess it)
  • Like the mobile wave a decade ago, failing to explore AI properly will leave you behind

“so like my advice to people is you've got to try You' got to set aside the time or it'll pass you by”

Paul Adams · 19:00

“if people don't explore AI properly it it will leave them behind”

Paul Adams · 19:30
#ai#learning#productivity#advice
Takeaway21:00

Go Back to Basics: Map What Your Product Does Against What AI Can Do

Rather than getting sidetracked by cool AI tech, Adams advises starting from the core premise of your product — what problem it solves and why people love it — then asking whether AI can do that. For some products AI is a full replacement, for others augmentation, and mapping product jobs against AI capabilities (write, summarize, answer queries, scan images, take actions) tells you where you sit relative to the 'meteor.'

  • Start from the product's core premise, not the shiny technology
  • Ask whether AI can do the job fully, partially, or not yet
  • AI can write, summarize, answer queries, find facts, scan text and images, and increasingly take actions
  • Any product with a workflow (most B2B SaaS) or multimedia is 'in the meteor's path'

“I'd start with uh the thing your product does like what's the core premise behind it why do people use it you know what problem…”

Paul Adams · 21:30
#ai#product-strategy#framework#b2b-saas
Takeaway36:30

Don't Bolt AI On — Have Everyone Learn It

Drawing a direct parallel to the mobile wave (when the mobile team was a marginalized side group nobody believed in), Adams warns against treating AI as a separate bolted-on team. Beyond deep ML specialists, Intercom tries to have everyone learn about AI, reflecting his broader belief in generalists who can pick up new things.

  • Mobile was once dismissed as a small side team — a warning against marginalizing AI the same way
  • You still need great ML engineers as a deep specialism to build truly great things
  • But don't isolate AI in a dedicated team; have everyone learn about it
  • Adams is a strong believer in generalists and setting teams up that way

“what we're trying not to do is have like the the kind of like AI inbox team who and they're the only people who work…”

Paul Adams · 38:00
#ai#team-structure#generalists#leadership
Takeaway48:30

Deliberately Read the AI Skeptics to Balance Your Optimism

Adams admits he's a 'delusionally optimistic' thinker who has jumped on bandwagons before, so he deliberately reads people who are skeptical of AI or think it's bad for humanity. Seeking out the alternative opinion is his way of balancing optimism with negativity and avoiding another wrong bandwagon.

  • He's been guilty of jumping on wrong bandwagons in the past
  • He notably never bought crypto and questions what web3 even was
  • He now reads skeptics and people who think the technology is dangerous
  • The goal is to balance his natural optimism with counterweight negativity

“and I try these days to read the alternative opinion uh people who are skeptical or or think it's bad”

Paul Adams · 49:00

“I'm kind of a delusionally optimistic thinker so I try and balance that with negativity”

Paul Adams · 49:30
#ai#critical-thinking#learning#advice
Takeaway75:30

Two Mottos on a Monitor Post-it: 'Only Work on What Matters Most'

Adams keeps two mottos on a post-it stuck over the bottom of his monitor so he's forced to look at them: 'only work on what matters most' and 'stop worrying about things you can't control.' They lower his temperature when he wakes up to a stressful email and stop him firing off dumb, red-energy replies.

  • He physically covers monitor real estate with the post-it so he can't ignore it
  • Motto one: only work on what matters most
  • Motto two: stop worrying about things you can't control
  • He links the second to Seven Habits' circles of control and influence

“only work what matters most stop worry things you can't control it just like reduces the temperature”

Paul Adams · 76:30
#productivity#focus#mottos#self-management