Asana initially emphasized experimentation and first-principles learning rather than committing prematurely to performance targets. Metrics became more useful only after early usage supplied realistic baselines for the scale phase.

Premature reporting can displace necessary exploration
A new technology may require internal capability building
Early data points enable more realistic targets later
Headline AI targets may be arbitrary or deliberately easy

“if we overcommit on metrics and we just start reporting on metrics all the time we're not actually going to like spend the time learning”
Ethan Dewal · 29:30

“AI metrics are throwing spaghetti against a wall”
Ethan Dewal · 30:00

#ai metrics#experimentation#organizational learning#roi

Hot Take30:30

The Two-Year Bet: An Always-On Engine Creates the 10x Rep

Dewal expects AI to affect the entire sales cycle rather than one isolated stage. An always-on system could process CRM changes, direct attention toward the right accounts, and reduce administrative work enough for representatives to spend most of their time selling.

No representative can manually understand every change across hundreds of accounts
AI can continuously process CRM and account signals
Painful manual work exists at every sales stage
The target is moving selling time from roughly 30% toward 70% or 80%

“in two years the 10x sales rep will exist”
Ethan Dewal · 30:30

“hopefully instead of 30% of their time being done selling we can get back to 70 or 80%”
Ethan Dewal · 32:00

#future of sales#copilots#crm#sales productivity

Explainer· 3

Explainer02:00

AI Created a Third Path for Personalized Outbound

Traditional outbound forced sellers to choose between slow one-to-one research and scalable but generic sequences. GPT-4 revealed a third option: producing persona-aware, individually personalized messaging efficiently enough to use at scale.

Manual one-to-one outreach could take roughly 30 minutes per prospect
Sequence-based outreach sacrificed individual relevance for volume
AI made scalable one-to-one personalization conceivable
The seller still needed a system that produced consistent results

“that day I saw the third pathway which is you could do Persona based hyper personalized messaging one to one very efficiently”
Ethan Dewal · 03:00

“you just had to figure out the system to get it to work for you”
Ethan Dewal · 03:00

#outbound#personalization#sales#gpt-4

Explainer18:00

Automation Removes the Manual Handoffs Around AI

Early AI tools made users collect data, paste it into a chatbot, and move the response back into another system. Automation improves the experience by triggering an already validated capability and delivering its output without those handoffs.

Manual prototypes put integration work on the user
An upcoming meeting can trigger call preparation automatically
Automation should surround a capability that users already value
Prompts plus automations can become custom internal software

“every time I have a meeting trigger that call prep uh uh uh document to be created in in an automated way”
Ethan Dewal · 19:00

“you're kind of building custom AI software for yourself as soon as you hit this uh automation phase”
Ethan Dewal · 19:00

#automation#workflows#sales enablement#internal tools

Explainer24:00

The Practical AI Value Equation: Highest Quality in Seconds

Dewal frames sales-tool value through the activity requirements of a role. The strongest tools produce the quality associated with careful expert work while compressing the effort from an hour or more to seconds or minutes.

Sales activities have expected volume and quality levels
High-quality execution is limited by the hours available
AI can combine institutional knowledge with data processing
The emotional reaction comes from seeing formerly manual work disappear

“if they did that activity at the highest quality that's what I want to prod for them in seconds”
Ethan Dewal · 24:00

“that is literally an hour of effort that Humanity has never been able to like automate and like now we can”
Ethan Dewal · 26:00

#roi#productivity#sales operations#time savings

Story· 1

Story34:00

Eight Stuck Waymos Illustrate What Humans Still Handle Effortlessly

Dewal describes autonomous vehicles becoming gridlocked in a parking lot because none could negotiate who should move first. Humans routinely resolve similarly messy, low-speed situations while accounting for numerous subtle risks, illustrating the remaining gap in adaptable real-world judgment.

Autonomous vehicles delivered a strong ride experience in normal conditions
A small coordination problem caused multiple vehicles to stall
Humans handle ambiguous parking interactions with little conscious effort
Competence in controlled tasks does not imply mastery of open-ended complexity

“there was like a big traffic jam of like eight of them they couldn't figure out how to like turn past this little thing”
Ethan Dewal · 34:30

“we handle that just so effortlessly a literally could not think to handle that complexity”
Ethan Dewal · 35:00

#autonomous vehicles#human judgment#ai limitations#waymo

Q&A· 1

Q&A32:30

Will Customers Buy From an AI Avatar That Sells Better?

The hosts ask whether an AI avatar could outperform a human in lower-complexity sales by knowing every product detail and answering consistently. Dewal declines to make a long-term prediction until another major model leap reveals how well AI can handle deeper, unpredictable interactions.

AI avatars may improve on chatbots and static knowledge bases
Breadth of product knowledge favors machine assistance
Customers may still prefer humans despite an inferior transactional experience
A live sales pilot is needed to test actual buyer behavior

“until I see that next jump I'm not placing a lot of long-term bets of what I think AI could or could not do”
Ethan Dewal · 33:30

“it's just an unknown problem until you actually run a pilot and have an avatar try to sell to some of your users and look…”
Kieran Flanagan · 37:00

#ai avatars#customer experience#sales automation#human interaction

Takeaway· 2

Takeaway03:30

Transparency Helps Sales Teams Give Up Control

AI skepticism declined when employees learned how the tools were constructed, where their data came from, and what their limitations were. Customization also gave users ownership instead of forcing them to trust an unexplained black box.

Teach users how the tool was built
State capabilities and limitations explicitly
Allow small adjustments for individual workflows
Pair technical power with a simple user experience

“once kind of our team understood that this isn't a magic black box like this is how we built it this is what it can…”
Ethan Dewal · 05:00

#adoption#education#change management#trust

Takeaway22:00

If Sales Value Is Not Obvious in Seconds, Adoption Will Stall

Salespeople protect workflows tied directly to their quotas, so an AI launch needs a concrete and immediate payoff. A screenshot of the output, a minimal trigger, and a clear statement of the time saved can communicate value more effectively than a lengthy rollout.

Sales teams resist workflow disruption because they are accountable for results
Launch messaging should show the exact output users will receive
The interaction should require very little effort
Time-to-value must be immediate and emotionally obvious

“this is what you'll get if you fill out this form”
Ethan Dewal · 23:00

“if you can't make the like 20 to 30 second like Tik Tok influencer style clip of what this is and the value and why…”
Kieran Flanagan · 25:30

#sales adoption#change management#launches#time savings

Frameworks in this episode

Four-Part Bottom-Up AI Roadmap

Goal-Example-Data Prompt Formula

Pain-First AI Use Case Selection

Insights & moments

◆Hot Take· 2

New AI Programs Should Prioritize Learning Before Metrics