ExpertsFMWhat Makes a Consumer AI Product Stick? | Josh Elman
The a16z Show · September 19, 2026
5 Selected Lessons:
heuristic · 3 framework · 1 priority · 1
Elman's ranked trends he's most excited to back in consumer AI
Source
The a16z Show — What Makes a Consumer AI Product Stick? | Josh Elman (September 19, 2026) · at 14:23 ▸ · at 17:04 ▸ · at 20:21 ▸ · at 30:19 ▸
When this applies
when allocating limited attention/capital across multiple plausible emerging categories
The takeaway
When many categories look promising, rank them by proximity to solving a persistent, clearly-felt daily friction for the end user.
Why it works
He orders categories by how directly they reconnect digital tools to real human experience and relationships, starting from personal agents (most direct daily utility) to interactive/generated entertainment to vertical AI-powered services (travel, shopping, home).
One level up
Prioritization under uncertainty favors bets that visibly reduce daily friction over bets that are merely technologically novel.
Where it stops holding
This stops working when a technically novel category creates no felt daily friction reduction for users, regardless of how exciting it is to builders.
Detail
hierarchy:
- Personal agents (health, finance, scheduling, daily check-ins)
- Interactive/generated entertainment (AI microdramas, user-shaped story worlds, Roblox-like creation)
- Vertical AI services (travel, shopping, fashion, home) with concierge-style personalization
- AI-facilitated human connection (role-play prep, agent-to-agent coordination for real-world meetups)
rationale: He walks through these in order, starting 'I start with being really excited about the personal agent' then 'the second area I get really excited about is the future of entertainment.'
In their words
There are so many ways that AI is going to reinvent the experiences that we do every day
Earn the right to do more (step-by-step trust building)
Source
The a16z Show — What Makes a Consumer AI Product Stick? | Josh Elman (September 19, 2026) · at 33:16 ▸
When this applies
when introducing any product that eventually wants to handle many tasks but currently must start narrow
The takeaway
Scale trust and scope incrementally: perfect a small set of use cases before expanding, since each win earns permission for the next ask.
Why it works
Nailing a few narrow tasks builds a track record of reliability; each success expands the scope users are willing to delegate, while early overreach destroys trust before it's earned.
One level up
Ambition should be sequenced, not declared — big platforms are built by compounding narrow wins, not by claiming broad capability upfront.
Where it stops holding
This stops working when a narrow use case has no natural adjacent expansion path, trapping the product in a niche forever.
Detail
procedure:
- Pick a few tasks and execute them extremely well
- Let success in those tasks earn user trust for a slightly broader scope
- Repeat, expanding scope only as fast as trust is earned
- When the product can't do something, explain clearly rather than failing silently
benchmarks: {}
rationale: Discord earned the right to be a chat hub by first nailing game-voice-chat; Siri's incremental utility (calls, timers, reminders) is contrasted with its failure when users tried to extend trust too far too fast.
In their words
I'm a big believer that if you want to become a very big company, you can't start by trying to do everything, but you have to start by doing a few things really, really well and earn the right to do more.
You can clone the product, not the trajectory
Source
The a16z Show — What Makes a Consumer AI Product Stick? | Josh Elman (September 19, 2026) · at 48:09 ▸
When this applies
when any fast-follower can replicate your visible output almost instantly
The takeaway
Defensibility in fast-copy markets comes from staying ahead on vision and deepening relationship/identity with users, not from feature secrecy.
Why it works
Competitors can only copy what is observable today; they cannot see or replicate the founder's roadmap, accumulated user relationship, or evolving product personality.
One level up
When imitation is cheap, compete on direction and relationship rather than on the artifact itself.
Where it stops holding
This stops working when the product has no real differentiated vision or personality to build relationship around — then being copied simply means being replaced.
Detail
rule: Build personality, relationship, and a forward vision competitors can't see, rather than relying on feature novelty.
threshold: Cloning now happens in days, not months or weeks
rationale: Users stay with a product because it has 'the essence of something that was really special,' even after trying clones.
In their words
But I can only clone what I see. I can't clone where it's going and I can't clone the next thing.
AI inference costs demand new consumer pricing models
Source
The a16z Show — What Makes a Consumer AI Product Stick? | Josh Elman (September 19, 2026) · at 41:55 ▸
When this applies
when a product's marginal cost of serving each user is variable and rising rather than fixed and falling
The takeaway
When marginal cost per user action is non-trivial and variable, pricing strategy must shift from pure free-distribution toward converting engaged users to paying ones, timed to when costs and willingness-to-pay cross.
Why it works
Unlike the cloud-infrastructure era where marginal serving costs collapsed toward zero, AI inference imposes a persistent variable cost per interaction, so giving the product away free is not sustainable at scale the way pure software once was.
One level up
The right monetization moment in any business is when falling serving costs meet rising willingness-to-pay from proven product value — not before.
Where it stops holding
This stops working when model/inference costs continue to drop faster than user willingness to pay rises, in which case free distribution can remain viable longer than expected.
Detail
rule: Expect ad, purchase, and subscription models to combine once falling inference costs intersect with consumers' now-higher willingness to pay for digital value.
threshold: n/a — described as an intersection point 'over the next couple of years'
rationale: Consumers are now much more used to paying for digital goods (mobile gaming, subscriptions) than 10-30 years ago, unlike enterprise where expensing softens payment friction.
In their words
And I think over the next couple of years, we're going to intersect with how much cheaper these can get to build and serve, with how much customers are willing to pay for things that are of value or entertainment for them.
Bay Area for scaling, not necessarily for starting
Source
The a16z Show — What Makes a Consumer AI Product Stick? | Josh Elman (September 19, 2026) · at 51:34 ▸
When this applies
when deciding whether to locate a venture in an industry's established hub versus elsewhere
The takeaway
Separate the question of where to originate an idea from the question of where to access the specific expertise needed for the next stage of growth — proximity matters most for the capability you currently lack, not for validation of the idea itself.
Why it works
Great ideas and early execution can originate anywhere (Sweden, China, etc.), but the dense concentration of people who've personally lived through scaling a startup into a durable platform gives outsized access to that specific expertise in one place.
One level up
Resource/location decisions should be driven by which specific capability-gap you need filled next, not by geographic prestige or habit.
Where it stops holding
This stops working for consumer products whose success depends heavily on deep cultural fit with a non-SF market, where importing 'SF tech culture' could actually hurt mainstream adoption elsewhere.
Detail
rule: I don't think it's the best place in the world to start a company, but I think it's a very important place in the world to scale a company.
threshold: n/a
rationale: Notes B2B companies especially benefit from SF concentration of buyers, while consumer companies must also understand cultures outside SF to reach mainstream users.
In their words
I don't think it's the best place in the world to start a company, but I think it's a very important place in the world to scale a company.