Lowercase-c vs Capital-C: AI-Proofing Your Skill Stack
Sort every task into what AI does 80% well and what only a human can do — then invest accordingly
- Difficulty
- Moderate
- Time to result
- ~ongoing to results
- Steps
- 5
- Confidence
- 91%
Vo's triage for which professional skills AI will absorb and which it will amplify. She splits communication into lowercase-c — the functional trading and synthesis of information so other people can do their jobs — which language models are already excellent at, and Capital-C — being influential, convincing, bold, and getting a system of humans to follow you down a path — which is far harder to replace. She extends it: strategy synthesis is highly automatable (contrary to popular polling), while boldness and seeing a future that a system trained on priors cannot see are not. The operative question is never 'can AI do this?' — it's 'is 80%-good functionally tractable the best way to do this, or can a skilled human with a purpose-built toolkit make it 3x, 4x, 10x better?'
Origin
Claire Vo's own framework, developed while building ChatPRD explicitly as a way to stress-test how AI will change the product function she leads. She takes a position contrarian to a Lenny poll in which respondents voted communication most-replaceable and strategy least-replaceable — Vo argues strategy synthesis is exactly what these tools are exceptional at.
Core principles
- 01Short-term pessimistic, long-term optimistic: change is coming faster than anyone expects, and technology still makes society better.
- 02Lowercase-c communication (synthesizing information, coordinating who needs what, generating it across modalities) is going away.
- 03Capital-C communication (influence, conviction, boldness, charisma, making humans follow you) is very hard to replicate.
- 04Strategy — synthesizing what you know into a plan and communicating it — is highly automatable, contrary to what most people assume.
- 05Seeing the future is not something a thing trained on priors can do.
- 06Assume nearly every PM task is 80% good and functionally tractable by AI. The value question is whether 80% is the best you can do, or whether a human plus a purpose-built toolkit makes it 10x.
- 07Don't spend your magic on solved areas — scaffold those with AI and ship.
How to run it
- 1
List every job your role actually does
Write out the day-to-day: writing PRDs, setting goals, proposing a roadmap, aligning a team behind it, developing strategy and vision, communicating timelines, finding blockers and unblocking, getting buy-in from up high, securing budget and headcount, giving feedback on product and design.
- 2
Split each item into lowercase-c and Capital-C
Lowercase-c: the functional trading and synthesis of information that lets other people do their jobs. Capital-C: are you influential, are you convincing, are you bold, can you get a system of humans to follow you down a path. Nearly every job splits into some of both.
Pro tip Roadmap writing is mostly lowercase-c. Getting a room of skeptical executives to commit to the roadmap is Capital-C.
Watch out Do not assume strategy work is safe just because it feels intellectual. Synthesis-plus-distillation-plus-communication is precisely the model's strength.
- 3
Assume the lowercase-c parts are 80% tractable today
Treat every lowercase-c task as something AI can already do at 80% quality. Then ask the real question: is 80%-good the best way to do this, or can a certain type of person with a certain skill set, backed by a purpose-built toolkit, make it 3x, 4x, 10x better?
Pro tip For solved areas — Vo's example is data export to Snowflake — scaffold it with AI, customize to your context, and ship. That is not where your human magic changes the outcome.
Watch out The framing is not 'which of these can AI do' but 'which of these do you want to hand the keys to.'
- 4
Invest your development time in the Capital-C stack
Deliberately build boldness, conviction, charisma, influence, and the ability to see a future the priors don't contain. These are the durable differentiators — plus the judgment of whether you are building the right stuff, quickly, and whether it is delivering.
Pro tip The identity shift Vo expects: PM value stops coming from the ideas you personally manifest and how, and starts coming from whether the right thing shipped fast and worked — no matter the toolchain.
- 5
Reshape the team, not just your own workflow
As a leader, hold yourself to the bar of understanding what this technology disrupts, using the tools to build a better team, and actually shifting the size and shape of your organization in response. Vo cites a team that had to spend a week trying to automate any role before they were allowed to open the JD.
Pro tip Sustainability in organizations is the leader's responsibility. Inefficient hiring builds unsustainable cost and helps no one — a small startup extends its runway by not adding the PM it can't afford.
Watch out Many people don't want to hear that a team may not need to hire another PM. Lead from the front on it anyway rather than on the back foot.
In the wild
Lenny ran a poll on Twitter and LinkedIn asking which PM skill — communication, execution, strategy, or product sense — AI would most likely take over. Communication won by a mile; strategy was voted least replaceable. Vo takes the opposite view: strategy is 'here's everything we know about the world, competitors, the market and our advantages, here's a plan to win' — which is synthesis, distillation and communication, exactly what these tools are exceptional at.
→ Her conclusion: the durable human edge is not strategy-as-analysis but boldness, seeing a future not present in the priors, and the charisma to make it happen.
Vo built ChatPRD partly to stress-test how AI changes the product function she leads, and partly to learn how to build non-deterministic products. It began as a chat prompt she used to spec a complex technical product between the start and end of a single meeting, became a custom GPT she shared with her team, then a standalone app with per-user customized assistants.
→ Thousands of daily users; PMs report it saves dozens of hours of document writing; one single PM on a growing team reported they no longer expect to have to hire another PM. Vo herself uses it precisely for the solved areas where her human magic wouldn't change the outcome.
Common mistakes
Assuming your intellectual work is the safe part
Most PMs voted strategy least-replaceable. Vo argues it is among the most replaceable, because strategy as practised is synthesis and communication — the model's core competence. Investing your development time there is a bet against the trend.
Anchoring identity on manifesting your own ideas
Vo expects an identity shift: PMs today believe their value is the ideas they personally manifest and the way they do it. The value is moving to whether you are building the right stuff, building it fast, and whether it delivers — regardless of toolchain.
Spending your magic on solved problems
Nobody's genius is required to design Snowflake data export. Scaffold known-solved surfaces with AI, customize, ship — and spend human attention where boldness and judgment actually change the outcome.
Is it for you?
Best for
Product managers and product/engineering leaders deciding which skills to invest in over the next few years, and leaders planning team size and shape against AI capability
Not ideal for
Roles where the deliverable is regulated, safety-critical, or requires provenance/liability that an 80%-good generated artefact cannot carry
From the transcript
“communication lowercase C I feel like is one of the places that's going to be replaced”
“but Capital C communication of are you influential are you convincing are you bold”
“it's it's sort of the human aspect though of boldness seeing the future in a way that a thing trained on priors cannot”
“I believe all of those are 80% good functionally tractable the question is is 80% good”
From the episode
Bending the universe in your favor
Claire Vo (LaunchDarkly, Color, Optimizely, ChatPRD)