Pull the Thread on New Tools
Judge a new AI tool by where it'll be in a week or a month, not where it is on day one.
- Difficulty
- Easy
- Time to result
- ~weeks to results
- Steps
- 3
- Confidence
- 88%
New AI tools ship broken, with sharp edges, and are easy to dismiss on first use. The disciplined evaluation is to feel for underlying product-market fit — the flash of joy or utility beneath the bugs — and if it's there, keep returning week after week to judge the tool by its trajectory rather than its current state.
Origin
Claire Vo, a self-described anti-hype skeptic, from her arc on OpenClaw (which deleted her calendar on day one) and her earlier reversal on Claude co-worker.
Core principles
- 01A rough tool with a spark of joy signals product-market fit that hasn't been reached yet
- 02Distinguish 'not useful' from 'useful but broken' — only the first is a real reject
- 03Evaluate on trajectory: where it is in a week and a month, not today
- 04Returning repeatedly, not one trial, reveals the unlock
How to run it
- 1
Feel for the spark beneath the bugs
On first use, separate whether the tool is fundamentally not useful versus useful but broken. A jolt of joy or utility amid the failures is the 'ugly, apparent feeling of product-market fit.'
Pro tip Complaints like 'it's buggy / it forgot / it can't do X yet' indicate a product that hasn't caught up to its PMF — not an absence of PMF.
Watch out Don't write a tool off wholesale on a bad first session; Claire nearly did this with Claude co-worker before finding the unlock.
- 2
Return week after week
If the spark is there, commit to revisiting the tool repeatedly over days and weeks rather than judging from one install, because these tools improve fast.
Pro tip An 'accountability cost' helps — spending $500 on a Mac mini made Claire actually follow through on setup.
- 3
Extrapolate the trajectory
Make the keep/drop decision based on where the tool is heading — a week and a month out — taking as given that it will improve, not on its rough current state.
Pro tip If the only gaps are 'broken' rather than 'not valuable,' the trajectory is favorable — bet on it improving.
In the wild
Claire's first OpenClaw install took eight hours and deleted her family calendar. Instead of quitting, she recognized the underlying utility, kept returning, and — as the tool matured — went from a leading skeptic to running nine agents across three machines. She'd nearly dismissed Claude co-worker the same way before it clicked.
→ A tool she'd have rejected on day one became, in her words, the most mind-blowing AI experience since ChatGPT.
Common mistakes
Writing a tool off after one rough session
Early AI tools are buggy by nature; judging them on a single bad first use throws away tools that are on a steep improvement curve and about to become genuinely useful.
Confusing 'broken' with 'no product-market fit'
Complaints that it's buggy or forgetful describe a product that hasn't caught up to its PMF, not one that lacks it — mislabeling the two leads you to abandon the right tools.
Is it for you?
Best for
Operators and builders deciding which of the flood of new AI tools deserve sustained investment
Not ideal for
Situations needing a reliable production tool right now, where trajectory can't compensate for present-day breakage
From the transcript
“you really have to pull the thread on these tools and you have to spend enough time with them to see not where they are…”
“it just hit me with enough joy and a enough utility when it wasn't deleting my calendar that I knew something was there”
“That's not not product market fit. That's a product that hasn't caught up to its product market fit”
From the episode
From skeptic to true believer: How OpenClaw changed my life
Claire Vo