Tacit judgment
The most valuable part of expert performance is rarely written down. It sits in pattern recognition, exception handling, and what gets weighted when conditions change.
Expert capability systems
A Blackkite companyShape the system that turns expert judgment into reusable AI capability.
Skillify is building AI using expert-derived reasoning, starting with how macro PMs update scenario odds after new events. We are speaking with a small number of experts to help shape the first capability.
No generic demo. We will make the conversation relevant to your domain and workflow.
Why experts matter
Generic AI can summarise. It does not reliably know what matters in edge cases.
What generic AI misses
Foundation models can restate the obvious. They are less reliable when the job is to notice which assumption just broke and how that should update the decision.
Why this matters now
If expert judgment is not captured, validated, and kept usable, it remains fragile, expensive, and hard to deploy where it matters most.
How Skillify works
Make the mechanism explicit, then test it against reality.
Capture
Experts walk through real decision moments, not vague abstractions.
Compile
Reasoning becomes structured heuristics, scenario logic, and reusable artifacts.
Evaluate
Outputs are tested against generic AI and baseline workflows.
Deploy
The capability becomes usable in a real workflow rather than a research memo.
Improve
Expert review and new cases strengthen the system over time.
What we are building first
A concrete starting use case, not a broad AI promise.
Given a PM’s pre-event scenario table and a new macro event, the system updates regime probabilities and explains the reasoning shift better than a generic model.
- Updated scenario odds
- What changed
- Which assumption broke or strengthened
- Confidence level
- Optional second-order implications
Scenario update artifact
Illustrative v1 outputPre-event
| Scenario | Odds |
|---|---|
| Sticky inflation | 40% |
| Soft landing | 35% |
| Growth scare | 25% |
Event
CPI misses on the downside. Wage detail softens. Services breadth improves.
Post-event
| Scenario | Odds |
|---|---|
| Soft landing | 48% |
| Growth scare | 28% |
| Sticky inflation | 24% |
Reasoning shift
- Inflation persistence evidence weakens
- Cut path becomes more defensible
- Growth concern remains a tail, not the base case
- Confidence: moderate
What we are asking from experts
A selective interview process with concrete expectations.
Intro call
5–15 minutes to confirm fit, relevance, and whether the starting use case is worth your time.
Expert interview
45–60 minutes focused on real decision moments rather than generic opinion.
Historical case walkthroughs
1–3 concrete examples. No polished deck required. Cases can be anonymised.
Optional follow-up
Only if useful to both sides. Deeper involvement is possible, not assumed.
What experts get
Real influence on an important capability layer.
Shape the first capability
Help define how expert reasoning is actually captured, structured, and tested.
Influence the evaluation standard
Push the system toward what matters in live judgment, not generic model theatre.
Early access to outputs
See how your reasoning translates into artifacts and workflow tools.
Potential deeper involvement
For the right experts, continued collaboration or founding-expert status may make sense.
“Artificial intelligence is not a substitute for human intelligence; it is a tool to amplify human creativity and ingenuity.”
Skillify is built around that premise: expert judgment matters, and the goal is to make it more legible, testable, and useful without flattening what makes it valuable.
Knowledge handling and trust
How expert knowledge is handled matters from the start.
Confidential by default
Interviews, case walkthroughs, and draft artifacts are treated as private unless a different agreement is made explicitly.
Provenance matters
We care about where a heuristic came from, how it was derived, and what confidence belongs around it.
Expert control, not extraction theatre
The goal is not to take expert knowledge out of context. It is to structure it in a way that remains useful, reviewable, and responsibly deployed.
Who is behind this
Built inside Blackkite Ventures.
Skillify is being developed within Blackkite Ventures as a focused effort to convert tacit expert judgment into validated AI capability.
Gary Pratt
Founder
gary@blackkite.ventures Former macro PM with 10 years of portfolio-management experience, including BFAM Partners and ExodusPoint.Why this exists
The best workflows often depend on judgment that is real, valuable, and largely undocumented. Skillify is an attempt to preserve and operationalise that layer without pretending generic AI already solved it.
FAQ
Why are you reaching out to me?
Because the first Skillify workflow is narrow and expert-specific. We are not trying to talk to everyone. We are trying to speak with people whose judgment is difficult to replicate with generic AI.
How long is the interview?
A standard flow starts with a 5–15 minute intro call, followed by a 45–60 minute expert session if there is clear mutual fit.
How will my information be handled?
Confidential by default. Examples can be anonymised, and nothing is published or shared publicly without agreement.
What exactly are you building?
The current starting use case is a scenario-update capability for macro workflows: given a pre-event scenario table and a new event, the system updates the odds and explains why the reasoning should shift.
Is there compensation?
Where compensation applies, it is discussed directly. We do not imply economics that have not actually been agreed.
Will my name be public?
Not by default. Attribution and public reference only happen with explicit agreement.
We are speaking with a small number of experts now.
If your work depends on judgment that generic systems do not reliably capture, we would value the conversation.