Let AI build it.
Then catch up.
Get to where you can explain and judge the code, analyses, and plans AI produced for you. Our experts ask questions in the areas you haven't noticed — closing the comprehension gap that AI-assisted work leaves behind.
“It works, but nobody can explain it”
With AI, the cost of building is heading toward zero — but the cost of understanding hasn't changed. Left alone, working software and documents that nobody can explain — what we call comprehension debt — pile up by structural necessity. And the tricky part: until someone asks, you can't see what you don't understand.
The illusion of understanding
With a finished deliverable in front of you, it feels like you could explain it. Then someone asks “what happens with this input?” — and you discover you can't.
AI-dependent organizations
Every incident and spec change ends in “ask the AI that wrote it.” Engineers stop growing review and design judgment, and the codebase lives in nobody's head.
Unexamined decisions
AI-produced analyses and proposals flow into executive decisions without their assumptions or evidence ever being checked. Deciding on AI's say-so is a business risk in itself.
Questions that raise understanding through four levels
We call the share of a deliverable you can explain, predict, modify, and critique its cognitive coverage — think of it like test coverage. Your coach analyzes the deliverable, finds your blind spots, and asks questions pitched at the level you haven't reached yet. We don't hand out answers — working through a question is the fastest way understanding actually sticks.
Level 1 — Explain
Say what it does
“Can you state this module's responsibility in one sentence?”
Level 2 — Predict
Anticipate its behavior
“What does this do when the input is empty?”
Level 3 — Modify
Know where to make a change
“If the requirement changed like this, what would you touch?”
Level 4 — Critique
Name its weaknesses and alternatives
“Under what load or requirements does this design break down?”
Why a person asks the questions — not an AI
Reading between the lines
A written answer can be produced by AI, and a superficially correct answer doesn't prove real understanding. Pauses, hesitation, the order of an explanation — a human interlocutor reads these, judges what you actually understand, and follows up on the spot. It's the same reason oral examinations have been held face-to-face for centuries.
Questions sharpen over time
Where you tend to get stuck, which questions draw you in, how far you can stretch — these accumulate in a coach who works with you continuously. With every session, the questions land more precisely on the spot just beyond your reach.
AI works backstage
We do use AI — before each session, Tied's internal tooling analyzes the deliverable's structure and drafts candidate blind spots and questions. Your coach selects, refines, and focuses on the conversation. People outside, AI inside — that's the design principle of this service.
Two programs
For Engineers
Catch up with your code
Your own pull requests become the curriculum. Regular oral-exam-style sessions on AI-written code build the skill that matters now: verifying and steering, not just producing.
- ・Questioning sessions built on your PRs and design docs
- ・Team reports that surface “areas nobody can explain”
- ・Question sets reused as onboarding material for new members
For Executives
Catch up with your decisions
Judge AI-produced analyses, plans, and proposals instead of taking them at face value. Sparring sessions before important meetings work through assumptions, evidence, alternatives, and risks.
- ・30-minute sparring before board and leadership meetings
- ・Organizational “comprehension debt” assessment reports
- ・For investors: portfolio comprehension-debt reviews (with tech DD)
How it works
01 Contact us
Reach out through the contact form. A vague feeling like “we're shipping with AI, but the team isn't keeping up” is a perfectly good starting point.
02 Agree on scope and target depth
Together we decide which deliverables (code, documents) to cover and how deeply they should be understood, triaging by importance.
03 Questioning sessions
We analyze the deliverable beforehand and design the questions, then meet for a 30–60 minute session. We don't hand out answers — we probe and follow up as you reason through them.
04 Records and reporting
We track how understanding develops over time, with personal retrospectives for individuals and area-level aggregate reports for teams and leadership.
Frequently asked questions
If AI can answer these questions, why have a human ask them?
Questions about areas you haven't noticed can't come from your own prompts — blind spots are blind precisely because you can't see them. In a face-to-face conversation, pauses, hesitation, and the order of an explanation reveal whether you truly understand, and your coach can follow up on the spot. And because the same coach works with you over time, they learn where you tend to get stuck, so the questions get sharper with every session. None of this can be replicated by a chatbot.
Is this only for engineers?
No. We offer a program for executives and business leads, focused on interrogating AI-produced analyses, proposals, and plans — their assumptions, evidence, alternatives, and risks — as a sparring session before important decisions.
How much time does it take?
Sessions run 30–60 minutes, typically two to four times a month. We don't cover every deliverable — we triage by importance (change frequency, blast radius in an incident) so the program fits alongside your day-to-day work.
Will session results be used for performance reviews?
This is a learning service, not an assessment. For team engagements we contractually agree on disclosure rules: individual details are visible only to the person and their coach, while teams and leadership see aggregates by area of the codebase or business only.
How much does it cost?
Pricing depends on the scope of deliverables covered, session frequency, and reporting. Contact us and we'll propose a plan after an initial conversation.
Let AI do the building.
Keep the understanding.
“I feel like the team isn't keeping up with the AI” is a perfectly good
place to start.
Tell us where you are, and we'll propose the right program.