For people who think in systems.

The first metacognitive AI.

Cognitive AI reasons. Metacognitive AI examines the reasoning — its frames, its ground, its blind spots — so you can navigate the problems that resist simplification.

Keywords: metacognitive AI · wicked problems · sensemaking · integrative inquiry

“The formulation of a wicked problem is the problem!”
— Rittel & Webber, Policy Sciences 4 (1973)
§ 01 · Why metacognitive

Software has learned to reason. Paradoxically, that’s making leaders’ lives harder.

You use cognitive AI every day. On the problems that keep you up, you’ve felt where it stops — a confident answer that flattens what you needed to hold, blind to its own frame and to what you can see.

Table 1 — Three kinds of software.
Pre-cognitive softwareRan what it was told.
Cognitive AIReasons.
Metacognitive AI1Examines the reasoning.

Alethic is the third kind. It surfaces its frames, shows its ground, names its blind spots — and leaves the judgment with you.

§ 02 · Why we exist

The problems that matter most aren’t information problems.

The defining challenges of this era — climate transition, institutional trust, demographic change, the legitimacy of governance, the coherence of a pluralistic society — do not suffer from a shortage of information. We are drowning in it.

They suffer from something harder: too few people able to hold contradictory truths at once, reconcile incommensurable values, and act under irreducible uncertainty. These are what Horst Rittel called wicked problems2 — problems that resist definition, have no stopping rule, and whose “solutions” are judged not as true or false but as better or worse.

General-purpose AI is impressive, and getting better fast. But it hits a ceiling on problems like these — and the ceiling isn’t intelligence. It’s how the tool relates to complexity. Cognitive AI resolves complexity to hand you an answer; for a wicked problem, a premature answer is worse than none, because it makes the problem look tamer than it is.

More intelligence won’t get you through this. A different kind of tool will.
§ 03 · What Alethic actually does

Structurally different — from the first principle.

01

It integrates. It doesn’t just retrieve.

Most AI answers by finding and combining existing content. Alethic holds multiple frames, traditions, and disciplines in tension — and helps you see where they converge, where they clash, and where the clash itself is the signal.

02

Designed for the conversation, not the output.

Alethic is a single thinking surface — from a quick orientation in the middle of a working day to deep analysis on a problem that plays out over years. You stay in the driver’s seat of your own thinking. Alethic sharpens it.

03

Not a neutral tool. A considered one.

Every AI encodes a worldview; most pretend they don’t. Alethic draws openly on the most rigorous integrative traditions of human inquiry, and tells you when it’s using them. You can disagree with it. You cannot be deceived by it.

§ 04 · One capability, many problems

Not eight features. One capability.

These aren’t eight features. They’re one metacognitive capability — reasoning on the frames, grounds, and blind spots of thought — pointed at the jobs that matter most.

For your own thinking

01Navigate a wicked decision.

Hold competing frames in tension, see what your current framing is hiding, and act under irreducible uncertainty without collapsing the problem prematurely.

02Pressure-test your own thinking.

Surface the premises beneath a strategy and stress-test them — the governing assumptions you can’t see from inside your own frame.

03Reason across disciplines you don’t have.

Integrative synthesis no single field reaches — a partner that knows where traditions converge, where they clash, and where the clash is the signal.

04Reach a position you can defend.

Produce a judgment you can stand behind and show the work for: named frames, disclosed lineage, examined assumptions.

05Communicate depth without losing it.

Translate complex reasoning into the register a board, a funder, or the public can receive — without dumbing it down or laundering the nuance out.

As you bring it into your institution

06Legitimate an AI-informed decision.

Give the institution the provenance, disclosure, and defensibility to stand behind a judgment — before a board, a regulator, a court, or the public.

07Govern the epistemics of AI use.

A metacognitive layer over the organization’s cognitive-AI use: what frame it reasoned from, on what ground, and what it structurally cannot see.

08Build shared institutional reasoning.

A persistent, examinable model of a complex situation the institution refines over time — reasoning that compounds and can be inspected, not one-off chats.

§ 05 · How it works

A surface for structured thought.

One chat. Two depths.

Ambient modes for working-day cognition. Deep analysis for the problems that deserve more. One surface — the depth is yours to set.

Designed around the human, not the model.

The interface never pretends the AI is a person, a colleague, or an oracle. It is a surface for structured thought. You are the subject; Alethic is instrumental — and you will see and think more deeply for using it.

Depth by design.

When Alethic reasons from a scientific, philosophical, or knowledge tradition, it tells you which one. When it surfaces a frame, it names its genealogy. You always know what you’re working with.

§ 06 · Where we draw the line

AI can’t legitimize itself. You can.

There is a quiet error running through the frontier of AI — the assumption that a system can be its own source of reliability, its own final authority on whether its reasoning holds. It can’t. Legitimacy routes through human agency, or it isn’t legitimacy at all. Alethic is built on the other side of that line: it puts everything it can see in front of you, and leaves the judgment — and the authorship of the judgment — where it belongs. That is why the more capable the AI becomes, the more it matters that a human remains the one who decides.

§ 07 · Who it’s for

Not a demographic. A developmental profile.

Alethic isn’t built for a demographic. It’s built for a developmental profile — the person whose interior coherence already exceeds what the systems and institutions around them can support. Who thinks in systems. Holds paradox without collapsing it. Cares more about getting things right than about appearing to have already gotten there.

These are people who live in an inverse complexity gap3: more complex than their surrounding environment, rewarded for simplification, and quietly punished for what they actually see.

If your problems don’t fit the tools you’ve been given, they were probably never designed for you.

§ 08 · Pricing

One membership. The whole instrument.

One membership opens every part of Alethic. Start free for 3 days or 15 messages, whichever comes first — no card required. Add payment when you're ready for unlimited access.

Alethic Membership
$49/ month
Excludes VAT / sales tax, added at checkout based on your location. Cancel anytime.
  • Ambient and deep-analysis reasoning in one surface
  • Frames, grounds, and blind spots surfaced on every answer
  • Answers grounded in the integrative knowledge graph
  • Unlimited messages — no trial cap
Start free
— the Alethic team
References
  1. 1Flavell, J. H. (1979). Metacognition and cognitive monitoring: A new area of cognitive-developmental inquiry. American Psychologist, 34(10), 906–911.
  2. 2Rittel, H. W. J., & Webber, M. M. (1973). Dilemmas in a general theory of planning. Policy Sciences, 4(2), 155–169.
  3. 3Kegan, R. (1994). In Over Our Heads: The Mental Demands of Modern Life. Harvard University Press.