The Everypact Manifesto

On the Liberation of Legal Knowledge

by Elvinas Jonaitis

I. The Gate

[001]For centuries, access to expertise has been controlled by those who possess its keys. Philosophers from Foucault to Illich have examined how professional expertise becomes a mechanism of control, shaping what counts as legitimate knowledge, who may claim competence, and how ordinary people come to see themselves as dependent on specialists for matters that affect their own lives.

[002]The law, that most consequential of human systems, has been the most guarded domain of all. When knowledge is accessible only through sanctioned interpreters, interpreters become indispensable. The structure serves those who stand at the gate, and the gate justifies itself by its own existence.

[003]Yet gates are built. And what is built can be unbuilt.

II. The Paradox of Efficiency

[004]Consider the standard logic of legal efficiency: Give lawyers better tools, and justice becomes more accessible. This sounds right. But it conceals a flaw.

[005]The premises:

A.The price of legal services depends on perceived value, meaning what clients believe they're getting.

B.Perceived value depends on information asymmetry. When clients can't evaluate legal work themselves, they rely on the lawyer's judgment about what's needed.

C.In markets with information asymmetry, prices track perceived value rather than production cost. (analogy of an auto repair: if you can't tell whether the mechanic did two hours of work or twenty minutes, you pay for what they say they did.)

D.Better tools given exclusively to lawyers reduce the time required for legal work.

E.Tools held exclusively by lawyers do not reduce information asymmetry. The client still can't evaluate the work.

[006]What follows:

From A and B: if information asymmetry persists, perceived value persists.

From C: if perceived value persists, prices persist.

From D and E: lawyer-exclusive tools increase lawyer efficiency without reducing asymmetry.

Therefore: Efficiency gains from lawyer-exclusive tools accrue to lawyers. Clients pay the same, or more, for faster work. The gate remains.

[007]To change who benefits, we must change who holds the tools.

III. The Dissolution

[008]The structural conditions that made legal gatekeeping possible are eroding. Not through reform or good intentions, but through technological change that undermines the gate's foundation.

[009]The premises:

A.Gatekeeping requires controlling two things: access to knowledge and the means of applying it.

B.New computational architectures (transformers, embeddings, large language models) enable semantic understanding accessible to non-experts.

C.The marginal cost of distributing knowledge has collapsed to near zero. What once required law libraries now requires an internet connection.

D.Mature interface design can translate complex knowledge into navigable experience.

E.These capabilities are being deployed and adopted at increasing scale.

[010]What follows:

From B and E: the means of understanding legal knowledge are becoming accessible beyond trained professionals.

From C and E: the knowledge itself is reaching wider audiences.

From D and E: the experience of accessing that knowledge is becoming intuitive.

From A: when access to knowledge and means of application become widely available, gatekeeping loses its structural foundation.

Therefore: The conditions that sustained legal gatekeeping are eroding, not because anyone decided they should, but because the ground has shifted beneath them.

IV. The Distinction

[012]We must be precise, for precision serves justice. Not all legal work is the same. Not all complexity is artificial.

[013]The premises:

A.Some legal complexity is inherent to genuinely difficult problems: novel questions of law, high-stakes negotiations, cases at the edge of precedent.

B.Some legal complexity is artificial, an artifact of opaque language, convoluted processes, and information hoarding. Think of a simple lease agreement buried in eighteen pages of legalese, or a straightforward trademark filing wrapped in procedural mystery.

C.Artificial complexity can be dissolved through accessible tools, clear interfaces, and open knowledge.

D.Essential complexity requires trained expertise and experienced judgment.

[014]What follows:

From B and C: for matters involving artificial complexity, tools can eliminate the need for gatekeepers.

From A and D: for matters involving essential complexity, lawyers remain irreplaceable.

Therefore: The question is not whether lawyers are needed. The question is for what they are needed.

[015]We honor lawyers by distinguishing the work that truly requires their gifts from the work that a broken system assigned to them by default.

V. The Limits of Tools

[016]One might object: If tools grow powerful enough, won't essential complexity also yield? Isn't this just a matter of time?

[017]We must address this directly, because our answer is no, and the reasons matter.

[018]First argument: Why AI cannot resolve essential complexity

[019]The premises:

A.AI systems derive their capabilities from patterns in training data.

B.Essential legal complexity arises in genuinely novel situations: unsettled law, contested facts, irreducible uncertainty about how particular decision-makers will respond.

C.Patterns from past data cannot determine outcomes in genuinely novel situations. (this is by definition - if past patterns determined the outcome, the situation wouldn't be genuinely novel.)

D.At the frontier, legal questions become normative, concerning what justice requires rather than merely what courts have done.

E.Normative judgments cannot be derived from descriptive data alone. You cannot get "ought" from "is".

F.Legal counsel involves accountability. Someone must bear professional, legal, and moral responsibility for advice given.

G.AI and other automated tools cannot bear responsibility. It cannot be sued for malpractice, cannot face disbarment, cannot feel the weight of having failed a client.

[020]What follows:

From A, B, and C: AI cannot resolve the uncertainty inherent in essentially complex legal matters.

From D and E: AI cannot make the normative judgments that essential complexity demands.

From F and G: even if AI could produce equivalent outputs, it cannot fulfill the accountability function that counsel requires.

Therefore: The need for human lawyers in essentially complex matters is structural. It is not a temporary limitation waiting for better technology.

[021]Second argument: Review versus Analysis

[022]There is a deeper reason why essential complexity cannot be automated while artificial complexity can. It concerns two fundamentally different kinds of legal work.

[023]The premises:

H.Legal tasks divide into two categories. Review involves checking documents against known standards, identifying presence or absence of terms, and matching patterns to established criteria. Analysis involves extending existing holdings to new situations and reasoning about how principles apply to novel facts.

I.Pattern-matching produces reliable outputs for review tasks. (Does this contract contain an arbitration clause? Pattern-matching can find it.)

J.Analysis requires tracing reasoning through an extension, understanding why a principle reaches a new situation, not just whether it looks like it might.

K.AI systems produce outputs by pattern-matching. They generate text that resembles reasoning but do not trace reasoning through extensions.

L.Artificial complexity (opaque language, convoluted processes, information hoarding) consists of matters addressable by review against known standards.

M.Essential complexity (novel questions of law, cases at the edge of precedent) consists of matters requiring analysis of how law extends to new situations.

N.Legal standards for novel arguments require certification that the argument represents genuine reasoning.

[024]What follows:

From I and K: AI can reliably perform review. AI uses pattern-matching, and pattern-matching produces reliable outputs for review.

From J and K: AI cannot perform analysis. Analysis requires tracing reasoning through extensions. AI produces pattern-matched text, not traced reasoning.

From L and the first derivation: AI can handle artificial complexity, since artificial complexity consists of review-type work.

From M and the second derivation: AI cannot handle essential complexity, since essential complexity consists of analysis-type work.

From N and the second derivation: legal systems cannot accept AI outputs as satisfying certification requirements for novel arguments.

Therefore: The distinction between artificial and essential complexity aligns with the distinction between review and analysis. Tools are reliable for the former and structurally incapable of the latter.

[025]A US federal court recently captured this distinction with more clarity. When a lawyer overstates a case's holding, they can reason through how that holding might extend to novel situations. Even if their reasoning is flawed, it is reasoning. When AI overstates a holding, it is producing a plausible-looking sentence whose content may or may not be true. A lawyer's imperfect argument is still an argument. An AI's plausible-seeming analysis is a pattern that happened to fit.

[026]Tools can perform review. Tools cannot perform analysis. Tools can check against the known. Tools cannot reason into the unknown.

VI. The Core Argument

[027]Here is where the logic converges.

[028]The premises:

A.If tools are given exclusively to lawyers, information asymmetry persists and efficiency gains accrue to lawyers. (Section II)

B.If tools are given directly to people, information asymmetry dissolves for matters where tools suffice. (Section III)

C.Most legal matters that individuals and small businesses face involve artificial complexity, not essential complexity. (The overwhelming majority of legal needs are not Supreme Court cases or billion-dollar negotiations. They are leases, contracts, filings, disputes over knowable rules.)

D.Tools can now be built that handle artificial complexity with high fidelity. (Section III)

E.For matters involving artificial complexity, tools in the hands of laypeople can produce work of equivalent quality to tools in the hands of lawyers. (Section IV and V)

[029]What follows:

From A: the current approach, giving better tools to lawyers, preserves the gate.

From B, C, D, and E: a different approach, giving equally powerful tools to people, opens the gate for the majority of legal needs without sacrificing quality.

Therefore: Building tools for people, as powerful as what lawyers have, dissolves information asymmetry and transfers benefits to those who need legal help.

[030]This is the Everypact thesis.

VII. The New Possibility

[031]A world where a person facing a legal matter does not first face fear. Fear of cost. Fear of complexity.

[032]This is within reach.

[033]The tools exist. The interfaces invite rather than intimidate. The knowledge flows to those who need it, packaged for the person whose life depends on understanding. The workflows that once required professional intermediation now run through well-designed channels. The right document, the right process, the right understanding, arriving at the moment of need, at a cost proportional to actual value.

[034]And when essential complexity demands essential expertise, the connection to qualified counsel is seamless. A genuine matching of problem to solution.

[035]Lawyers are liberated to do the work that only lawyers can do. Everyone else gains access to what should never have required them.

VIII. The Commitment

[036]We are Everypact. We build for the people.

[037]Our commitment is to create tools for individuals and businesses that are as powerful as what lawyers have. When those tools are simply the best available, lawyers may use them too. We build excellence and direct it toward those who have been underserved.

[038]This means:

Curation without compromise. Assembling the best tools with judgment and making them accessible to all.

Context that serves the user. Packaging legal knowledge for the seeker's understanding.

Workflow that flows. Automation that guides rather than mystifies.

Search for meaning. The power to find intent and not just keywords.

Design that dignifies. Interfaces that treat users as capable adults.

IX. The Invitation

[039]The gate is no longer locked.

[040]To those who would keep the gate closed: history is not on your side. The abstraction of knowledge has always, eventually, served the many rather than the few.

[041]The law belongs to the people. It always has.

[042]We are here to make that belonging real.

This is the Everypact Manifesto.