‹ Back to site
1 / 23
The Knowledge Layer for IT Development · Witold Reichhart · governedintelligence.eu · © 2026
A governed basis for delegated AI in software delivery

The Knowledge Layer
for IT Development

AI coding agents are about to act - they will open pull requests, bump dependencies, and ship deploys. They need a layer beneath them that governs what they act on, decides what they are allowed to do, and records why they did it.

Witold Reichhart · governedintelligence.eu · first published May 2026

Arrow keys to navigate · M for menu · T for theme · N for notes · scroll for depth

The shift

Capability is solved. Admissibility is not. Your coding agents are about to act - and the model that makes them capable cannot make them admissible.

Capability is whether the agent can produce the change. Admissibility is whether the basis it stood on - the patterns, the contracts, the assumptions it relied on - can be defended and relied upon. The first is the model's job, and it is essentially done. The second is not the model's job at all.

What the model gives you

Capability

AsksCan the agent produce the change?
Delivered byThe model - and the models are good enough to act.
Gets youA change, written.
Stops atA draft a human still has to vet. The model has no notion of whether what it relied on was current, authoritative, or in scope.
What the model cannot give you

Admissibility

AsksCan the basis the change stood on be defended?
Delivered byA layer beneath the agent that governs the basis. Not the model - it is probabilistic, it cannot compute standing.
Gets youA change that can ship, and be defended after the fact.
The unsolved halfThis is what the rest of the deck builds.

The agents are about to act - not already acting - and that is the opening. Stand the admissibility layer up before they scale, and every consequential change is admissible from the first commit. Add it afterwards, and you inherit a backlog of changes nobody can defend.

Capability gets the change written. Admissibility gets it shipped. The full benefit of delegated coding - agents that operate, not merely assist - sits on the admissibility side of that line.
The case for the layer

Why the layer has to exist. When AI moves from tool to operator, the job of carrying the basis for a consequential act loses its owner - and no layer the institution already runs can take it in.

Two halves to the case. The architectural half is a structural claim: one specific function goes un-owned the moment the human leaves the act, and each existing layer fails to adopt it for a reason that does not go away. The strategic half is the cost of leaving it un-owned - an AI portfolio that stalls at deployment, and a regulatory exposure that is turning from good practice into obligation.

The architectural argument - one function, no owner.

Every consequential act rests on a basis - the facts, rules, interpretations, and authorities the actor relied on. In the tool world a person at the point of action held that basis: the senior engineer knew the auth pattern had changed, and which library version still carried the CVE. Delegated AI removes the person. The basis still has to be held. Walk the five layers the institution already runs and ask which one holds it.

Application & process layer
Governs  the flow of work, screen by screen.
When  build time.
Can't hold the basis - the wrappers that carried process logic are dissolving into agents. This layer is shrinking, not taking on work.
Data governance
Governs  records - what is stored, and who may see it.
When  at rest.
Can't hold the basis - it governs the dataset, not the claim. No authority chain, no standing, no test of fitness at the moment of action.
Spec-driven engineering
Governs  what the system shall do.
When  build and release time.
Can't hold the basis - a spec is deontic and ships at release; the basis is epistemic and moves between releases.
Model risk management
Governs  the reasoner - the model itself.
When  validation and monitoring.
Can't hold the basis - MRM governs the model, not the claims the model reasons from.
Identity & policy runtime
Governs  the actor's permission to act.
When  action time.
Can't hold the basis - it tests whether the actor may act, not whether the basis is fit to be acted on.
none of the five holds it
The function with no owner
The basis - at the moment of action
Held by none of the five layers above. This is the Knowledge Layer - the one layer that governs the basis itself, at the instant of the act.

Each layer governs a different object, at a different moment. Stack all five and the basis at the moment of action is still held by no one. A function with material consequence and no owner is not a process gap to close with effort - it is a missing architectural layer.

Outcome A · leave the function homeless
The basis stays in a human head. A person sits in every consequential loop. Delegation is cancelled in all but name - the AI assists, it does not operate. The investment case for autonomous action never closes.
Outcome B · build the layer
The basis moves into a governed institutional layer that holds it at machine speed. Delegation holds. Every consequential act carries a basis the institution can defend after the fact.

There is no third outcome in which delegated AI runs consequential actions and the basis has no home. The choice is the layer, or a human in every loop.

The strategic argument - the cost of leaving it homeless.

Three reasons the layer is worth building deliberately, and now.

Stake 1

Where the AI money is stuck

Models perform. Pilots pass evaluation. Projects stall - and they stall at the deployment gate, not the build gate. The deployment gate is the admissibility gate. Without the layer, the portfolio is a row of pilots that never convert into operating systems.

Stake 2

What compounds, what does not

Every competitor buys the same models - capability is a rented advantage that resets each release. A governed claim substrate is built from the institution's own knowledge and earns a longer track record with every action. It compounds, and it does not transfer. Durable advantage accrues in the layer, not the model.

Stake 3

The clock is regulatory

Action-time evidence is moving from good practice to obligation - EU AI Act Article 12, DORA Articles 11-12, BCBS 239, SR 26-2, PRA SS1/23, in force or in transposition now. The institution will build this capability. The open question is whether it builds one architecture on purpose or reconstructs it in fragments under audit pressure.

Capability is the cost of entry - everyone pays it. The governed layer is the asset, and an asset is built once, on purpose - not reassembled in fragments while a supervisor waits.
The requirements

What an admissible change must carry. Eight requirements the problem itself sets - and the test the next slide applies to every tool you run.

These are not a vendor's checklist or this deck's invention. Write down what it takes to defend a delegated change after the fact, and these eight fall out. A competent reviewer, a security lead, or an architect names each one on their own. The three marked action-time are the requirements that decide admissibility at the instant the change is made.

R1

Provenance

Every fact the change relied on traces to a source. Which decision said this is the canonical pattern, authored by whom, and when.

R2

Confidence and standing

The basis carries a computed confidence, not a binary "it exists." How much weight a claim bears is derived from evidence and track record.

R3 · action-time

Action-time admissibility

The basis is checked at the moment of the change - is this claim fit to rely on now, for a change of this consequence. Not at build time; at the act.

R4

Named authority

Each claim has a named owner with the standing to author it. The canonical pattern is owned by the architecture board, not by whoever last edited the wiki.

R5 · action-time

Signed manifest

The change carries a tamper-evident record of the basis it relied on - the claims, their standing, their authority - bound to the commit.

R6 · action-time

Replay

The basis as it stood at the moment of the change can be reconstructed afterward, when an incident surfaces weeks later.

R7

Currency and decay

Claims expire. A pattern is superseded, a library deprecated, an interface sunset - the basis must know a claim has aged out, not treat last quarter's truth as current.

R8

Tiered by blast radius

The bar scales with consequence. A production deploy to a critical service is held to higher standing than a feature-branch merge. The gate is proportional.

The room cannot dismiss this list without dismissing one of its own seats. Each requirement is something a competent risk, security, or architecture function already asks of a consequential change. The next slide tests what you run against the eight; the slides after specify what clears all of them at once.

Hold the eight. The next slide takes every approach a software organisation already runs and tests it against them - and the gap is what is left.
The gap

Nothing you run today closes this. Thirteen approaches a software organisation already operates - the ways it governs knowledge, and the ways it keeps change safe - tested against what an admissible change requires.

Each column is something an organisation already runs - on the left, the things people count on to govern knowledge; on the right, the tooling that keeps change safe. Each row is one of the eight requirements just established. Read down a column or across a row - the conclusion holds either way. The three shaded rows are the action-time requirements - admissibility, the signed manifest, and replay - and they are where the gap is.

A1Data governanceA2Knowledge baseA3ADRsA4Service catalogA5AI governanceA6Code reviewA7CI/CDA8Static analysisA9Dependency scanA10Agent frameworkA11Policy-as-codeA12Build provenanceA13Change advisory
R1 · Provenance
R2 · Confidence / standing
R3 · Action-time admissibility
R4 · Named authority
R5 · Signed manifest
R6 · Replay
R7 · Currency / decay
R8 · Tiered by blast radius
Full coverage Partial coverage Does not cover

One full mark across the whole table. Dependency scanning tracking CVEs and outdated versions - the one thing it is built to do. A category-defining strength of a single column, and not coverage of the gap.

The action-time trio has no full coverage anywhere. The tools that gate at all - CI, policy-as-code, change advisory - gate the test result, the declared rule, or the change ticket. The knowledge stores - data governance, the wiki, the catalogs - hold content but gate nothing. None reaches the basis the change reasoned from.

The gap is not a weak spot in one tool. It is a missing layer - the one that holds the basis, computes its standing, and binds it to the change. The slides that follow specify it.
What the layer is

What the Knowledge Layer is. Raw institutional knowledge goes in. A signed, defensible basis comes out at the moment of action. The machine in between is the layer.

Read it left to right. The strip below fixes where the layer sits - between the players who act and the systems that record. The machine under it is what the layer does: ungoverned knowledge as feedstock, four jobs that turn it into governed claims, and an action-time gate that hands the actor a basis it can stand behind. Every outcome writes back, so the layer compounds the more it runs.

Where it lies
Above
Players
Humans and AI agents - whoever does the consequential work. They act on what the layer delivers.
The layer
The Knowledge Layer
The governed substrate that delivers the basis. The action-time gate is its upper boundary.
Below
Systems of Record
Payments, contracts, filings. The layer stands on them and reads from them; it does not replace them.
What it does
Goes in
Ungoverned knowledge
Regulations & rulings
Policy & interpretations
Records & data
SME & tacit knowledge
Model outputs
Mixed authority. No provenance discipline, no standing, no decay.
The Knowledge Layer four jobs, three primitives
01 Capture
Stamp provenance at intake - source, authority, method, time.
02 Structure
Type each claim into the claim graph - ten claim types.
03 Govern
Score, compute standing, set decay, resolve contradiction.
04 Serve
Deliver the right claims to the act, with the gate and the manifest.
What flows through - the claim
typed provenance-stamped confidence-scored standing-computed bitemporally tracked
The claim - not the document or the row - is the unit on which admissibility is decided.
The standing function
Confidence is computed, not assigned, and recomputed continuously as evidence and authority move. The standing function is itself a model under MRM - 1LoD owns it, 2LoD challenges it, 3LoD audits it.
confidence standing
0 50 100
Knowledge
The governed claim graph - built by Capture, Structure, and Govern.
Context
The signed bundle delivered at the act - built by Serve.
Memory
The encode-back loop - recomputes standing from outcomes.
Comes out, at the act
The action-time gate
Confidence standing × consequence tier, run through the circuit breaker → admit · block · escalate.
Tier 1 · ≥85consequential
Tier 2 · ≥70material
Tier 3 · ≥50routine
Tier 4 · anyinformational
The signed manifest
The bounded, inspectable bundle of relied-upon claims, bound to the action and replayable from the event log. The basis the actor stands behind.
No signed manifest, no action
Encode-back loop The committed outcome writes back as an operational-validation event. Standing recomputes, the pattern library compounds, and contamination control keeps the agent's own traces out of corroboration. The layer gets better the more it runs.

For the architect. In standard policy-architecture terms the action gateway is the PEP, the layer's manifest service is the PDP, the claim graph is the PIP, and the standing-function-version registry is the PAP. The layer is a control plane that fits the pattern - not a paradigm to adopt wholesale.

The layer cannot lift its own gate. No retrieval routes around it. Evidence is produced at the act, not reconstructed after it. That is the whole of what makes a delegated action defensible - and the reason the gate has to be a layer, not a feature.

IT development · the worked example

The same architecture, in the place it pays back fastest. Point the Knowledge Layer at AI-assisted software delivery and the loop hardest to prove elsewhere - knowledge that earns its standing through outcomes - closes in days, not quarters.

Agentic coding is moving IT from AI-as-tool to AI-as-operator. As agents take on pull requests, dependency bumps, and deploys with no human at the act, the basis they rely on - what is canonical, what is deprecated, what is fragile - is the same un-owned function the case for the layer describes. IT is the strongest first instance: the feedstock is already machine-readable, the outcome signal is fast and objective, and the estate is small enough to start without touching regulation or company-wide knowledge.

The reframe - the repository is a System of Record, not the layer.

Repositories, CI logs, dependency manifests, and the service catalog sit below the layer. Aim the substrate at what the code says and it is retrieval over a repo - which the organisation already has. The layer governs the interpreted layer above the repo: what the institution is currently prepared to let an agent rely on.

AGT
AI coding agent - the operator
Opens pull requests, bumps dependencies, modifies infrastructure, ships deploys. No human at the act.
KL
The Knowledge Layer
Governs the interpreted layer - what is canonical, current, owned, and fragile. The basis the agent acts on, delivered with standing at the moment of the change.
SoR
Systems of Record
The repository, CI logs, dependency manifests, the service catalog. Machine-readable already; the layer reads from them and does not replace them.

The repository shows the codebase contains the old auth pattern and the new one. Which one is canonical is a claim - authored, dated, and decaying - and it is not in the code.

What the claim is, in IT - five types.

Only the second type is cheaply re-derivable from source. The other four decay, contradict each other, and carry authority - the test for whether the substrate earns its place. Architectural canon is the sharpest: the IT equivalent of a policy interpretation.

Type 01 · the sharpest
Architectural canon
The canonical pattern for a cross-cutting concern - auth, data access, error handling, deployment. The code holds every pattern ever shipped; which one is current is a judgment, not a fact.
Authority  architecture board
Decays on  migration & platform events
Type 02
Lifecycle & deprecation state
An API, library version, endpoint, or service is current, deprecated, or sunset. Factual but fast-moving; the feedstock - deprecation notices, semver, CVE feeds - is already structured.
Authority  owning team
Decays on  release cadence; CVEs intra-day
Type 03
Ownership & accountability
Who owns a service, who may approve a change to it, who is accountable when it breaks.
Authority  the organisation
Decays on  reorganisation
Type 04
Operational reality
The gap between documented and actual behaviour - the real connection-pool ceiling, the undocumented coupling, the config with a large blast radius. Tacit; lives in postmortems and on-call heads.
Authority  on-call / incident review
Decays on  infrastructure change
Type 05
Risk annotation
This module is fragile; this area has caused three incidents this year; changes here need extra scrutiny. Institutional memory, made explicit.
Authority  incident review
Decays on  refactor; sustained clean record

The action, the gate, the tiers.

The consequential delegated action is an agent proposing a change. The gate fires at pull-request open and at deploy. The signed manifest records the claims the change relied on, at what standing, under whose authority. The consequence tier is blast radius.

01 ProposeAgent proposes a change - a PR, a dependency bump, an infra change, a deploy - with a declared change class.
02 ResolveThe layer resolves the claims that govern this change - canon, lifecycle, ownership, risk.
03 ScoreThe standing function computes admissibility per claim against current evidence and authority.
04 BuildThe Context manifest is assembled - the relied-upon claims, their standing, the authority chain.
05 GateManifest against blast-radius tier. Admit, route to human review, or block.
06 ShipOn admit the change merges or deploys. The signed manifest is bound to the commit.
07 EncodeThe build, incident, and rollback outcome writes back to Memory. Standing recomputes.
Tier 1 · admits ≥85
Production deploy to a critical service, schema migration, security or IAM config change, any change during a freeze.
Tier 2 · admits ≥70
Merge to main of a non-trivial change, a major-version dependency bump, a change to a module flagged fragile.
Tier 3 · admits ≥50
Feature-branch merge, test addition, documentation change.
Tier 4 · any standing
Code search, draft suggestions, analysis not yet submitted.

Why IT makes the case - the encode-back loop actually closes.

This is the reason to start here rather than anywhere else.

Every change has a fast, objective outcome - a green build, an incident, a rollback, an error-rate spike. That outcome writes back as an operational-validation event. Claims relied on by changes that shipped clean accrue standing on the operational-validation dimension; claims relied on by changes that caused incidents lose currency. Within weeks the standing scores are a measured track record of which institutional knowledge is trustworthy.

Change ships
The merge or deploy commits, its signed manifest bound to it.
Outcome observed
Green build, incident, rollback, or error-rate spike - within hours.
Standing recomputes
Claims behind clean changes gain standing; claims behind incidents lose currency.

The compliance domain waits for a supervisory examination to learn whether a claim was sound. IT learns from the next deploy. The slowest part of the architecture to prove anywhere else is the fastest part to prove here.

Start small - shrink the scope, not the architecture.

The minimal first instance
One platform. Two claim types - architectural canon, plus lifecycle and deprecation state - roughly 100 to 200 claims. Gate agentic-coding pull requests and deploys within that platform. The authority body is the architecture board: real, named, and accountable.

Shrink the claim count, the domain, and the action scope as far as is useful. Keep the gate and the encode-back loop. Drop either and it stops being the Knowledge Layer and becomes retrieval with extra steps.

Anticipated objection - "this is a better service catalog, or RAG, or policy-as-code." Catalogs and RAG return what exists, not what is admissible to rely on now. Policy-as-code gates the artifact, not the basis the agent reasoned from. The substrate's distinct contribution is standing - computed, decaying, outcome-validated - and the action-time manifest.

IT is the proving ground; the regulated estate is the destination - the same architecture runs both. And the agentic engineering platform the organisation is already standing up is missing exactly this: a governed context, and a governed memory.
IT development · what you get

What you get. Six things change once delegated coding runs on a governed basis instead of an assumed one.

Capability is settled - the agents write code. These are the changes that show up when the basis they act on carries standing: in how much output you trust, in whether architecture decisions hold, in how institutional knowledge survives, and in what the organisation owns at the end.

Benefit 01

Trust becomes selective

Today the choice is binary - trust all agent output, or review all of it. Blanket trust ships confident garbage; blanket review reinstates the bottleneck delegation was meant to remove. Standing is the third path: a high-standing basis flows, a weak or decayed one routes to a human.

You review what the layer flags - not everything.
Benefit 02

The deployment gate stops being a person

The senior reviewer is today's admissibility gate - the one who knows the threshold moved or the module is fragile. That person does not scale to the volume agents produce. The gate becomes machine-speed and consistent.

Human judgement is spent only where it is genuinely needed.
Benefit 03 · for the architect

Architecture decisions become enforced

A board decision today lands in an ADR and a wiki page, then decays quietly while agents and engineers keep shipping the old pattern - it is still in the codebase, still in the training data. As a governed claim, that decision becomes a runtime fact: the gate refuses changes that relied on superseded canon.

Your authority stops being advisory.
Benefit 04

Institutional memory stops walking out

Fragility, real blast radius, undocumented coupling - today this lives in a few senior heads and in postmortems nobody re-reads. As risk and operational-reality claims, it becomes part of the basis every agent and every new engineer acts on.

Operational knowledge compounds instead of leaving with people.
Benefit 05

Incidents become explainable

When a change causes an incident weeks later, you replay its manifest - which claim, at what standing, under what authority, on what date. You see at once whether the agent acted on a bad basis or a sound one overtaken by events.

Post-incident review stops being archaeology across logs, tickets, and chat.
Benefit 06

The capability is yours, and it compounds

Every competitor buys the same coding models - capability is rented, and it resets each release. A governed claim graph is built from your codebase, your decisions, and your incident history, and it earns a track record with every deploy.

The half of the agentic-engineering investment nobody else is building.

The models give the organisation capability. The layer gives the agents a governed context and a governed memory - the institutional grounding that decides whether delegated coding becomes an asset or an exposure.

IT development · how you stand it up

How you stand it up. One platform, two claim types, a gated pilot in six to eight weeks - the architecture whole, the scope small.

The way to get this wrong is to govern everything at once. The way to get it right is to shrink the scope hard and keep the architecture intact. Nothing in the first instance touches regulation, company-wide knowledge, or a new governance function.

The minimal first instance
Scope
One platform - the internal service platform, or one bounded estate.
Claim types
Two - architectural canon, plus lifecycle and deprecation state.
Claim count
Roughly 100 to 200 governed claims.
Authority
The architecture board - real, named, already deciding these.
The build - six to eight weeks to a gated pilot
Phase 1 · weeks 1-3
Capture & Structure
Ingest from ADRs, CODEOWNERS, dependency manifests, deprecation notices, and the board's decision log. Type the first ~150 claims and build the dependency edges, so a moved claim re-drives its dependents.
Gate - not yet wired
Phase 2 · weeks 3-5
Govern & gate in shadow
Score confidence, set decay, compute standing. The gate observes - it writes the manifest and computes a verdict, but does not block. Measure false-accept and false-reject before anything is enforced.
Gate - shadow mode, observes
Phase 3 · weeks 5-8
Serve & gate live
Turn the gate live for the high-consequence tiers first - production deploys, security changes, changes during a freeze. Widen tier coverage as the false-accept and false-reject rates hold.
Gate - live, top tiers first
Encode-back loop runs from the first shadow-mode week. By go-live the standing scores already carry a track record from real build, incident, and rollback outcomes - the gate opens on evidence, not on assertion.
What it sits on - already in your estate

The layer adds the claim model, the standing function, and the gate. It does not rebuild the plumbing - the prerequisites already run in your estate.

Source control Service catalog CI/CD pipeline - the surface the gate sits in Identity Observability & incident tooling - the outcome signal

Shrink the claim count, the domain, and the action scope as far as is useful - but keep the gate and the encode-back loop. Drop either and it stops being the Knowledge Layer and becomes retrieval with extra steps.

Anticipated objection - "this is a better service catalog, or RAG, or policy-as-code." Catalogs and RAG return what exists, not what is admissible to rely on now. Policy-as-code gates the artifact, not the basis the agent reasoned from. The layer's distinct contribution is standing - computed, decaying, outcome-validated - and the signed manifest.

Small, measured, and reversible at every step - and at the end of it, a deployment gate that is an architecture you designed, not an accident you reconstruct after an incident.
IT development · implementation in practice

Implementation in practice. What an engineering team ingests, builds, and runs - in order, and on the stack it already has.

The previous slide is the shape of the pilot: one platform, six to eight weeks, the gate in shadow mode before it blocks. This slide is what the team does inside that shape - the order knowledge is ingested, the components built, the technology they map onto, and the handful of people who run it.

What to ingest, and in what order.

Order by authority and tractability. The board's decisions are the canon and are already written. Dependency feeds are structured and automatable. Tacit knowledge in runbooks and postmortems is the hardest to type, so it comes last - the first instance is steps one and two only.

Step 1 · first instance
Architecture decisions and the board's decision log.
→ architectural canon claims
Step 2 · first instance
Dependency manifests, CVE and deprecation feeds.
→ lifecycle claims
Step 3 · expansion
CODEOWNERS and the service catalog.
→ ownership claims
Step 4 · expansion
Runbooks and incident postmortems.
→ risk and operational-reality claims

What to build, and on what.

Seven components. The architecture is fixed; the products are not - the Microsoft and AWS columns are examples of how each component lands on a stack a team already runs.

ComponentWhat it doesMicrosoft stackAWS stack
Capture connectorsRead each source and stamp provenance at intake.Functions or Logic Apps, GitHub APILambda, EventBridge, GitHub API
Claim storeThe typed, bitemporal graph of claims.Azure SQL temporal tables, or Cosmos DBAurora PostgreSQL bitemporal, or Neptune
Event logAppend-only and ordered - the spine that makes replay possible.Azure Event HubsAmazon Kinesis or EventBridge
Standing functionScores each claim and recomputes standing as evidence moves.Azure Functions or Container AppsAWS Lambda or ECS
Manifest serviceAssembles and cryptographically signs the Context manifest.Functions with Key VaultLambda with KMS
The gateA required check in the pull-request pipeline, an admission step at deploy.GitHub Actions check, Azure PipelinesGitHub Actions check, CodePipeline
Encode-back pipelineConsumes build, incident, and rollback signals; recomputes standing.Azure Monitor with FunctionsCloudWatch with Lambda

Who runs it.

A pilot team is three to five people, plus the architecture board's time. A pilot, not a programme.

1 person

Pilot lead

Owns the first-instance charter, the architecture-board liaison, and the go-live decision.

1-2 people

Layer engineers

Build and run the claim store, the standing function, the gate, and the capture connectors.

1-2, part-time

Claim curators

Author and maintain the claims with the board. Usually senior engineers, not a full-time hire.

existing body

Architecture board

The authority that approves architectural canon. Already exists - no new headcount.

Across the three phases of the previous slide: the curators and the board lead the opening weeks - ingest and author the first claims; the layer engineers carry the middle - the store, the standing function, and the gate in shadow mode; the lead runs go-live and the tier-by-tier widening.

The stack columns are examples, not a mandate. The architecture needs a bitemporal claim store, an append-only event log, a signing service, and a gate in the pipeline - build each on the products your platform already runs.

The decision

Your agents are about to act. What you settle now is whether the basis they act on is an architecture you designed - or an accident you reconstruct after an incident.

Start in IT. The feedstock is machine-readable, the outcomes are fast, the scope is yours to bound - one platform, two claim types, a gated pilot in six to eight weeks. The same architecture extends to the rest of the institution once it has earned the reference here.

The Knowledge Layer · IT projection · May 2026