Authority and action control
Who may act and within what limits; graduated autonomy; separation of duties; containment; emergency human override; restore.
In private development
Agents do the work.
Your company keeps the authority.
When an agent takes on work that matters, Lagstyr keeps the whole story in one governed case: the evidence it worked from, who allowed it, who made the call, what actually changed in your other systems, and whether it paid off—across the systems your company already uses.
The proposition
Machine-speed execution needs company-grade authority.
An agent can reason, call tools, and cross systems in seconds. But access is not authority, a trace is not a management record, and a confident output is not a decision. The more capable agents become, the more important it is that the company can answer: Who allowed this? On what evidence? What changed? What remains owed? Did it work?
Lagstyr is being designed first for owner-led and tightly managed companies moving agents beyond drafting and into work with real consequences—where founders, COOs, CFOs, and company secretaries need the complete chain, not another summary.
Where Lagstyr sits
The tools you know each hold one piece. None closes the loop.
Agent identity tells you who may call an API. Observability tells you what the model did. Approval tools tell you someone clicked yes. None of them can tell you what the company now owes, whether the change actually landed in the system that owns it, or whether it was worth doing. Lagstyr closes that loop—and leaves every specialist system's authority exactly where it belongs.
One authoritative management record
From mandate to outcome, without losing the thread.
Lagstyr sits above the specialist systems that hold company facts. It records the authority and operating context that lets people and agents perform consequential work across them.
Lab evidence · supplier renewal
From source document to governed work.
The built components are exercised end to end against our own synthetic documents. They show how facts extracted from a contract become official company records only when a person approves them—the AI can propose, but it can never quietly make its own output official.
01Ingest
Source documents enter unchanged, with their identity, classification, and origin recorded.
02Read
Embedded text and supported scanned pages are read inside the installation, and every page records how it was read.
03Extract
Renewal dates, notice periods, clauses, and obligations keep a pointer to the exact source passage.
04Govern
An accountable person accepts or rejects the exact proposal they were shown—it cannot change between review and decision.
05Operate
Accepted facts and owned obligations become visible in the contract's journey.
06Measure
The outcome view reports what is evidenced, and leaves unevidenced value visibly absent.
Built & lab-tested
Evidence to governed obligation
The full journey—reading documents, extracting terms, human approval, recording facts and obligations, and tracking the work—runs end to end on controlled test data.
Built, not yet proven live
Actions on outside systems, and outcomes
The machinery for making changes in outside systems—and confirming they really happened—is built and tested. It has not yet run against a real supplier's systems, and no real buyer has yet measured its value.
Not available
Customer proof
A real buyer's documents, a live connector round trip, and production operation are not claimed yet — they belong to the gates ahead.
Product capability in private development
Built around the company’s authority, not the agent’s convenience.
Every label below follows the same four-label vocabulary: built & lab-tested means it works end to end on controlled test data; it does not mean deployed for a customer or generally available.
Operator and approver work
Queues that route work needing judgement to the right person, and decision cards that bring the evidence, proposal, limits, and conflicts into one view.
Evidence and management record
Stable identity, governed records, page-level provenance, history, reports, obligations, and decisions—the chain behind consequential work.
Verified effects
Precisely defined actions, safe to retry, with receipts—and the owning system read back afterwards to confirm what changed.
Governed change
Registers of every AI system in use, impact assessments, exceptions, and proposal-and-approval for material changes. No agent improves itself silently.
Measurement and learning
Costs, benefits, feedback, and outcome reporting that make wider autonomy an evidence-based decision—switched on only when an installation records real evidence.
Earned autonomy
People control the moments that matter.
Oversight should be proportional to consequence. Routine work can proceed inside exact limits. Material commitments, exceptions, and changes to authority go to an accountable person with the evidence already in view.
T0 · retrieve
Read permitted sources. Search, classify, and assemble evidence without changing the source.
T1 · draft
Prepare a recommendation or complete action. A person independently decides and commits.
T2 · act with approval
Execute the exact action an authorised person approved, then verify the effect.
T3 · bounded autonomy
Narrow, reversible action inside hard limits, with continuous monitoring and automatic containment.
Agents cannot grant themselves tools, widen their limits, or rewrite the authority model. Human authority remains irreducible.
Customer-controlled architecture
The company record belongs inside the company boundary.
Lagstyr is being designed for customer-controlled deployment. Specialist systems keep their rightful authority; the runtime keeps working even if Lagstyr’s vendor is unreachable; model providers are selected and governed by the organisation.
Your cloud
Bring-your-own-cloud direction
The intended distribution model (BYOC) places the whole deployment—core service, workers, database, and storage—inside infrastructure the customer controls, not ours.
Operating profiles
Partner-operated or customer-operated
Control Starter and Authority Platform define two intended ways to run the same kernel. They are profiles, not separately available products.
No runtime reach-in
Authority does not phone home
No vendor licence check, heartbeat, telemetry requirement, or management endpoint is intended to sit on the company’s runtime path.
Where things stand
Evidence before claims.
Most of Lagstyr is built. We test it end to end in isolated environments with realistic synthetic data—no customer data, and no customer installation yet.
There is no production installation, public trial, or independent security certification, and we are not accepting installations yet. The next gate is a small design-partner pilot; the development page records the whole path.
Read the path to availabilityFree resource · no email required
The AI-native operating guide.
A vendor-neutral, five-document reference model for leaders building an AI-native organisation: architecture, phased delivery, proportionate governance, and honest benefits discipline.
Read and download the guide
Free · ungated · Markdown
Register interest
Follow the evidence. Be there when the gate opens.
Registering brings you occasional development and availability updates—and when the design-partner pilot opens, registered interest is where we will look first.
We store your email address only to send Lagstyr development and availability updates. No tracking, no list sharing, unsubscribe in one click. See our Privacy Policy.
Questions now? Write to the team.
