# AuthLN - full site content > AuthLN proves authentication and authorization: who was allowed in, what was allowed to > act and on whose authority, on a record no one can rewrite. This file is the readable > content of every substantive page, generated at build time from the site itself. The > short index is at https://authln.com/llms.txt ## Read this first - Factor Assurance is shipped and in production. The Agent Access Review is available now. Agent Authority is coming soon and is not available yet. - Factor Assurance outcomes are Authorized, Denied and Walked away. - Two issued US patents: 11,956,366 and 12,118,550. Pending or provisional applications are not counted in that total. - SOC 2 Type II certified. - Figures labeled modeled or illustrative are not measured customer results. --- # AuthLN - AI is how you grow. AuthLN is how you trust it. URL: https://authln.com/ Frontier security platform ## AI is how you grow. AuthLN is how you trust it. AuthLN gives a company the trust to put AI to work, and protects it from AI aimed at it. What we do ### See it. Prove it. Govern it. We don't ask you to trust us. We give you the tools to prove, for yourself, who or what is acting in your company's name. - See it 01 Agent Access Review Available now Discover every AI agent running in your company and whose authority each one uses, so leadership can see what is operating in its name. Agent Access Review - Prove it 02 Factor Assurance Shipped A gate in front of the sign-in you already run. Every attempt has to come through a check on a phone a real person holds. Factor Assurance - Govern it 03 Agent Authority Coming soon Every AI action traces back to the person who allowed it, with authority that only narrows from one agent to the next, on a record no agent can change. For design partners. Agent Authority Why it matters ### AI multiplies whoever uses it. It works both ways. For you 2.8x AI leaders are increasing the decisions their AI makes on its own at 2.8 times the rate of their peers. Source: PwC 2026 AI Performance Study Against you 1 in 4 malicious breaches is now AI-enabled, up 56% in a year. Source: IBM Cost of a Data Breach Report 2026 At scale One attacker's AI agents reached 395 organizations, 11 of them in 26 seconds. Source: GreyNoise, September 2026 ; Blackpoint Cyber Why it matters Trust decides which side of the multiplier you are on. How it works The platform ### One platform proves who comes in, and is designed to govern what acts inside. - A person signing in AI attempts stop at the door - Factor Assurance the door: who is allowed in Quantum-safe Shipped - Your organization Agent Access Review Available now Agents acting for one person - Sub-agents started along the way - Agent Authority the gateway, in every path: what may act, and on whose authority Coming soon - Apps Data APIs Two parts. The door and the gateway each work on their own. Start with the door, or with an Agent Access Review. What you get ### Four returns on trust. When a company can trust its AI, it can move at its own pace. This is what that's worth. **Move faster** Approve AI projects sooner, because you can see what is running and who stands behind it. AI leaders are increasing the decisions their AI makes on its own at 2.8 times the rate of their peers. PwC 2026 AI Performance Study **Get time back** Your team stops chasing false alarms and works on what matters. About 7 in 10 of the authentication alerts analysts work today would stop reaching them. Modeled See the model and its assumptions **Protect revenue** Reduce the risk of a breach, and of one AI mistake that compounds before anyone sees it. The average breach now costs $4.99M, a record. IBM Cost of a Data Breach Report 2026 **Show oversight** A clear answer and a record when regulators, auditors or insurers ask how AI is controlled. Every agent and the person whose authority it uses, in one report. Agent Access Review Start here All four start today, with the Agent Access Review and Factor Assurance . The record Proof of both ### Every attempt, grant, action and refusal, on a record no one can rewrite. A sign-in attempt leaves a record whether it was authorized, denied or walked away. Agent Authority is designed so that every grant, action and refusal down a chain of agents lands the same way. The record is written once and held where the thing being recorded cannot edit it. Transparency is the product, not a feature: you check the record instead of taking anyone's word. How the record works The system that acted Apps, agents and your identity provider act here. None of them can edit what lands on the right. The record · write-once Illustrative entries - 8f2c62 Sign-in Authorized outcome, correlation ID, timing - 8f2c55 Sign-in Walked away the missing device signal is the finding - g-0417 Grant authority, from a named person what it covers, and who it was handed to Coming soon - a-0418 Action sub-agent 4: update ticket within the grant, on the record Coming soon ### Security and trust Patents Two issued U.S. patents: US 11,956,366 and US 12,118,550. SOC 2 SOC 2 Type II certified. Trust Center Policies, controls and documents for your security review. trust.authln.com Available now ### How many agents are running in your environment today, and who owns them? Agents are being turned on across the tools you already use. They act with people's access, and most teams cannot say what each one can reach or how to stop it. From read-only exports and with nothing installed, an Agent Access Review answers it. And the questions that follow - 01 How many changes can one agent, and the sub-agents it starts, make in our systems? - 02 When we revoke a person’s access, do the agents acting for them lose it at the same moment? - 03 If an agent is compromised, what did it do, and under whose authority? - 04 When did an agent first reach sensitive data, and whose grant let it? --- # Factor Assurance - Authority to Enter | AuthLN URL: https://authln.com/factor-assurance Factor Assurance Authority to enter ## The decision layer in front of your identity provider that proves the attempt. Factor Assurance decides who is allowed in. Every sign-in attempt resolves to Authorized, Denied or Walked away, and arrives in your SOC as one event with an outcome and an ID. It stands in front of Okta, Microsoft Entra ID, Google Workspace and JumpCloud, which stay exactly as they are. Authorized no friction, on to your identity provider Denied still verified, still denied Walked away saw the gate, left. Still recorded. Shipped Factor Assurance is shipped and in production. Why you need it ### AI makes the old ways in cheap to exploit. Every way in, through one door. A real person proves it is them, on a phone they hold, in front of the sign-in you already run. Automated, non-human attempts stop at the door, before they become alert noise for your SOC. - 1 in 4 malicious breaches is now AI-enabled, up 56% in a year. IBM Cost of a Data Breach Report 2026 - 54% clicked fully AI-written spear phishing, the same rate as emails written by human experts. Heiding, Schneier et al., 2024 - 99%+ of unauthorized access attempts are blocked by multifactor sign-in. Attackers go for the ways around it. Microsoft Digital Defense Report 2025 The cost. $4.99M for the average breach, a record. About $6M when it is AI-enabled. IBM Cost of a Data Breach Report 2026 What you get All three ### Detection expires. The gate you impose does not. Most systems record a success or a failure. Factor Assurance records every attempt, including the walk-away: the attempt that saw the gate and left. **The gate** Every attempt must satisfy one gate. Your people clear it in one scan. It can be your passkey and your MFA, and the decision layer above both. **Signal** The passkey lives on the person's phone, bound to the device. Every event arrives resolved, with an outcome you can act on, not a failed login to investigate. **Decision record** Instant detection, before the attempt reaches your identity provider. Every attempt resolves to Authorized, Denied or Walked away, on a durable, correlated record that is auditable months later. Who this is for ### What the CISO can finally answer. For the person accountable for board reporting and defensible evidence. - How many people tried to get in last quarter? Every attempt is recorded, including the ones that never crossed a threshold and never became an alert. - How hard was it for them to try? Every attempt has to satisfy the gate and settles to a record, so the deterrent is measured, not asserted. - Which identities are worth most to an attacker? Walk-away behavior shows where they press on and where they give up, per user and per workload. Who this is for ### What changes in the analyst queue. For the analyst who wants alerts worth opening, time to close, and evidence in the ticket. The queue today - Failed authentication - Authorized - Failed authentication The failures carry no outcome, so none of them can be ranked. The queue with AuthLN - Denied 8f2c41 - Walked away 8f2c55 - Denied 8f2c58 - Walked away 8f2c5e - Authorized 8f2c62 - Walked away 8f2c71 Each one carries an outcome and an ID that survives. Structured for your SIEM. Illustrative IDs **What an analyst stops doing** Your analysts spend fifteen minutes working a failed-authentication alert to find out nothing happened. Every attempt through the gate arrives with the outcome already attached, so those fifteen minutes go back to them. - Users - Analyst-hours back per year - Capacity - Value per year - 500 - 648 - 0.4 FTE - ~$42,000 - 2,500 - 3,240 - 1.9 FTE - ~$211,000 - 10,000 - 12,960 - 7.6 FTE - ~$842,000 - 50,000 - 64,800 - 38.1 FTE - ~$4.2M Modeled, not measured. 0.6 authentication-noise alerts per user per month, 80% of them volume-driven, 90% deterred, 15 minutes each at $65 fully loaded. On that model, about 7 in 10 of the authentication alerts analysts work today would stop reaching them. A pilot replaces these assumptions with your own numbers. Two paths diverge How it works ### A passkey answers whether the key fit. We answer what happened. - Question - FIDO2 passkey alone - Passkey + Factor Assurance - Question it answers - Is this the right key? - Was this the right decision, and what happened when the answer was no? - Outcomes captured - Success or failure - Authorized, Denied, Walked away. The third state is the one most systems never capture. - Abandoned attempt - Invisible - Recorded. A walk-away from an untrusted origin is itself a signal. - Location check - Single device, no cross-check - Phone GPS against browser geo-IP, with an impossible-travel gate. - What survives after - A log line - A durable, correlated decision record, auditable months later. Why companies choose AuthLN ### Keep your identity provider. Close the doors it has to leave open. Established providers serve huge installed bases, so older ways in stay open to keep older systems working. We built AuthLN for AI, with no past to protect. - Topic - Established identity providers - AuthLN - Older ways in - SMS, push and password reset stay open so older systems keep working. - Every way in can run through one check, set up the way we recommend. - Passkeys - Researchers have bypassed them by forcing a weaker fallback, or by replaying a sign-in. Proofpoint, 2025 ; SpecterOps, Black Hat USA 2026 - Set up as we recommend, there is no weaker method to fall back to. - Getting started - An upgrade or a migration. - Sits in front of the sign-in you already run, with nothing replaced. - Built for - A world where attacks took skilled people and time. - AI-driven attacks, from the start. For you. Keep what you already run. Add what the established providers can't. Where it sits Okta example ### In front of your identity provider. Nothing gets rerouted. The sign-in path ≈ 1 second for the decision layer, in front of the session - 01 User requests the app Browser or app, managed or not. - 02 Okta evaluates policy Your tenant, exactly as it works today. - 03 AuthLN Factor Assurance The attempt is challenged, and the challenge has two answers. Paired the pairing holds, on to Okta Not paired no pairing to present, denied - 04 Okta issues the session For authorized attempts. Nothing is rerouted. Two ways to deploy: layered with Okta Verify, or scoped to privileged and high-risk access only. The same gate stands in front of Microsoft Entra ID, Google Workspace and JumpCloud. What your SOC sees One event per attempt, joinable to your identity provider's system log. outcome · correlation ID · timing IP, ASN, geo · VPN and Tor flags In your tenant, not ours Enforcement policy. Group scoping. We hold no directory and no lifecycle. Your directory, your policy and your lifecycle are unchanged. The credential Physical and digital ### An AI never leaves the digital. Authority lives in both. Passkeys and tokens can be generated, copied and replayed. Policy and logic can be read, modeled and reasoned around. Bind authority to a physical anchor, and the attempt has to come from a body, not a model. **Device** A key that cannot be exported or lent. **Proof** A proof that cannot be rented, rotated or forged. **Presence** A device in the place, at the moment. Where it applies ### Wherever something authenticates. The credential **Device-bound passkey** The private key is generated inside the hardware security perimeter and cannot be exported, behind a class-3 biometric. Secure Enclave on iOS, StrongBox on Android. **Proximity-based authentication** Device GPS is evaluated at the moment of the attempt and cross-checked against network geolocation. **Quantum-safe passkey** ML-DSA-65-ES256 hybrid. Post-quantum signatures standardised in NIST FIPS 204, with COSE algorithm identifiers registered in RFC 9964. Where it runs **Workforce sign-in** Everyday logins in front of your IdP. Every attempt lands in the console with its outcome and its ID. **Service and machine accounts** Gates non-human authentication so automated abuse is deterred. Token, OAuth and agent coverage is roadmap. **High-value transactions** A proof-backed step-up behind the firewall, on the actions that matter, at front-door difficulty for an attacker. **Supply chain accountability** Custody is signed at each handoff and re-verified at the next checkpoint. Each scan carries a proof. The same gate, the same proof, wherever the attempt happens. The record ### Every attempt leaves a record. A walk-away still leaves a record. The missing device signal is the finding: the shape of the record tells you it was not them. How the record works Every attempt - Outcome + correlation ID - Dwell + 4-stage timing - IP, ASN and geo-city - VPN, Tor and proxy flags when the network exposes it When they engage - Device GPS + accuracy - Origin trust class when it resolves What it asks of you ### No rip and replace. The gate sits in front, the IdP stays put. **No keys to hold** The proof is produced on their device, never on yours. **No new credentials** Your passkeys and your IdP stay exactly as they are. **No friction for your people** Authorized attempts clear in one scan. The pilot 30 days ### One group. Thirty days. Then look at the record. - One application, one user group. - A named owner on your side. - An hour at each end. - One measured number. - Days 1-3 Scope Pick one application and one user group. - Days 4-7 Deploy The gate goes in front of your existing IdP. - Days 8-28 Observe Every attempt proved, resolved and recorded. - Day 30 Review Your own numbers in place of ours. Next ### Factor Assurance proves who got in. Next: what acted, and on whose authority. Once agents are in, they act with people's access. An Agent Access Review shows which agents are running, what each can reach and whose authority they act on. Agent Authority is designed to prove every action from there. --- # Agent Authority - Authority to Act | AuthLN URL: https://authln.com/agent-authority Agent Authority Authority to act ## Every agent action traces back to a person, within limits that only narrow. Coming soon Agent Authority is coming soon. Nothing on this page is available yet. A person grants an agent authority. Every hand-off passes on less, never more, and every action is checked before it reaches your systems. On the record. Every grant, action and refusal down the chain lands on a record no agent can rewrite. The spread ### One approval. Everything it starts. Your systems. A person approves one agent. That agent starts as many sub-agents as the work needs, and each one inherits the reach it was given. **Nobody approved the rest** A person looked at one agent. Everything it starts inherits that reach, and no one is asked a second time. **They outlive the task** A sub-agent that finishes its work keeps what it was handed until somebody goes looking for it. **The record says one name** Every call arrives under the same credential, so the log tells you a person acted and nothing more. More rules do not change the spread. A limit carried by everything the grant starts does. What it returns ### The answer is accurate. The account is still missing. The agent writes its own account, and nothing outside it verifies the accuracy. What it leaves you unable to say - What work actually happened, and from what source - Whether data left the building - Whether anything was refused, and on whose authority An account you cannot check is not evidence. The record has to be written outside the system that acted, where the agent cannot reach it. The answer Coming soon ### Eight problems. One place to stop them all. Every one of them - Overreach - Hijacked agents - Runaway agents - Stolen credentials - No way back - No trusted record - Unknown agents - Endless authority Each one is an action taken with a person's authority, unlimited and unrecorded. Meets the same three things **01 A person's grant** Authority comes from a named person and only narrows as work passes down. **02 A limit no agent can raise** Each agent can do only so much, and no agent can give itself more. **03 A record out of reach** Written once and held where the thing being recorded cannot edit it. We do not solve eight problems eight ways. Every action meets one gate, and the gate answers all eight. Coming soon The outcome ### Identity says who it is. The record says what it did. Every action is tied to the person who allowed it and recorded as it happens. None of it is accounted for by the agent. - Who - Action - Outcome - sub-agent 1 - read CRM - Allowed - sub-agent 4 - update ticket - Allowed - sub-agent 2 - send email - Allowed - sub-agent 4 - export contacts - Allowed - caller, unverified - request customer records no grant, flagged - Refused Without it, the row names whatever identity the call used, a shared agent or the person's own token, and the refusal isn't there at all. - 01 Actions cannot hide Nothing happens outside a person's grant, so every action traces to the person who granted it. - 02 Your own AI gets protected too A manipulated agent can be talked into attempting something. It cannot be talked into authority it was never granted, and everything it tries is on the record. - 03 You stop taking its word The reason for an action is recorded as it happens, not narrated by the agent afterwards. Coming soon Four questions ### The same four questions, answered. - Problem - The question - The answer - 01 Runaway - How many changes can one agent, and the sub-agents it starts, make in our systems? - No more than a person granted, and it stops there. - 02 No way back - When we revoke a person’s access, do the agents acting for them lose it at the same moment? - Yes. One revocation stops every agent and sub-agent acting under that person’s grant. - 03 No record - If an agent is compromised, what did it do, and under whose authority? - Every grant, action and refusal is on a record the agent can’t reach, so you can trace each step back to a person. - 04 Overreach - When did an agent first reach sensitive data, and whose grant let it? - A sub-agent holds only what it was handed. The first reach of sensitive data is flagged to the person whose grant allowed it. One design answers all four. What you can do ### Know what your agents can do, limit it, and prove it. **01 Know** See every agent and whose authority it acts on. **02 Limit** Only narrows Limit how much each agent and its sub-agents can do. Stop every agent acting for a person, at once. **03 Prove** Prove what happened without taking the agent's word. Your own team, your auditors and your customers. Design partners ### Agent Authority is coming soon. We are forming a small group of design partners who want these questions answered, and whose environments shape the final product. The way to start today is an Agent Access Review, to see which agents are running, what each can reach and whose authority they act on. --- # Agent Access Review - Know What Is Acting | AuthLN URL: https://authln.com/agent-access-review Agent Access Review Available now ## How many agents are running in your environment today, and who owns them? Get an inventory of the agents acting in your environment, a map of what each can reach and under whose authority, and a fix plan your people own. We map it from read-only exports. Nothing installed. You get a plan to close the access gaps we find. Can you answer the question And the questions that follow ### The questions the review answers - 01 How many changes can one agent, and the sub-agents it starts, make in our systems? - 02 When we revoke a person’s access, do the agents acting for them lose it at the same moment? - 03 If an agent is compromised, what did it do, and under whose authority? - 04 When did an agent first reach sensitive data, and whose grant let it? Agents are being turned on across the tools you already use. They act with people's access, and most teams cannot say what each one can reach or how to stop it. An Agent Access Review answers it. What you get ### Answers from your own records, and a plan your people own. You receive a report, a full agent inventory and an evidence index. Every finding traces to the export rows it rests on. Where your records can't show something, the report says so and counts it as a gap, not as zero. - Findings report - Agent inventory, as a spreadsheet - Evidence index: every finding traced to its source records - Two-hour fix-plan workshop with your owners - Written confirmation that your data was deleted **Your people own the plan** We run the review with your team, so every item has an owner before we leave. **A game plan you can run** The inventory, the access map and the plan, in your hands at the end of the review, with findings mapped to the access control and audit requirements CMMC examines. The report, section by section ### Ten sections. Each one answers a question. Where you stand Sections 1-2 **1 Summary** Answers: Where do we stand? You'll see: The answer in one paragraph and six numbers: agents found, agents unknown to security, agents with nothing limiting their changes, leavers whose agents kept working, agents that reached sensitive data, and gaps found. Illustrative “61 agents found. Security knew about 22.” **2 Answers to the five questions** Answers: The questions above, from your own records. You'll see: Each question answered, marked Answered, Partly answered, or Can’t tell from your records. Illustrative “Measured for 14 leavers. For 6, an agent or token kept working, from 9.5 hours to 59 days.” What is there Sections 3-6 **3 Agent inventory** Answers: How many agents are running, and who owns them? You'll see: Every agent by kind, whether security knew about it, whether it acts under its own identity or a person’s, what limits it, and who owns it. Illustrative “12 coding agents on laptops. None approved, none known to security.” **4 Delegation map** Answers: Under whose authority does each agent act, and which agents start others? You'll see: Chains from person to agent to sub-agent to what it reaches, with each link marked as seen in your logs or possible but not recorded. Illustrative “A coding agent runs with the developer’s own access to the manufacturing database. Whether it starts sub-agents is not recorded.” **5 Revocation lag** Answers: Do agents stop when their person leaves? You'll see: For each person who left in the window, how long their agents and tokens kept working after their account was disabled, and why. Illustrative “A cloud key on a departed analyst’s account was still in use 59 days later.” **6 First contact with sensitive data** Answers: When did an agent first reach your most sensitive data, and whose grant let it? You'll see: For each system on your crown-jewel list, which agent reached it first, when, and through whose grant, or why your records can’t tell. Illustrative “A widely shared drawings site let the tenant assistant return controlled drawings to a user who had never opened the site.” What to do Sections 7-10 **7 Gaps and fix plan** Answers: What do we fix first? You'll see: Gaps ranked by severity, each marked as a visibility gap (you can’t see what agents do) or a control gap (agents can do more than intended). Each has a fix using tools you already own, an owner and a date agreed at the workshop. Illustrative “Disabling a person doesn’t reach cloud keys or developer tokens. Fix within 30 days. Owner: identity lead.” **8 CMMC practices affected** Answers: What does this mean for our assessment? You'll see: Each gap mapped to the CMMC Level 2 practices it bears on. This is evidence for your own readiness work. It is not an assessment, and AuthLN is not a C3PAO. **9 What the fix plan can’t close** Answers: What’s left after we fix everything? You'll see: The limits that remain with today’s tools: a limit on one agent doesn’t carry to the agents it starts, each system revokes access on its own schedule, and the record is kept by the systems the agent acts in. **10 Evidence and method** Answers: How sure can we be? You'll see: What was received, what didn’t apply, what wasn’t available, how records were linked across systems, and how your data was handled and deleted. Examples are illustrative and drawn from a fictional sample report. Figures show the shape of each answer, not measured results. A full sample is available on request. What we need from you ### Read-only exports from tools you already run. Your administrators export records and send us the encrypted files. We join them into one map of who delegated what to which agent, and what each agent then did. A record that is missing, kept too briefly or editable is a finding, and so is a source you don't have. - 9 Send first Gets the review started. Users, groups and admin roles; leavers with exact cut-off times; connected apps and their consents; AI platform accounts and key IDs, never keys; and the 20-minute questionnaire, among others. - 8 Send next Per system in scope. Service accounts and cloud keys, developer tokens and bots, agents built in business platforms, AI tools on employee computers, and cloud audit logs, among others. - 8 If you have it Sharpens the map, not needed to start. Role and group changes, agent run traces, data labels and data-loss alerts, network traffic to AI services, and more. Data handling ### Before you send: what we never need. **Never send secrets** No passwords, API keys, tokens or connection strings. We need each credential's name, owner, scopes and last use. Remove any column holding a secret value. **Leave out content** We need who acted, when and on what, never the content itself. Drop prompts, message bodies and file contents. **Keep the IDs that link records** We join records on user principal names, app and client IDs, and role names. To pseudonymize people, use one keyed hash (HMAC) across every export. **Read-only, nothing installed** CSV or JSON as the system exports it, or a time-limited read-only audit role agreed at kickoff. The last 90 days for every source, 30 at minimum. **Inside GCC High or GovCloud** If anything is CUI or export-controlled, we can work inside your GCC High or GovCloud tenant instead, and only the report leaves. **Deletion, in writing** We agree the channel and encryption at kickoff. We delete everything at the end and confirm it in writing. Scope. One identity tenant (Entra ID, Okta or Google Workspace) plus the connected systems you pick at kickoff. Commercial, GCC High and GovCloud tenants count separately. HR, device and network inputs don't count as systems. What it costs ### A fixed scope, agreed up front. Scope is agreed at kickoff: one identity tenant plus the connected systems you pick. Pricing on request Book a call and we will confirm scope and price for your environment. Available now ### Know what is acting with your people's access. From read-only exports, with nothing installed. You get an inventory, a delegation map, a fix plan your people own, and written confirmation that your data was deleted. --- # The Record - Transparency Is the Product | AuthLN URL: https://authln.com/the-record The record Transparency is the product ## A record no one can rewrite. Every sign-in attempt, every grant, every action and every refusal lands on a record the actor cannot rewrite. Transparency is the product, not a feature. You check the record instead of taking anyone's word for what happened, including ours. Why it sits outside ### Written by something other than the thing that acted. When an agent writes its own account, nothing outside it can verify the accuracy. The record is written once and held where the thing being recorded cannot edit it, so it can be checked. The system that acted Apps, agents and your identity provider act here. None of them can edit what lands on the right. The record · write-once Illustrative entries - 8f2c62 Sign-in Authorized outcome, correlation ID, timing - 8f2c55 Sign-in Walked away the missing device signal is the finding - g-0417 Grant authority, from a named person what it covers, and who it was handed to Coming soon - a-0418 Action sub-agent 4: update ticket within the grant, on the record Coming soon - r-0419 Refusal caller, unverified: request customer records no grant, flagged Coming soon Without it ### An account you cannot check is not evidence. Today the only account of what an agent did is usually the agent's own log, written by the system that acted. It acts, it writes the account, we read the account, and we call it fine. The record breaks that loop: it is written outside the system that acted, so anyone who needs to can check it. What gets recorded ### For a sign-in, for a grant, for an agent action. **For a sign-in** - Outcome + correlation ID. Authorized, Denied or Walked away. - Dwell + 4-stage timing - IP, ASN and geo-city - VPN, Tor and proxy flags, when the network exposes it - Device GPS + accuracy, and origin trust class, when they engage A walk-away still leaves a record. The missing device signal is the finding. **For a grant** - The person who gave it. A named person. - Who it was handed to, and every hand-off after it, each passing on less - Any revocation, which stops every agent acting under that grant Authority comes from a named person and only narrows as work passes down. **For an agent action** - The agent that acted, and whose grant it acted on - The action, and what it reached - The reason, recorded as the action happens - A refusal: the moment it was stopped, and why The reason is recorded as it happens, not narrated by the agent afterwards. Who it is for ### Your own team, your auditors and your customers. **Your team** Every event arrives resolved, with an outcome and an ID you can act on and join to the logs you already keep. **Your auditors** A durable, correlated decision record, auditable months later, written once and out of reach of the thing it describes. **Your customers** Proof of what happened, and on whose authority, that does not rest on the word of the system that acted. Start here ### Start with what is running today. Book an Agent Access Review. From read-only exports, with nothing installed. You get an inventory of your agents, a map of what each can reach and under whose authority, and a plan your people own. And the questions that follow - 01 How many changes can one agent, and the sub-agents it starts, make in our systems? - 02 When we revoke a person’s access, do the agents acting for them lose it at the same moment? - 03 If an agent is compromised, what did it do, and under whose authority? - 04 When did an agent first reach sensitive data, and whose grant let it? --- # Security and Trust | AuthLN URL: https://authln.com/security Security and trust ## What we hold, what we don't, and how to check. We ask you to check the record instead of taking anyone's word. The same goes for us: here is what we hold, what we never hold, and where to check. Assurance ### Patents, audits and the Trust Center. Patents **Two issued U.S. patents** US 11,956,366 and US 12,118,550. SOC 2 **Type II certified** SOC 2 Type II certified. Trust Center **trust.authln.com** Policies, controls and documents for your security review. Data handling principles ### The least we need, for the shortest time. **No keys to steal** AuthLN does not store end-user private cryptographic keys. The proof is produced on the person's device, never on yours. **Your identity provider stays yours** We hold no directory and no lifecycle. Your directory, your policy and your lifecycle are unchanged, and enforcement policy and group scoping live in your tenant, not ours. **Read-only, and never secrets** An Agent Access Review runs from read-only exports, with nothing installed. We never need passwords, API keys, tokens or connection strings, and we ask you to leave out content: prompts, message bodies and file contents. **Deleted, and confirmed in writing** Review data is deleted at the end of the engagement, and we confirm it in writing. Authentication telemetry is retained no longer than 45 days by default, as our Privacy Policy sets out. **Inside your own boundary** If anything is CUI or export-controlled, the review can run inside your GCC High or GovCloud tenant, and only the report leaves. Data processing aligns with NIST SP 800-53 Rev. 5 and ISO/IEC 27001:2022. See the Data Processing Addendum . Documents ### Documents - Incident Response How we identify, contain and resolve security events - Privacy Policy What we collect and why - Terms of Service The terms of using AuthLN - Data Processing Addendum How we process data on your instructions Questions ### Security questions? Ask us directly. Talk to us about our security program, subprocessors or how a review handles your data. --- # About AuthLN - The Auth Lineage Network URL: https://authln.com/about About Auth Lineage Network ## AuthLN is the Auth Lineage Network. Lineage means any action can be traced back through every agent and hand-off to the person whose authority it ran on. When AI agents hand work to other agents, no one can say who authorized each action or show that it stayed within its limits. We are building the network that can. The mark ### The fan in the logo is that lineage. One person on the left. Authority branching and narrowing to the right: each node smaller than the one before, every branch traceable back to where it started. The same shape runs through every graphic on this site, because it is the thing we prove. What we do ### Everything is about authority. **Authority to enter** Who is allowed in. Every sign-in attempt resolves to Authorized, Denied or Walked away. **Authority to act** What software may do once it is in, including AI agents acting for a person. **Proof of both** Every sign-in attempt, grant, action and refusal lands on a record the actor cannot rewrite. The team ### The expertise to run it, and the people ready to build it. **Mike Siers** Founder and CEO Repeat entrepreneur and military combat veteran. Inventor of AuthLN’s patented approach to identity security. **Dennis Edwards, PhD** Cofounder and CTO Thirty years across academia and industry: fifteen as a computer science professor, over a decade leading system architecture and cybersecurity. **RADM William E. Chase III, USN (Ret.)** Advisor Thirty-three years in military cyber operations. As Deputy Commander of Joint Force Headquarters - DoD Information Network, he was responsible for defending one of the largest and most-targeted networks in the world. Start here ### Start with what is running today. Book an Agent Access Review. From read-only exports, with nothing installed. You get an inventory of your agents, a map of what each can reach and under whose authority, and a plan your people own. And the questions that follow - 01 How many changes can one agent, and the sub-agents it starts, make in our systems? - 02 When we revoke a person’s access, do the agents acting for them lose it at the same moment? - 03 If an agent is compromised, what did it do, and under whose authority? - 04 When did an agent first reach sensitive data, and whose grant let it? --- # FAQ | AuthLN URL: https://authln.com/faq-1 FAQ ## Questions, answered. The questions we hear most. For anything else, talk to us. AuthLN proves authentication and authorization: who was allowed in, what was allowed to act and on whose authority, on a record no one can rewrite. It brings together Factor Assurance (authority to enter, shipped), Agent Authority (authority to act, coming soon) and the Agent Access Review (available now). AuthLN is the Auth Lineage Network. Lineage means any action can be traced back through every agent and hand-off to the person whose authority it ran on. The fan in the logo is that lineage: one person on the left, authority branching and narrowing to the right. The decision layer in front of your identity provider. Every sign-in attempt resolves to Authorized, Denied or Walked away, and arrives in your SOC as one event with an outcome and an ID. It is shipped and in production. No. It stands in front of Okta, Microsoft Entra ID, Google Workspace and JumpCloud, which stay exactly as they are. Your directory, your policy and your lifecycle are unchanged, and nothing is rerouted. An attempt that saw the gate and left. Most systems only record a success or a failure. A walk-away still leaves a record, and a walk-away from an untrusted origin is itself a signal. A device-bound passkey whose private key is generated inside the hardware security perimeter and cannot be exported; proximity-based authentication, which checks device GPS against network geolocation; and an ML-DSA-65-ES256 hybrid passkey with post-quantum signatures. All three are shipped. Not yet. Agent Authority is coming soon. It is built so every agent action traces back to a person, within limits that only narrow, checked at a gate and recorded where the agent cannot reach. We are forming a small group of design partners. A review of the agents acting in your environment, available now, run from read-only exports with nothing installed. You get an inventory of the agents acting in your environment, a map of what each can reach and under whose authority, and a fix plan your people own. Read-only exports from tools you already run. Never secrets: no passwords, API keys, tokens or connection strings, and no content such as prompts, message bodies or file contents. If anything is CUI or export-controlled, the review can run inside your GCC High or GovCloud tenant. We delete everything at the end and confirm it in writing. Outside the thing being recorded. The record is written once and held where the system that acted cannot edit it, so it can be checked by your team, your auditors and your customers. SOC 2 Type II certified. Two issued U.S. patents: US 11,956,366 and US 12,118,550. Start here ### Start with what is running today. Book an Agent Access Review. From read-only exports, with nothing installed. You get an inventory of your agents, a map of what each can reach and under whose authority, and a plan your people own. And the questions that follow - 01 How many changes can one agent, and the sub-agents it starts, make in our systems? - 02 When we revoke a person’s access, do the agents acting for them lose it at the same moment? - 03 If an agent is compromised, what did it do, and under whose authority? - 04 When did an agent first reach sensitive data, and whose grant let it? --- # Incident Response | AuthLN URL: https://authln.com/incident-response Trust and security ## Incident response AuthLN maintains a structured incident response program designed to quickly identify, contain, and resolve security events while minimizing impact to our customers. Our approach ### A defined lifecycle, applied consistently. - 01 Detection and reporting Events are identified through monitoring, internal reporting, and customer notifications. - 02 Triage and investigation Events are assessed, classified by severity, and investigated by our engineering team. - 03 Containment Immediate actions are taken to limit impact and stop ongoing threats. - 04 Recovery Systems and services are restored to normal operation. - 05 Remediation Root causes are addressed and controls strengthened to prevent recurrence. Severity classification ### We prioritize by potential impact. **Critical** Active threats or confirmed compromise requiring immediate response. **High** Credible risks or vulnerabilities with potential for exploitation. **Medium and low** Suspicious or unverified activity requiring investigation. Response and escalation. Our engineering leadership oversees all incident response activities, coordinating investigation, containment, customer communication, and post-incident review. Questions ### Security questions? Let's talk. Reach out about our security program, subprocessors, or compliance posture. --- # Book an Agent Access Review | AuthLN URL: https://authln.com/contact Contact Book a review ## Book an Agent Access Review, or ask us anything. Pick a time for a 30-minute call. Tell us about your environment and we will scope a review, a Factor Assurance pilot, or a conversation about Agent Authority design partnership. On the call - Which identity tenant and connected systems are in scope - Who on your side owns the plan - How files are sent, and how we delete them Prefer email? sales@authln.com Want to read first? See what the review covers . Booking needs JavaScript. Email sales@authln.com and we will find a time.