Control plane · allows
Governed Intelligence Platform
elan1 Govern
The governed write path for AI agents.
“May this record change?”
allow · refuse · approve · seal
Explore GovernAI agent governance · The governed write path
Every action, answerable.
elan1 is the governed write path for AI agents, built by ELAN1 AGENTIC TECHNOLOGIES PRIVATE LIMITED in Tamil Nadu. Every consequential action an agent proposes is checked against your rules, approved by a named person where your rules ask, and written to a tamper-evident, hash-chained proof record. A refusal costs 40% of one governed action.
Available now: elan1 Govern, elan1 Work, elan1 Decide and elan1 Run.
The problem
Agents can now write to the systems that run your business.
An agent that drafts an email saves time. An agent that posts a journal entry or changes a claim commits your business. Your auditor will not ask what the model said. They will ask three questions.
- Who allowed it?
- Under which rule?
- Where is the record?
elan1 answers all three on the write itself: a named approver, a versioned rule, and a 16-field record.
The one rule
A model may produce a write intent. It may never produce an allow.
That sentence is the design. An agent, a decision model or a third-party tool can propose a typed change. Only elan1 Govern can allow it, using versioned rules, not a model. If Govern cannot decide, the change waits for a person. We fail toward a human, never toward an action.
- No model sits on the allow path.
- A refusal returns a typed code and still writes a record.
- A refusal is metered at 0.4 of a governed action.
The gateOne line in · one line out
Intent at the gatePlays in view · 3 cycles
elan1 Decide proposes claim.settle · ₹42,000 · 13th this hour, rate 12
Refused · GATE.RATE_EXCEEDED (illustrative) · record written · 0.4 governed action
| Seq | From | Write | Outcome | Meter | Chain |
|---|---|---|---|---|---|
| 3106 | Decidedecision model | claim.settle₹42,000 | RefusedGATE.RATE_EXCEEDED* | Meter 0.4 governed action | prev 5897…6f15 → dece…2772 |
| 3105 | Third-party agentMCP · Mode B | vendor.bank_changeno amount | Heldusr_specimen_approver | Meter not yet metered | prev 6e0c…a92a → 5897…6f15 |
| 3104 | Workagent | purchase_invoice.post₹62,000 | Allowedcommitted | Meter 1 governed action | prev f714…3ac2 → 6e0c…a92a |
| 3103 | Decidedecision model | claim.settle₹2,10,000 | Heldusr_specimen_claims_head | Meter not yet metered | prev 5052…2656 → f714…3ac2 |
| 3102 | Third-party agentMCP · Mode B | payment.release₹58,000 | Allowedyour system writes | Meter 1 governed action | prev 0496…8661 → 5052…2656 |
| 3101 | Workagent | credit_note.issue₹12,600 | RefusedGATE.PRECONDITION_FAILED* | Meter 0.4 governed action | prev 152a…c038 → 0496…8661 |
*Refusal codes are illustrative. Each digest is SHA-256 of seq, write type, outcome and the previous digest: a short illustrative preimage, not the Seal v6 record.
One write path, four layers
Two layers propose. One carries the work. Only one allows.
elan1 Govern is the control plane. elan1 Work and elan1 Decide propose changes. elan1 Run carries every call under contract. Every consequential write passes through Govern.
Proposes · what should change
Enterprise Intelligence Platform
elan1 Work
Governed agents for the work that commits your business.
“What should change in the books?”
Meter: credit
Explore WorkProposes · the answer
Decision Intelligence Platform
elan1 Decide
Typed, calibrated decisions, routed by risk.
“What is the right answer?”
Meter: decision
Explore DecideCarries every call
Infrastructure Intelligence Platform
elan1 Run
The contract layer and Day-2 operation for the open-source AI stack.
“Did the call stay within contract?”
Meter: node-year
Explore RunWork and Decide propose. Govern and Run assure. Only Govern allows.
See the whole architectureHow it works · The Right Path
Ask, check, approve, act, record. Every time.
Every consequential change takes the same path, wherever it starts. Five plain steps map onto rungs 0b STOP to 9 METERED. Any rung can refuse.
- 01
Ask
An agent or the Decide router builds a typed write intent. elan1 checks that writes are not stopped, who is asking, and which kind of write this is.
0b STOP · 1 IDENTITY · 2 WRITE DOOR
- 02
Check
Nine pre-gates (G1–G9) run against your own data. Six of them can never be skipped. Versioned rules then evaluate the change, with no model. A guardrail measures the blast radius.
3 PRE-GATES · 4 evaluate() · 5 GUARDRAIL
- 03
Approve
The autonomy level decides whether a person must say yes. If one must, a named approver who is not the asker sees the difference, the gates and the policy.
6 AUTONOMY
- 04
Act
One writer, holding a single-use credential that lasts 10 minutes, makes the change. It goes to a sandbox first.
7 ACT
- 05
Record
A 16-field record, hash-chained to the one before it, and one governed action counted. A refusal writes a record too, and counts as 0.4 of a governed action.
8 PROOF · 9 METERED
Walk one write up the line
Walk through every rungProposedMeter: Nothing counted yet
- 0bSTOP
- 1IDENTITY
- 2WRITE DOOR
- 3PRE-GATES
- 4evaluate()
- 5GUARDRAIL
- 6AUTONOMY
- 7ACT
- 8PROOF
- 9METERED
A write intent is proposed. Press Next rung, or the → key, to send it up the path.
Write intent: “Post purchase invoice · ₹4,82,000 · proposed by agent.” Illustrative specimen. Not customer data. You advance it; nothing plays on its own. Use the → key or the button.
The record
Every write leaves 16 fields. Never a seventeenth.
Each record is hashed with SHA-256 over a length-prefixed preimage and chained to the one before it. Change a past record and the chain shows it. That is what tamper-evident means at L1.
Checked and recorded by elan1 Govern.
Plain English · Illustrative specimenL1 · Guarded
Record 48,213 in this tenant’s chain. An agent proposed posting a journal entry under declaration ledger1 GL posting, version 7, and policy version 12, at autonomy level A1. The entry was written as record JE-specimen-0193, version 1. Its payload and vocabulary are hashed, and it points to the previous record by digest.
Checked and recorded by elan1 GovernIllustrative specimen. Not customer data.
Illustrative specimen — not a customer recordL1 · Guarded
Proof record · Seal v6 · 16 fields “no seventeenth field, ever”
- F01
- tnt_specimen_01
- F02
- 48213
- F03
- sha256:7c41…a9e2
- F04
- wr_specimen_7Q2M
- F05
- journal_entry.post
- F06
- decl.ledger1.gl_posting
- F07
- 7
- F08
- 12
- F09
- agt_specimen_ap01
- F10
- agent
- F11
- A1
- F12
- JE-specimen-0193
- F13
- 1
- F14
- sha256:e09b…41cd
- F15
- sha256:3a6f…0b17
- F16
- 2027-07-14T10:42:07+05:30
Tap, point at or focus a field name to read what it holds. Every term is in the glossary.
Committed · metered: 1 governed action
Checked and recorded by elan1 Governtamper-evident (L1)
How to check itL2 · Available
- At L1. Recompute this record’s digest and compare it with prev_digest on the next record. If a past record changed, the chain shows it.
- At L2. Each record is also signed with Ed25519 and anchored to two RFC 3161 timestamp authorities.
- Offline, at L2. The free offline verifier checks it on your own machine, without us.
Tamper-evident at L1. Signed and checkable offline at L2.How verification works
The chain · four recordsL1 · Guarded
- Record 1041, credit_note.issue, ₹6,250, digest 8231…517c, the first record shown.
- Record 1042, journal_entry.post, ₹18,400, digest cbb7…fce1, links to 1041.
- Record 1043, payment.release, ₹36,500, digest f7c9…c30c, links to 1042.
- Record 1044, journal_entry.post, ₹9,800, digest 05bf…8549, links to 1043.
Head digest you already hold
05bf…8549
matches record 1044
Chain intact for seq 1041–1044.
Hiding one change means rewriting every record after it, and the digest you already hold still disagrees. That is what tamper-evident means at L1. From L2, roots anchored with two timestamp authorities show it too.
Illustrative specimen. Not customer data.SHA-256 recomputed in your browser
Assurance levels
Three levels of a claim. We tell you which one we are making.
Levels describe what you can rely on. They are not tiers you buy.
L1 Guarded
“Can I let this run?”
The Right Path, approvals, a single writer and a tamper-evident record you can read.
AvailableTamper-evident records you can read, today
L2 Attested
“Can I hand this over?”
A signature, anchoring, an evidence bundle and a free offline verifier: a signed receipt you can hand to your auditor.
AvailableSigned records you can check offline, without us
L3 Proven
“Does this survive the vendor?”
A receipt that outlives the vendor.
On auditor acceptanceReached when an outside auditor accepts an L2 bundle in writing
Mode A · EnforcedMode B · Attested advisory
In Mode A, elan1 holds the write credential, and only there do we say enforced. In Mode B, over MCP or HTTP, every write intent is evaluated and recorded. We call that attested advisory, never blocked.
What elan1 stands on
Governed by design. One control plane. Configured for the industry.
01 · Governed by design
The rule check, the approval and the record sit on the write itself, not in a report afterwards.
02 · One control plane
One kernel. No product builds its own gates, approvals, proof or meter.
03 · Configured for the industry
Packs configure Govern for a regulation. ledger1 covers the MCA audit trail, invoice1 covers GST and collect1 covers RBI collections. Each pack is configuration, never new code.
For your role
Start with your first question.
Leads with 03 · Configured for the industry
The agent drafts it. You commit it.
- Start with one process. Add the next when it holds.
- Approvals reach the process owner in the console, on WhatsApp or by voice.
- Packs set up the checks for your regulation.
Approvals · Illustrative specimen/work/approvals
| Time | Write | Amount | State |
|---|---|---|---|
| 10:42 | Post purchase invoice Awaiting finance head | ₹4,82,000 | Held |
| 10:31 | Post journal entry Approved by finance head | ₹62,400 | Committed |
| 10:18 | Change supplier bank G3 Permissions · 0.4 governed action | — | Refused |
Leads with 01 · Governed by design
A person approves it before it happens.
- Every write, and every refusal, leaves a 16-field record that is tamper-evident at L1.
- From L2, records are signed and checkable offline without us.
- We publish what we claim, and what we don't.
Proof record · 6 of 16 fields · Illustrative specimen/govern/records/1043
- F02seq
- 1043
- F03prev_digest
- sha256:4be1…90c7
- F05write_type
- purchase_invoice.post
- F09actor_id
- usr_specimen_approver
- F11autonomy_level
- A1
- F16recorded_at
- 2027-07-14T10:46:12+05:30
Committed · approved by a named person · tamper-evident (L1)
Leads with 02 · One control plane
One write path. Two legal crossings.
- KernelPort v1 has six calls: define, assert, seal, verify, scope, refuse.
- There is an MCP server and an HTTP API for the agents you already run.
- Ten fusion.*/v1 contracts, each with two licence-clean implementations.
KernelPort v1 · assert · Illustrative specimenkernelport.assert()
// illustrative — one of six calls: define, assert, seal, verify, scope, refuse kernelport.assert({ tenant_id: "tnt_specimen_01", write_type: "purchase_invoice.post", declaration_id: "decl_gst_purchase", actor: { id: "agt_specimen_ap01", kind: "agent" }, payload_hash: "sha256:c03a…11fe" }) // → held at rung 6 AUTONOMY: a named approver, not the asker, decides
Pricing
Paid per governed action. A refusal costs 40% of one.
Each layer charges for one thing it can count, and nobody pays twice for the same action. There is no per-seat price and no per-token price.
How pricing works| Layer | Unit | List price |
|---|---|---|
| elan1 Govern | Governed action | ₹1.60 per governed action beyond the L1 allowance. A refusal costs 40% of one: ₹0.64 per refusal at L1. |
| elan1 Work | Credit | ₹0.80 per credit beyond the plan allowance, a rendering of governed actions |
| elan1 Decide | Decision | ₹30 per 1,000 decisions beyond the plan allowance |
| elan1 Run | Node-year | ₹1,80,000 per node per year; ₹1,80,000 per GPU per year. Capacity, never tokens |
All prices are exclusive of GST. GST is added at the applicable rate on the invoice. A Work action that passes through Govern is counted once.
Design partners
We are looking for ten design partners.
Design partners are early adopters of the launched platforms. Start with one process and add the next when it holds. We meet for a 30-minute call every two weeks, and what you tell us shapes what comes next.
Apply to be a design partner- Indian businesses with ₹5–50 crore turnover, introduced by their CA.elan1 Work
- Operations owners in insurance claims and BFSI.elan1 Decide
- GPU operators and systems integrators.elan1 Run
- Agent vendors connecting through KernelPort.elan1 Govern
Design-partner register · IllustrativeTen places
- Place 01
- Place 02
- Place 03
- Place 04
- Place 05
- Place 06
- Place 07
- Place 08
- Place 09
- Place 10
No customer names, no logos. Here is what we can show you today.
Where we are
All four platforms are available.
elan1 Govern, Work, Decide and Run are launched. There are no logos here: we show what you can use, not who uses it.
- NowEvery platform and every product is available.We are here
- elan1 GovernL1 Guarded and L2 Attested: tamper-evident records, and signed records you can check offlineAvailable
- elan1 WorkERP, Revenue, Commerce and VoiceAvailable
- elan1 DecideAll six modules, from Decision Studio to Decision PacksAvailable
- elan1 RunAll five modules, with Runtime Controls in every subscriptionAvailable
- On acceptanceL3 Proven: reached when an outside auditor accepts an L2 bundle in writingOn auditor acceptance
Questions
AI agent governance, answered plainly.
elan1 is the governed write path for AI agents, made by ELAN1 AGENTIC TECHNOLOGIES PRIVATE LIMITED in Tamil Nadu. It sits between an agent and your records: every consequential change is checked against your rules, held for a named person where your rules ask, and written to a tamper-evident proof record. It comes as four platforms, elan1 Govern, Work, Decide and Run, and all four are available now.
AI agent governance is the set of controls that decides what an AI agent may change, who must approve it and what evidence is kept. For agents that write to business systems, it works best at the write itself: one path every change must take, with versioned rules, a named approver and a record of the outcome.
elan1 governs the action, not only the model. Inventory and risk-register tools record which models you use and which policies apply; elan1 sits on the write path, so each change an agent proposes is checked, approved where your rules ask, and recorded as it happens. The two can work side by side.
Only where your rules allow it. Each process has an autonomy level: routine changes within your limits can run alone, and anything your rules mark as consequential is held for a named person who is not the one who asked. Either way the change leaves a proof record, and in Mode A, where elan1 holds the write credential, the agent cannot write around the path.
No, and elan1 does not claim it is. At L1 Guarded the record is tamper-evident: records are SHA-256 hash-chained, so any later change to an earlier record shows when you recompute the chain. From L2 Attested, records are also signed, and an auditor can check them offline without asking elan1.
In India. elan1 runs in AWS ap-south-1 (Mumbai) across three availability zones, and proof records never leave the residency domain they were written in. Gulf and EU domains open only on a signed contract, and no US region is offered.
// GET STARTED
One agent. On the record.
Watch one write climb the Right Path, and another one get refused.