Skip to content
PilotLumenar is in pilot. Put one agent on one real workflow.

Agents that do the work, and show it.

Each agent writes its plan first, leaves a receipt for every step, and waits for a person before anything irreversible.

Sample data. Screens marked Example are samples while Lumenar is in pilot.

  • Runs on
  • llama.cpp today
  • vLLM and SGLang next
  • OpenAI-compatible API
  • Open models

Agents mostly re-read. In our own coding-agent sessions, 99.2% of all tokens were input, and 88.5% of that input had been sent before. So we built the layer underneath: it keeps each session on its cache, picks a model for each step, and leaves you a receipt.

  • 99.2%of all tokens in our own agent sessions were input, not output
  • 88.5%of those input tokens had been sent before
  • 1.43Bprompt tokens measured across 17,780 model calls

Measured on Lumenar’s own coding-agent logs.

What every agent does

Work you can check, step by step.

Agents are useful when you can see what they did and stop what they shouldn’t do. Each of these is built in, not bolted on.

Approval gates

Refunds, merges and anything else you mark irreversible wait for a person.

Step receipts

Each step shows what it did, the tool it called and its time. Each session shows steps · time · cost.

Plan first

No step runs before the plan is written down. You can approve it, change it or stop it.

Trace and audit log

The trail of every session is kept in an append-only log you can export.

Hand-offs to people

When a case needs judgement, the agent hands it to a named person with the context attached.

Stop, any time

A live timer runs while an agent works, next to a Stop button that ends the session at once.

How an agent works

Plan, run, approve, receipt.

Every session follows the same four steps, and each one leaves something you can read afterwards.

See the agents
  1. Step 1

    Plan

    The agent writes its plan first. A person can approve it or change it before any work starts.

  2. Step 2

    Run

    Each step is traced: what it read, which tool it called, how long it took.

  3. Step 3

    Approve

    Anything irreversible, like a refund or a merge, waits at a gate with Approve and Redirect.

  4. Step 4

    Receipt

    Every session ends with a receipt: steps, time and cost. The audit log is append-only.

The session layer

The layer underneath keeps long sessions cheap.

Requests from one agent session go to the GPU that already holds its KV cache, so old context isn’t computed again.

KV cacheSession routing

Cost of one 60-turn task

A model of how the cost falls as each mechanism is added. Not a measurement.

No reuse
$2.95
+ prefix cache
$0.47
+ cache and step routing
$0.30

Inferstack prototype

The session layer is running.

Inferstack is the gateway under our agents. On a laptop with no GPU, the second turn of a session already costs a fraction of the first.

  • 87%less served cost on the second turn, on the small model. 79.6% on the larger one.
  • 98.3%of the prompt came from cache on the second turn.
  • 0.24 sfor a repeated 4,298-token prompt from cache, against 73.4 s cold.

Measured on the Inferstack prototype: Qwen2.5 0.5B and 1.5B on llama.cpp, CPU only.

The agents

Each agent has a job title and a catalogue ID.

All agents

Support Agent

LMN-2041
Example

Answers tickets, looks up orders and drafts replies.

tickets.readorders.getpolicy.search

Waits for a person: Refunds and credits

Billing Agent

LMN-2077
Example

Handles invoices, failed payments and plan changes.

invoices.getpayments.retryplans.change

Waits for a person: Any refund or plan downgrade

Research Agent

LMN-2110
Example

Reads your data and writes the weekly summary.

warehouse.querydocs.searchreport.write

Waits for a person: Sending a report outside the team

Code Review Agent

LMN-2153
Example

Reviews pull requests and runs the tests.

repo.readtests.runpr.comment

Waits for a person: Merging or pushing to main

Onboarding Agent

LMN-2190
Example

Sets up new accounts and sends the welcome sequence.

accounts.createemail.draftcrm.update

Waits for a person: The first email to each account

Have a workflow in mind? Put one agent on it.Request a pilot

Software

Products with agents inside.

The same agents, inside calm products your team uses every day, on one account.

The software
Example

Lumenar Desk

A support inbox where an agent drafts the reply and a person sends it.

Example

Lumenar Ledger

Invoices and reconciliation, with an agent that matches payments and flags the odd ones.

Pricing

Pay for work that got done.

Example pricing while Lumenar is in pilot. Pilot terms are agreed with each team.

Compare plans
  • Pilot

    Example

    $0

    for 30 days

    One agent on one workflow, set up with us, to see the receipts on your own work.

    • 1 agent
    • Up to 500 resolved sessions
    • Weekly review with the team
  • Most teams start here

    Agents

    Example

    12 credits

    per resolved session

    Pay when an agent resolves a session, from one balance of Lumenar credits.

    • Any number of agents
    • Approval gates and audit log
    • $10 per 1,000 credits
  • Software

    Example

    $29

    per seat / month

    Lumenar products like Desk and Ledger, with agents inside.

    • Per-seat billing
    • Agents billed in credits
    • Every product, one account
  • Enterprise

    Example

    Custom

    annual

    Your models, your region, your terms.

    • Open models on your own GPUs
    • Data residency per workspace
    • A named engineer

Trust

Every step on the record.

  • Approvals

    Refund
    Waits
    Merge
    Waits
    Email to a customer
    Waits

    Approvals. Anything irreversible, like a refund, a merge or an email to a customer, waits at an approval gate. A named person approves or redirects it, and their name goes in the log.

  • Append-only audit log

    The plan
    Logged
    Each tool call
    Logged
    Each approval
    Logged

    Append-only audit log. Every session leaves a trail: the plan, each step, each tool call, each approval. Entries are added, never edited, and the whole log exports.

  • Open models, no lock-in

    vLLM
    Runs on
    SGLang
    Runs on
    API
    OpenAI-compatible

    Open models, no lock-in. The session layer runs open models on vLLM and SGLang and speaks the OpenAI-compatible API. The session log belongs to no single model, so you can switch models mid-session.

  • How session logs are used

    Run the session
    Yes
    Hand to another model
    Yes
    Show what happened
    Yes

    How session logs are used. A session log is the record of one agent’s work. It is used to run that session, to hand it to another model and to show you what happened. In a pilot we agree in writing what is kept, where, and for how long.

Planned, not yet earned: SOC 2 Type II · ISO/IEC 42001

Questions

Asked before a pilot.

Are these the real prices?

Not yet. Lumenar is in pilot, so every figure on this page is example pricing. Pilot terms are agreed with each team.

What is a resolved session?

One piece of work an agent finished, such as a ticket answered or a report written, and that wasn’t handed back unresolved. Sessions an agent hands to a person are not billed as resolved.

What are Lumenar credits?

One balance your whole organisation shares. Every agent draws from it, so you see one number, not one bill per agent.

Can we bring our own model?

Yes. The session layer runs open models on vLLM or SGLang and speaks the OpenAI-compatible API.

Put one agent on one real workflow.

A pilot is one agent, set up with us, on your own work. You keep the receipts.