Hiyaku
AI Tokenomics · Impact study

FlySBS × Hiyaku

Hiyaku builds one live model of a business and runs governed agents on it, so AI spend can be priced against what it actually produced.

01 / The problem

Three pressures, and a budget line answers none of them.

Token cost is rising +47%

AI spending reaches $2.59 trillion in 2026. Per-token prices have been falling the whole time and no enterprise's bill has followed them down.

Compute demand is rising faster 1,000×

An agentic task can consume a thousand times the tokens of a chat turn, and the same task run twice can differ by thirty. Demand compounds faster than price falls.

The outcome is inefficient 93%

of enterprises are already over their AI budget, and only a fifth can tie any of that spend to a business result.

Inside one task, with nothing governing it One agent · no gateway
A charter enquiry handed to a single agent with no control plane. Every step re-reads everything before it, so input grows with the square of the work, not with the work. Six steps cost 126k input tokens, all on a frontier model, none cached, and the bill has no enquiry, trip or customer to land on. The governed run of the same enquiry, shown in section 03, takes 41.2k. Figures are representative.

FlySBS had all three in miniature. A charter enquiry is decided within hours, first come, first served. Pricing one took two days — elapsed time spent assembling the cost by hand, line by line, while that window closed.

They were not losing on price. They were arriving after the decision had already been taken.

02 / What Hiyaku rethinks

Build the denominator before you read the meter.

Tokens are easy to count: they arrive on a vendor invoice, itemised, every month. Outcomes are hard to count, because in most companies they are scattered across spreadsheets, inboxes and the ledger, and nothing knows that a quote, a trip and an invoice are one piece of work.

You cannot divide by an outcome you have never modelled.

So Hiyaku inverts the usual order. Model the business first, give it one governed way to be written to, and only then let agents run on it. Attribution stops being a tagging programme and becomes a property of the architecture.

03 / How Hiyaku manages tokenomics

One enquiry, from landing to what it cost us.

The agents — enquiry to price forecast

Every token above is already charged to the enquiry it served.

    Tokens · this enquiry41.2k in · 2.1k out
    Routingsmall ×3 · frontier ×1
    AI cost₹4.18
    Charged toENQ‑4471 · aircraft · customer
    One charter enquiry, from a WhatsApp message to a price forecast. Each tower is one agent's input: a cube is a thousand tokens, dark cubes came from cache, amber cubes went to a frontier model. Switch to One ungoverned agent to run the same enquiry with no control plane and watch each step re-read the ones before it.
    AgentReadsWritesRouted toTokens in / out
    1 · IntakeThe WhatsApp messageEnquiry record, confirmed by OperationsSmall model6.4k / 0.4k
    2 · JourneyLeg, AircraftAir time, ground time, block timeSmall model5.1k / 0.3k
    3 · CostRate Card, CrewThe cost lines of the quoteSmall model11.3k / 0.5k
    4 · ForecastPast quotes and invoicesPredicted costs and the price bandForecast model, one frontier call to explain it18.4k / 0.9k
    Charged to ENQ‑447168% of input served from cacheSmall ×3 · frontier ×141.2k / 2.1k

    The same enquiry on the quote desk

    Quote desk — ENQ‑4471 One charter enquiry
    Captured from
    ChannelWhatsApp
    Received15:04:12 IST
    Logged byassistant, confirmed
    Source & segment tagged
    Journey
    Routecaptured
    Datecaptured
    Passengerscaptured
    Aircraftassigned
    Clock
    HoursDecision window opens
    Customerexisting
    Illustrative reconstruction of the quote desk. Identifiers, timings and monetary figures are representative and are not FlySBS's own.
    04 / From cost per token to cost per outcome

    Four moves, and the unit of account changed.

    Cost per token → Cost per flown trip

    One model of the business

    Customers, enquiries, quotes, trips, legs, aircraft, crew and invoices became one thing the whole company reads. Nothing was ripped out; the ledger stays the ledger.

    One way anything is written

    People and agents go through the same governed door. Every AI call lands against a customer, a trip and a cost, not a line on a bill nobody can allocate.

    Drafts, not decisions

    The assistant proposes and a person confirms. Full autonomy stays a decision the business takes deliberately. An unbounded agent loop is an unbounded bill.

    AI-driven price forecasting

    Leg and ground costs are forecast from FlySBS's own history, with a price band drawn off quotes won and lost, and the model retrains on what each trip actually burned. Trained on their data, not guessed at by a frontier model.

    05 / What it creates

    What FlySBS runs on now.

    ~190×

    Faster to a priced quote

    Two days of elapsed time became under fifteen minutes, with no estimator tied up for any of it.

    ±10%

    Forecast against the manual price

    Back-tested on FlySBS's own history before it priced a live one.

    In time

    The window it now meets

    A decision window measured in hours, answered by a quote measured in minutes.

    1:1

    Trip to profit and loss

    Every trip closes on real fuel and handling, so revenue per block hour is measured.

    Hiyaku

    A token is what you spent. A flown trip is what you bought.

    Talk to us about your AI spend →← All insights

    HIYAKU SYSTEMS PRIVATE LIMITED
    The Enterprise Autonomy Solution
    www.hiyakusystems.com · Newtown, Kolkata 700156, India
    Market figures: Gartner, McKinsey — AI Tokenomics & AI FinOps, Nov 2026.
    Films and quote desk are illustrative reconstructions; token and rupee figures are representative.
    FlySBS figures are engagement targets.

    About these figures. The four above are the engagement's targets, back-tested against FlySBS's own history — not results claimed from a finished deployment.

    Contact us

    Get started.

    Tell us about one decision your team makes every day: a reroute, a stock move, a payment hold. We will show you how far toward autonomy it can go.

    We use these details only to reply to you. See our privacy policy.