Skip to content

About

Making sellers ready for the agents already asking about their pricing.

Buyer agents already compare offers, ask for quotes and build shortlists before a person signs. We build PairRail Atlas so sellers can answer them from one governed catalog: their prices and their rules. Why this is happening now.

QUOTE · PairRail

Our mission: give every seller full ownership of how agents discover their offers, get quotes and ask to buy.

Agents should get the rigor of an enterprise deal desk on every buying protocol, even as protocols version, fork or get replaced.

What we believe

  1. When an agent buys for a person, that person should get a fair, clear price.

  2. A price an agent can read is a price a person can check.

Commitments that shape what we ship

  1. Seller-side first

    one catalog

    We build for the teams who set and own the prices, not a marketplace for buyer agents. One catalog serves every agent protocol, and you choose which offers each agent can see.

    Example: one catalog, what each agent sees
    No key
    3 public offers
    Customer key
    3 public + 2 customer
    Internal offers
    never shown

  2. No lock-in

    export

    Export your catalog and data whenever you like, and close your workspace from Settings. Approved quotes go to the billing system you already run.

    What the export file holds
    format
    pairrail.workspace-export
    includes
    catalogVersions · quotes · approvals · agentKeys · auditEvents
    Secrets and tokens
    never included

  3. Focused, not exclusive

    usage-priced

    We start with usage-priced sellers: compute, data APIs and AI infrastructure. Any seller with a price book can publish today, and Sandbox is free.

  4. Same question, same answer

    deterministic

    Two agents asking for the same thing on the same terms get the same price. Prices don’t change based on who seems willing to pay more.

    Example: two protocols, one price
    Both send
    offerId gpu.a100-80gb · units 1000 · egressGb 2000
    Agent A · MCP · qr_7f3a…0c21
    $1,801.00
    Agent B · ACP · qr_b62e…91d4
    $1,801.00

  5. AI never sets the price

    published rules

    AI helps you structure your catalog draft. It never invents or negotiates a quote: published rules calculate every price, the same way every time.

    Example: how that total is built
    1,000 h × $1.701, 400+ tier
    $1,701.00
    Egress · 2,000 GB × $0.05
    $100.00
    Total
    $1,801.00
    Priced from
    published catalog v12

  6. You keep the final say

    approval

    We won’t ship anything that lets an agent close a deal you didn’t approve. Above your threshold or below your floor price, a person on your team approves, and agents never see where those lines sit. Each agent key can carry a daily limit, and a kill switch pauses every agent at once. How approval works

    What the agent gets back

    Needs approval

    reason
    Quote total exceeds autonomous authority.
    requiredApprovers
    ["deal-desk"]
    Kill switch on
    503 AGENT_ACCESS_PAUSED

  7. Evidence over hype

    versioned

    Price changes go live as a new published catalog version, and the audit log records who made them. Every quote carries the rule and catalog version that produced it. If a catalog has unresolved conflicts, its readiness score and badge say so.

    Example: on Agent A’s quote
    quoteRequest.id
    qr_7f3a…0c21
    catalogVersion
    12
    policyVersion
    4
    binding
    false
    Audit log
    catalog.version.publish

  8. A person can check what the agent was quoted

    signed

    Every quote we return is signed. The buyer, or anyone they pass it to, can confirm it came from us and hasn’t been changed, on our public endpoint or with our published key. How to check a quote

    POST /v1/public/quote-verify
    Agent A’s signed quote
    valid: true
    Same quote, one digit changed
    valid: false
    reason
    bad_signature

signature · ed25519 · valid: true, how to verify a signed quote

People

Anish Posim Reddy

Founder, PairRail

Anish has spent several years building production ML systems and AI tools inside revenue and GTM workflows, the teams that answer pricing questions today. Working alongside them, he kept seeing the same gap. Models were getting better at asking commercial questions, while sellers still answered with PDFs, rate cards and Slack threads built for human deal desks.

He started PairRail to close that gap. He’s building it into the way every seller answers buyer agents, with the care it takes to ship infrastructure operators trust.

LinkedIn (opens in new tab)

See how agents read your pricing today.

Score your pricing in minutes, free, or see how Atlas works from import to approved quote.

Score your pricing, free See the product