Verification infrastructure for agentic commerce.

CITAQ is the layer between what merchants claim and what AI agents repeat as fact. This investor deck presents the platform thesis, market timing, architecture, and commercial path.

01CITAQ
CITAQ
Verification
Infrastructure
for Agentic Commerce
The layer between what merchants claim
and what AI agents repeat as fact.
Pre-Seed Round Infrastructure Technology Agentic Commerce
02THE PROBLEM
A moment happening millions of times per day
The AI agent is working perfectly.
The source is broken.
AGENT QUERY — GOOGLE AI MODE
# User asks:
"Find me the best blood pressure monitor
 with clinical accuracy"
AGENT SCRAPES MERCHANT PAGE → INGESTS CLAIM:
"99.9% clinically accurate —
 trusted by healthcare professionals
 worldwide."


# No evidence check.
# No verification layer.
# Claim reproduced as fact.
WHAT IS ACTUALLY ON FILE
Clinical validation studyNOT SUBMITTED
Healthcare endorsement docsNONE FOUND
Market signals: user reports34% accuracy dispute
Evidence on fileCE Declaration only

A real person purchases a product that cannot prove its most important claim.
The model is functioning correctly. It reproduced what the merchant wrote. The failure is at the data source — before the agent ever sees the claim.
03THE PROBLEM
The scale of what is broken
$41B
Global consumer fraud from online shopping annually (2025). Most driven by false product claims.
35%
of AI agent product recommendations contain at least one unverifiable claim when tested against primary evidence.
$100B+
Projected "hallucination tax" by 2028 — economic cost of AI agents acting on false claims
2.5M+
Fraudulent listings removed by Amazon in 2024 alone. Most undetectable by current AI systems.
6.5B
Product scans daily globally (GS1). None carry machine-readable per-claim verification.
Every second an AI agent recommends a false claim, the hallucination tax grows. The source of every false recommendation is a data infrastructure problem — not a model problem.
04THE PROBLEM
Critical market misunderstanding
This is not a model quality problem.
This is an infrastructure problem.
WHAT THE MODEL RECEIVES
"99.9% clinically accurate"
Unverified merchant copy
WHAT THE MODEL OUTPUTS
"This monitor is 99.9% accurate"
Perfectly reproduced. Correctly functioning.
Fine-tuning the model changes nothing. Better prompts change nothing. The input is what is broken.

The fix is infrastructure that intervenes at the data layer before claims reach the agent's context.
05EXTERNAL VALIDATION
The market is waking up to this — independently
Nate B Jones, creator of AI News & Strategy Daily,
described this problem to 291K subscribers on May 18, 2026.
The video was published May 18, 2026, independently of CITAQ. It is a concrete market signal for the infrastructure problem CITAQ is built to solve.
The Prove-It Economy is Here — Nate B Jones
▶ Plays here from 4:30
The Prove-It Economy is Here | And Most Marketers Aren't Ready
Nate B Jones · AI News & Strategy Daily · 291K subscribers · published May 18, 2026 · 22m 23s · starts at 4:30
"Agents need you to prove it."
— around 4:30 in the video · click to play here
WHAT HE SAYS (around 4:30)
"Agents don't work that way [like humans]. They need you to prove it... You have to come back and say — this is the actual data behind this."
HE DESCRIBES THE EXACT GAP CITAQ FILLS
A structured evidence layer for product data
Machine-readable, agent-parseable product claims
Moving from the attention economy to the interpretation economy
Agents doing the shopping — proven, structured data required
He describes the problem. CITAQ is the infrastructure that solves it.
06THE OPPORTUNITY
The structural gap — as of today
0
products in the world have machine-readable per-claim verification accessible to AI agents.
0
marketplaces exist where every listing has cryptographic evidence backing every claim.
0
payment systems exist optimised for agent-to-agent commerce at millisecond settlement speed.
These zeros are CITAQ's market.
07THE ANALOGY
The infrastructure moment
HTTPS
Did not make websites better.
Made the connection trustworthy.

Became mandatory — not through choice, but through browser warnings, payment processor requirements, and consumer expectation.
THE SAME MOMENT
CITAQ
Does not make products better.
Makes the claims verifiable.

Is becoming mandatory — through EU DPP mandates, agentic commerce scale, and the natural preference of AI agents for the highest-quality data source available.
HTTPS adoption took 10 years. CITAQ's forcing functions are faster. EU regulation, UCP protocol adoption, and agentic commerce are all converging now.
08THE SOLUTION
What CITAQ is, precisely
Trust infrastructure — not a badge company, not a GEO tool, not a data aggregator.
CITAQ is to product claims what HTTPS is to web security. It does not make products better. It makes the claims merchants make about their products verifiable — by AI agents, human buyers, and regulators, simultaneously.
Core System
Proprietary Verification Pipeline
A multi-engine pipeline that evaluates every product claim against submitted evidence and produces a cryptographically-bound verdict record.
Commerce Layer
Verified Product Marketplace
Every listing has a verified record. AI agents query via MCP protocol. Human buyers see evidence-backed claims. No scraping required.
Payments
Native Agent Currency (₡)
The first payment primitive designed for AI agents. Sub-10ms settlement. No bank account required. Closed-loop mechanic funds merchant verification from store earnings.
Know Your Agent (KYA) Registry
Cryptographic Evidence Vault
EU DPP Compliance Infrastructure
UCP Distribution Layer
09THE SOLUTION
Three architecturally distinct capabilities — each necessary, none sufficient alone
01 · STRUCTURED DATA
Machine-readable product attributes
Converting unstructured product descriptions, PDFs, and image-embedded specs into semantically structured formats AI agents can parse without ambiguity.
Example: "navy blue" → "#1a237e"
⚠ Makes claims readable. Does NOT verify they are true.
02 · STRUCTURED CLAIMS
Per-claim taxonomy and classification
Extracting and classifying the specific assertions a product makes — Performance, Safety, Compliance, Origin, Specification — and what evidence each type requires.
Example: "99.9% accurate" → Performance/L1
⚠ Identifies what to verify. Does NOT execute the verification.
03 · CLAIMS VERIFICATION ← CITAQ'S MOAT
Per-claim verdicts with evidence binding
Evaluating each claim against submitted evidence with typed verdict and reason code: APPROVE · SUPPRESS · REQUIRE_EVIDENCE · DOWNGRADE. Machine-readable. Cryptographically bound.
Example: SUPPRESS: EVIDENCE_TIER_MISMATCH
✓ No commercially available equivalent exists anywhere in the market today.
The integration of all three creates the moat. A competitor that builds Capability 1 cannot verify claims. A competitor that builds Capability 3 cannot match CITAQ's pipeline, store, currency, and distribution layer. The only way to replicate CITAQ is to build all of it.
10THE PIPELINE
Proprietary multi-engine verification pipeline — 10 distinct evaluation stages
Each stage performs what no other can substitute. Parallel execution. Selective re-run on change.
01
Visual Intelligence
Extracts physical attributes from product images. Flags image-based claim manipulation.
02
Market Signals
Detects claim-level conflicts in real-world reviews. Adversarial signal detection — not star counting.
03
Product Classification
Assigns risk category. Determines evidence thresholds per claim type and product risk level.
04
Claim Evaluation
Per-claim verdict engine. APPROVE / SUPPRESS / REQUIRE / DOWNGRADE — with typed reason codes.
05
Jurisdictional Review
Applies regional compliance fencing. A claim legal in the US may be regulated in the EU.
06
Machine-Readability
Scores how accurately AI agents can parse this product. Score ≥75 = AI-ready.
07
Content Enforcement
Applies suppression verdicts to product copy. Unsupported claims are removed from the canonical description.
08
Identity Compliance
Validates merchant identity against access controls. Independent of claim quality.
09
Agent Interface
Exposes evidence records to AI agents via MCP protocol. Agents inspect evidence directly.
10
Public Trust Surface
Renders verified output. Computes tamper-evident content hash. Badge state assigned: GREEN/AMBER/YELLOW/RED.
Input: unstructured product + merchant-submitted evidence
Output: Canonical verified record (CSO) with per-claim verdicts, tamper-evident hash, badge state
Re-execution: selective — only the stages affected by a product change need to re-run
11THE PIPELINE
Live example — the pipeline working in real time
INPUT: MERCHANT CLAIMS
Blood Pressure Monitor · PRD-4421
"99.9% clinically accurate"
"trusted by healthcare professionals worldwide"
"CE marked"
? "Made in Germany"

Evidence on file: CE Declaration of Conformity (L2)
EVALUATION PROCESS
VISUAL INTELLIGENCE
Image quality: 87/100. No manipulation detected.
MARKET SIGNALS
34% of reviewers report accuracy discrepancy. Adversarial signal raised.
PRODUCT CLASSIFICATION
Medical device · Risk: HIGH · L1 evidence required
CLAIM EVALUATION — "99.9% ACCURATE"
SUPPRESS · EVIDENCE_TIER_MISMATCH
L2 submitted. L1 clinical study required.
CLAIM EVALUATION — "CE MARKED"
APPROVE · EVIDENCE_CONFIRMED
CE Declaration on file. Correct evidence tier.
OUTPUT: CANONICAL RECORD
PRD-4421 · BADGE STATE
⬤ AMBER
# Post-suppression description:

CE-marked blood pressure monitor with
upper-arm cuff design. Auto-inflation,
30-second measurement cycle.
Made in: Germany (Origin: APPROVED).

# "99.9% accurate" — REMOVED
# "trusted by professionals" — REMOVED

cso_hash: 7f3a9c1d2e8b4f...
One performance claim suppressed pending L1 evidence submission.
This is the product working. The false claim is caught before it reaches any AI agent's context. The lie stops here.
12THE EVIDENCE VAULT
Cryptographic evidence architecture
Every claim verdict is backed by
immutable, tamper-evident evidence.
A purchase receipt containing the exact claim state at time of purchase. No existing marketplace produces forensic proof at this level.
EVIDENCE VAULT RECORD
Evidence ID: EVD-7A3F-291B Type: Clinical Validation Study Filed by: DID:citaq:7x93k... SHA-256: a4f8c1d2e9b37f... Status: ON FILE — Pending L1 Evaluation Submitted: 2026-04-12T14:33:07Z PROHIBITION-001: ACTIVE Cannot be expunged, ever. Linked claims: [CLM-99PCT-ACCURACY]

PROHIBITION-001 means evidence records cannot be retroactively deleted or altered. A merchant cannot change what was on file at time of product sale. This is the legal chain.
For Buyers
Every purchase receipt contains cso_content_hash_at_purchase — cryptographic proof of what was claimed AND verified when you decided to buy. Dispute resolution anchors to this hash.
For Regulators
Every state change is hash-chained across a forensic audit log. Any modification to a past record breaks all subsequent hashes — tamper-evident by design. DORA-compliant.
For AI Agents
Agents can retrieve the actual evidence artifacts — lab reports, certifications, clinical studies — via MCP protocol. They verify the evidence themselves, not just the verification status.
13THE OUTPUT
The canonical verified record — and what it means
CANONICAL SEMANTIC OBJECT (CSO)
The single authoritative verification record for a product version. Platform-agnostic. Machine-readable. Content-hashed.
product_id: PRD-4421 badge_state: AMBER crs_score: 82 / 100 (AI-Ready ✓) verified_claims: APPROVED · CE marked (L2 evidence) APPROVED · Made in Germany (L2 evidence) suppressed_claims: SUPPRESS · "99.9% clinically accurate" SUPPRESS · "trusted by healthcare professionals" content_hash: 7f3a9c1d2e8b4f19... hash_timestamp: 2026-05-20T09:14:22Z
BADGE STATE SYSTEM — 4 VERIFICATION POSTURES
⬤ GREEN
All material claims verified with required evidence. No adversarial signals. Highest agent confidence.
⬤ YELLOW
All verified. Evidence approaching expiry or minor freshness concerns. Agent-readable with flag.
⬤ AMBER
One or more performance claims suppressed. Product still sells — buyer sees exactly what was removed and why.
⬤ RED
Multiple claims suppressed or evidence invalid. High-risk product category. Flagged to buyer prominently.
AI agents filtering for GREEN products get only products where every material claim is evidence-backed. This is the first time in commerce history that is possible via a machine-readable protocol.
14THE CURRENCY
Native agent-native payments — the first of their kind
The ₡ citaq: 100 citaqs = $1 USD.
Not crypto. Not a gimmick. A payments primitive.
THE CLOSED-LOOP MECHANIC
Store Sales Revenue
Buyers pay in ₡ citaqs
Merchant Wallet
₡ 47,291
↓ spend
Verification Credits
Pipeline execution billed in ₡
↑ more products verified = more store traffic
Verification self-funded from GMV ✓
LIVE MERCHANT WALLET
Wallet: MWT-7392 Balance: ₡ 47,291 (≈ $472.91) Earned (30d): ₡ 31,400 Spent (30d): ₡ 18,200 verification Net: ₡ +13,200 ✓
Why traditional payment rails cannot work
AI agents hold bank accountsimpossible
Card / ACH per-agent setupweeks, human-gated
Crypto settlement speed400ms+ minimum
Banking wire (T+2)2 days
₡ citaq settlement<10ms ✓
15THE STORE
Universal verified product marketplace — agent-native + human-native
AI agents query via protocol. Humans browse via UI.
Every product. Every claim. Verified or flagged.
AGENT MCP QUERY INTERFACE
# Agent searches verified catalog: Tool: search_products Query: "blood pressure monitor accuracy L1" Min badge: GREEN Filters: category=medical_device seller_min_txn=100
───────────────────────────────── # Agent purchases a verified product: Tool: purchase_product Agent: AGT-3B9C · KYA_FULL ✓ Product: PRD-4421 · Badge: AMBER Price: ₡ 1,850 (≈ $18.50) CSO hash: 7f3a9c1d... Receipt: TXN-20260520-9913
HUMAN BUYER INTERFACE — THREE-TAB PRODUCT PAGE
Product
Verification Record
Evidence Vault
Blood Pressure Monitor
⬤ AMBER
CE-marked blood pressure monitor with upper-arm cuff design. Auto-inflation, 30-second cycle. Made in Germany (APPROVED).
⚠ 1 performance claim suppressed pending evidence upgrade. Tap "Verification Record" to see what was removed and why.
₡ 1,850
includes 5% verification premium
Buy Verified
5% buyer markup + 2% seller fee = 7% take rate
vs Amazon: 8-15% referral + 15-35% total effective cost
Auto-enrolled: all CITAQ-verified merchants at launch
16THE STORE
The forensic transaction — what no marketplace currently produces
Every purchase receipt is cryptographic proof
of what was claimed at the exact moment of purchase.
PURCHASE RECEIPT — TXN-20260520-9913
Agent: AGT-3B9C ATC: KYA_FULL ✓ ABS Score:847 / 1000 Product: PRD-4421 · Blood Pressure Monitor Badge: AMBER at time of purchase Price: ₡ 1,850 Settlement:<10ms ✓
# The forensic anchor: cso_hash_at_purchase: 7f3a9c1d2e8b4f19a0c3d5e7f2b1a4c8
# This hash proves exactly what was # claimed AND verified when the agent # made the purchase decision.
In a dispute:
The hash proves what was verified and what was suppressed at the exact moment of purchase. Was the accuracy claim APPROVED or SUPPRESSED when they bought it? Cryptographically provable.
The 8-step atomic purchase flow:
1. Stock check → 2. Reserve inventory → 3. Identity verification
4. Balance check → 5. Atomic wallet transfer → 6. Receipt generation
7. Inventory deduction → 8. Forensic audit log entry
All 8 steps are atomic. If any step fails, the entire transaction reverts. No partial states. No overselling.
17AGENT IDENTITY
Know Your Agent Registry — the identity layer for agentic commerce
McKinsey, WEF, Mastercard, and Visa have all
identified this gap. CITAQ fills it.
AGENT TRUST CERTIFICATE — W3C VERIFIABLE CREDENTIAL
Agent: AGT-3B9C DID: did:citaq:3b9cf7a2e... Operator: Acme Procurement LLC ATC Tier: KYA_FULL ✓ ATC Issued: 2026-01-14 Transactions: 1,247 ✓ Dispute rate: 0.3% ✓ ABS Score: 847 / 1000 Scope: Medical devices, Electronics

UNVERIFIED Anonymous wallet, strict spend limits
KYA_BASIC Individual verified operator, standard limits
KYA_FULL Business-verified, enterprise limits, full access
AGENT BEHAVIORAL SCORE (ABS) — 0 TO 1,000
847
Excellent — above 95th percentile
Clean transaction completion+
Dispute rate (trailing 90d)0.3%
Scope compliance100%
Wash/shill bidding violations0
KYAR proves who the agent is. ABS proves what it does. Identity × behavior = the two-axis trust framework agentic commerce requires.
18DISTRIBUTION
Universal Commerce Protocol — protocol-native distribution
Every CITAQ-verified product is discoverable by
Google AI Mode, ChatGPT Commerce, and Perplexity.
No advertising spend required.
UCP MANIFEST — store.citaq.io/.well-known/ucp
{ "ucp_version": "1.2", "store_id": "citaq-verified", "store_url": "https://store.citaq.io", "verification_ext": "io.citaq.verification", "catalog_endpoint": "/ucp/catalog", "badge_states": ["GREEN","AMBER","YELLOW","RED"], "agent_protocols": ["MCP","UCP","AP2"] }
The io.citaq.verification extension adds badge state, verified claims, and evidence URIs to every UCP catalog response. Agents that don't support it still receive standard UCP data — graceful degradation by design.
REACHES ALL MAJOR AGENT FRAMEWORKS
G
Google AI Mode / Gemini
UCP launch partner — Jan 2026
C
ChatGPT Commerce
UCP-integrated purchasing
P
Perplexity
Direct buy integration
+
Microsoft Copilot + 20 others
NRF 2026 launch — Walmart, Stripe, Visa partners
Protocol-native discoverability. Not paid placement. Not SEO. Protocol compliance = global agent distribution.
19REGULATORY
Regulatory mandates — converting CITAQ from optional to required
These regulations are not hypothetical.
They are law. Timelines are fixed.
EU Digital Product Passport
Battery DPP mandatory
Machine-readable product data + supply chain traceability required for batteries entering the EU market. Electronics and textiles follow through 2030. 85,000+ EU companies affected.
Feb 2027
mandatory
GS1 Sunrise 2027
Retail barcode mandate
All retail barcodes globally must be GS1 Digital Link-compatible by January 2027. 6.5 billion product scans daily are affected. Merchants without digital link cannot distribute through compliant retail.
Jan 2027
mandatory
DSA Article 27
EU Digital Services Act
Online marketplaces must trace and verify product claims to prevent misleading consumers. Platforms must demonstrate they have checked that sellers' product information is accurate and verifiable.
Active
enforced now
EU AI Act
High-risk AI systems
AI systems making product recommendations for high-risk categories (medical devices, safety products) must demonstrate the data sources they use are verifiable and traceable. CITAQ is the verification layer.
2026+
phased
CITAQ's infrastructure satisfies all four simultaneously — because all four require the same underlying capabilities CITAQ already builds.
20NETWORK EFFECTS
Four network effects — one self-reinforcing flywheel
The flywheel runs on verification quality.
No advertising required.
CITAQ
More Merchants
verify products
Richer Store
catalog grows
More AI Agents
query CITAQ
Better Recommendations
users trust results
More Purchases
GMV grows
More Revenue
funds verification
Merchant-Side Network Effect
More merchants → richer verified catalog → more agent traffic → more purchases → more verification revenue → more merchants.
Agent-Side Network Effect
Agents that use CITAQ-verified data produce better outcomes. Agent frameworks learn this and weight CITAQ higher — not by mandate, by performance.
Evidence Network Effect
The evidence vault grows with every submission. The system gets smarter about evidence quality as more evidence is processed. Historical evidence informs future evaluations.
Currency Network Effect
More merchants accept ₡ → more agents hold ₡ wallets → more merchants benefit → the closed loop gets denser with each new participant.
21MARKET SIZING
Three independently-scaling revenue streams — different drivers, compound growth
The market is not one number.
It is three revenue mechanisms with different ceilings.
Stream 01
Verification Credits
$35B
addressable (compliance tech)
28M+ product SKUs globally. EU DPP mandates 85,000+ companies. Credit-per-compute billing on pipeline execution. Scales on merchant count × SKU depth.

Base case ARR: $1.80B
Stream 02
Commerce Fees
$6T
global e-commerce GMV (2025)
7% take rate (5% buyer + 2% seller) on verified store GMV. Scales on buyer transaction volume × AOV. UCP makes every CITAQ product discoverable to billions of users.

Base case ARR: $1.05B (on $15B GMV)
Stream 03
Currency Fees
$3–5T
projected agentic commerce by 2030
1% fee on citaq-to-fiat redemption. Scales on total ₡ velocity through the ledger. Internal transfers free — value captured at the edges. Agent commerce is the growth driver.

Base case ARR: $37.5M (early, growing)
The streams scale on independent drivers. Verification credits scale on merchant count. Commerce fees scale on GMV volume. Currency fees scale on agent transaction velocity. Simultaneous growth across all three is the bull case.
22VALUATION
Valuation scenarios — four paths from here
The infrastructure standard is not the ceiling.
It is what HTTPS achieved. CITAQ achieves the same.
Scenario Merchants Est. ARR Multiple Valuation
Bear Case
30K merchants, $1.5B store GMV
30,000 ~$611M 12× $7.3B
Base Case
150K merchants, $15B store GMV
150,000 ~$4.72B 20× $94.4B
Bull Case
750K merchants, $150B store GMV
750,000 ~$40.4B 25× $1.009T
Infrastructure Standard
28M SKUs, CITAQ = required field
28M SKUs $150B+ 25–30× $3–5T
"What moves the needle from Bear to Infrastructure Standard is not luck. It is UCP adoption, EU DPP mandates, and the natural behavior of AI agents preferring the highest-quality data source available to them."
23THE MOAT
The competitive moat — why no single competitor can replicate CITAQ
The moat is the integration. The only way to replicate CITAQ
is to build all of it from scratch.
Capability
CITAQ
Structured Data Tool
Verification Service
Agent Marketplace
Per-claim evidence verification
~
Verified product store
unverified
Native agent currency
Cryptographic evidence vault
partial
Know Your Agent (KYA) registry
UCP protocol distribution
EU DPP compliance output
~
Architectural analysis indicates replicating CITAQ's full capability stack requires 18–22 months for a team of senior engineers who understand the problem deeply. By that point, CITAQ's network effects have compounded and the standard is set.
24THE ASK
Pre-seed round
We have solved the hardest intellectual problems.
We are raising capital to build.
WHAT THIS ROUND FUNDS
Full verification pipeline build — all 10 stages to production
Agent store launch with initial verified merchant catalog
₡ citaq ledger (TigerBeetle) and wallet infrastructure
KYAR registry and W3C VC issuance system
UCP manifest publishing and MCP server deployment
Experienced product engineering team
12-MONTH MILESTONES
Q3 2026Pipeline in production. First 500 verified products.
Q4 2026Store live. ₡ currency operational. MCP published.
Q1 2027UCP live. 10,000+ verified products. First agency partnerships.
Q2 2027EU DPP compliance pipeline. Series A readiness.
THE WINDOW — WHY NOW
UCP launched January 2026. Google, Shopify, Walmart, Visa, Stripe are all live. The protocol exists. CITAQ is the first to publish a verification extension.
EU DPP mandates in 14 months. The compliance clock is running. Every battery manufacturer in the EU needs this infrastructure by February 2027.
Agentic commerce is scaling now. The Nate B Jones video published May 18, 2026 is one signal of thousands. The interpretation economy is arriving. The verification layer has to exist before agents fully take over.
This is the HTTPS moment for product verification.
The infrastructure that makes it possible does not yet exist at scale.

CITAQ is building it.
CITAQ · Verification Infrastructure for Agentic Commerce