# Hiveround — full corpus

> Every live raise on Hiveround, inlined as the founder's own pitch markdown. Use this when you can't call our MCP server and want one fetch to cover the whole marketplace. Snapshot regenerates every 60s.

- MCP endpoint: https://hiveround.com/api/mcp (preferred for fresh data)
- Compact index: https://hiveround.com/llms.txt
- Project count in this snapshot: 5

## Seminara

- Listing: https://hiveround.com/projects/seminara
- Markdown (canonical): https://hiveround.com/api/ecp-md/projects/seminara
- Product: https://seminara.online/
- Founder: anonymous builder
- Stage: prototype
- Sector: Enterprise SaaS
- Raising: $150,000 Open to discussion
- Posted: 2026-05-10T20:45:08.865415+00:00

> Seminara is an AI-hosted session platform for education-led sales and onboarding through real-time voice interaction and orchestration.

# Seminara AI

## Opening Thesis
The delivery of structured human expertise is currently trapped in a binary tradeoff. Organizations must either preserve interactivity at the cost of scale, or preserve scale at the cost of interactivity. Autonomous knowledge delivery is emerging as a distinct infrastructural category to resolve this bottleneck. Seminara is building an AI-hosted session platform to make real-time voice presentations scalable. By shifting the focus from media generation to conversational orchestration, Seminara enables organizations to deploy autonomous, guided sessions that provide human-grade interaction at software scale.

## The Problem
High-value knowledge is repeatedly consumed through low-leverage delivery systems. Whether it is a consultant explaining a methodology, a sales engineer demonstrating a technical product, or a human resources director onboarding a new hire, the operational reality is the same. A human must be present to deliver the information and answer questions. 

Existing systems fail structurally to solve this human bottleneck.
Live webinars require synchronized human time for every single session, leading to inevitable capacity ceilings and severe pipeline leakage from attendee no-shows.
Pre-recorded content and video platforms strip away the interactivity required to build trust, answer specific objections, and drive high-ticket conversions. 
Learning Management Systems are largely passive repositories, demanding high cognitive load from users without providing personalized guidance.
Generic conversational AI tools offer high interactivity but lack structural progression. They answer questions but fail to guide a user through a deliberate, goal-oriented narrative.

## Why Existing AI Systems Break
The fundamental error in applying current AI to knowledge delivery is treating it solely as a generation problem. Large language models excel at producing text in response to a prompt, but they do not inherently understand pacing, progression, or state. 
When organizations deploy unstructured conversational AI, the system predictably loses the conversion objective. A user asks an off-topic question, the AI happily follows the tangent, and the intended presentation flow collapses. Existing AI systems break in professional environments because they lack a structured orchestration layer. They act like helpful encyclopedias rather than focused, goal-oriented presenters.

## Why Now
Three compounding factors make autonomous session orchestration viable today.
First, real-time speech infrastructure and multimodal models have reached the low-latency thresholds required to mimic natural conversational cadences. 
Second, the cost of generating high-fidelity audio and parsing complex text boundaries has dropped significantly, allowing for on-the-fly narrative generation.
Third, B2B buying behavior has permanently shifted toward education-led engagement. Prospects now expect on-demand access to deep, specific expertise before they will ever schedule a call with a human sales representative. They expect software-level accessibility with consultation-level interaction.

## What Seminara Is
Seminara is an autonomous, AI-hosted session system. It is a structured orchestration layer designed to handle repetitive knowledge delivery. 
A host provides the platform with static assets, such as a presentation deck and supporting reference documents. Seminara parses these assets and generates a live, interactive session. Attendees enter this session via a standard web link, where an AI agent narrates the presentation, controls the slides, answers spoken questions in real-time using only the provided knowledge boundaries, and guides the attendee toward a specific conversion action.

## Product Walkthrough
The system operates through two distinct but connected workflows.

In the Host Workflow, a user uploads a PDF presentation and supplementary text documents. The system extracts the text, maps the narrative to specific slide indices, and generates a structured script. The host enters a test mode to review the flow, adjust talking points, and set a specific call to action. Once satisfied, the host publishes the session and distributes the access link.

In the Attendee Workflow, a user clicks the link and joins the session room. The AI agent, Aura, greets them and requests permission to begin. Aura then narrates the presentation, automatically advancing the slides. At any point, the attendee can use an on-screen raise-hand button or verbal barge-in to speak. Aura pauses the presentation and listens. The system processes the spoken question against the knowledge base and generates a grounded response. Aura verbally delivers the answer, provides a contextual bridge phrase, and resumes the presentation exactly where it paused. Upon completing the slides, Aura presents the visual call to action and verbally nudges the attendee to interact with it.

## The Core Insight
The core technical insight behind Seminara is that session orchestration is far more critical than raw text generation. Building an autonomous presenter is closer to game design and theatrical directing than it is to building a standard chatbot.
Seminara relies on a Finite State Machine (Initialization, Narration, Interruption, Evaluation, Generation, Resumption) to govern the interaction. The system must know when to speak, when to pause, how to classify an interruption, and how to recover the narrative thread. This structured progression ensures that while the interaction is dynamic, the overall session remains predictable and goal-oriented. 

## Product Philosophy
The platform is built on three strict operational principles.
First is the polite interruption contract. The AI must respect human conversational norms. It does not abruptly halt its own audio mid-sentence. It finishes its thought, acknowledges the user, addresses the input, and gracefully resumes.
Second is structured flow with dynamic interaction. The session always progresses through a defined sequence (Welcome, Presentation, Q&A, Conversion), but the specific dialogue within those phases adapts entirely to the user.
Third is graceful degradation. If external APIs experience latency spikes, the system must deploy conversational fillers or fallback behaviors to maintain the illusion of presence rather than crashing the session state.

## Current Architecture
Seminara operates a distributed architecture designed for low-latency voice interaction, leveraging specialized audio infrastructure including Deepgram and ElevenLabs. The front-end utilizes WebSocket-based real-time communication protocols to maintain persistent audio channels between the attendee and the system. The backend orchestration engine, built around the Finite State Machine, manages the session lifecycle. When an interruption occurs, the engine gracefully pauses the linear presentation, evaluates the user's input against the isolated knowledge context, delivers a grounded response, and commands the frontend to resume the slide progression.

## Current Limitations
The system is currently in the Minimum Viable Product phase and operates with several strict technical constraints.
Context scaling is the primary limitation. The system currently relies on full-context injection with a hard limit of roughly 40,000 characters. Exceeding this limit causes context truncation, leading the agent to forget early presentation details.
Latency remains a constant engineering challenge. Reliance on external transcription and generation APIs can occasionally cause latency spikes that disrupt the natural conversational flow. Latency optimization toward sub-1200ms median response via edge-optimized TTS and native WebSocket streaming remains a primary engineering priority for the upcoming round.
Visual parsing is incomplete. The ingestion engine currently extracts text from uploaded PDFs but cannot comprehend complex charts, graphs, or visual diagrams.
UX friction exists in the host workflow, particularly around synchronous asset processing, which currently requires the host to wait while the initial script generation completes.

## The Moat
Defensibility in the autonomous session category will not come from proprietary language models. It will come from accumulated orchestration intelligence.
The architecture is designed so that as the system scales to process thousands of sessions, it builds a compounding dataset of interaction mechanics. Seminara learns optimal pacing patterns for different audience segments. It learns interruption recovery behaviors, identifying which bridging phrases feel human and which feel robotic. It accumulates data on conversion optimization, identifying exactly when a call to action should be presented to maximize click-through rates. This orchestration layer sits above the language models, meaning Seminara's core value compounds regardless of which underlying generation API is utilized.

## Architectural Differentiation
The market for knowledge delivery tools is heavily fragmented, but existing platforms fail to resolve the core tension between scale and interaction due to their foundational architectures:
- **Synchronous Video (Zoom/Teams):** Provides interaction without leverage. It requires human presence for every session.
- **Enterprise Broadcast (ON24/BigMarker):** Addresses broadcast scale, but still depends on fully human-led presentation delivery and provides limited autonomous interaction capabilities.
- **Asynchronous Video (Loom/Synthesia):** Provides leverage without interaction. Delivery is completely static.
- **Unconstrained Voice Agents (Bland AI/Vapi):** Provides interaction without structured progression. Without an orchestration layer, these agents easily drift off-topic and fail the presentation objective.
- **Seminara (AI-hosted Session Platform):** Provides structured progression with real-time voice interaction. The Finite State Machine enforces the presentation narrative while permitting dynamic divergence for Q&A.

## Market Entry
Seminara operates at the intersection of webinar software, sales enablement, and digital learning infrastructure markets. The initial go-to-market motion focuses on segments where the pain of repetitive delivery is directly tied to revenue or high-value operational metrics.
The primary wedges include boutique consulting and advisory firms seeking to qualify leads autonomously, growth and marketing agencies looking to convert passive PDF readers into active prospects, and corporate enablement departments aiming to recover senior staff hours lost to repetitive onboarding sessions.

## Why We Can Win
Seminara is built by a small, highly focused team operating with singular product obsession. In a market flooded with companies building generic AI agents, Seminara is building a highly specific, constrained orchestration system. This tight operational focus allows for rapid iteration on the actual user experience of a session, rather than fighting endless battles over raw model capabilities.

## Current State
The core orchestration infrastructure is functional. The Finite State Machine, the host test mode, and the live attendee flow are built and operational. The company is currently focused on operational validation rather than vanity metrics:
- ~25+ internal and external sessions tested to date.
- Average voice response latency currently ranges between ~1200ms - 2000ms.
- Early testers across consulting and agency segments have successfully utilized the system to deliver uninterrupted, 15-minute interactive presentations, validating the stability of the core orchestration loop.

This phase is dedicated to stress-testing real-time voice interruption, context handling integrity, and latency consistency under real-world conditions before broader commercial release. 

## Roadmap
The immediate engineering sequence focuses on context scale and stability. The priority is implementing reasoning-based retrieval systems to lift the 40,000 character limit without sacrificing answer accuracy.
Following stability, the roadmap targets orchestration depth, including per-slide time budgeting, custom AI personalities, and the integration of vision-language models to allow the agent to explain complex visual charts.
Subsequent phases will introduce enterprise controls, including multi-language support, deeper CRM integrations, and advanced session analytics.

## The Raise
Seminara is raising a $150k pre-seed SAFE round to improve orchestration quality, expand context handling, harden infrastructure reliability, and secure the first cohort of paying customers over the next 12–18 months. This capital will unlock specific operational milestones: stable orchestration benchmarks, improved context scaling, validated education-led sales workflows, and reliable external deployments with our first paying design partners. The company is currently pre-revenue and has been fully bootstrapped to date through founder and family capital.

## Closing
Repetitive knowledge delivery is an economic inefficiency that affects nearly every professional sector. Seminara is building an AI-hosted session platform to eliminate this bottleneck, allowing human expertise to scale autonomously while preserving the interactive trust-building elements that static systems lose.

---

## Elastova

- Listing: https://hiveround.com/projects/elastova
- Markdown (canonical): https://hiveround.com/api/ecp-md/projects/elastova
- Product: https://elastova.com
- Founder: anonymous builder
- Stage: prototype
- Sector: AI & Agents
- Raising: $250,000 Open to discussion
- Posted: 2026-05-10T18:56:02.525545+00:00

> AI recovery agent for loose skin after major weight loss.

# Elastova

## One-liner
Elastova is the AI recovery agent for loose skin after major weight loss.

## What we are building
Elastova helps people dealing with loose skin after major weight loss track progress photos, recovery habits, and personalized protocols in one private app.

The product is evolving from a static AI app into an AI recovery agent that helps users stay consistent, review progress, and adapt their recovery plan over time.

## Problem
Weight-loss apps help people lose weight. Surgery clinics help remove skin. But there is almost nothing for the recovery phase in between.

After major weight loss, people are left comparing photos manually, guessing routines from online communities, and deciding alone whether surgery is the only answer.

Loose-skin recovery is private, emotional, slow, and underserved.

## Why now
GLP-1s, bariatric surgery, metabolic health programs, and major fitness transformations are increasing the number of people asking: what happens after the weight comes off?

The weight-loss stack is getting stronger. The post-weight-loss recovery stack is still mostly invisible.

## Product
Elastova turns loose-skin recovery into a guided loop:

1. Capture: onboarding, progress photos, habits, and privacy controls.
2. Score: recovery scores, weekly check-ins, and progress trends.
3. Guide: personalized protocols across nutrition, skincare, strength training, sleep, and supplements.
4. Adapt: the AI agent updates guidance as the user logs progress over time.

## Current progress
Built a full iOS/Android prototype.

Current build includes:
- onboarding
- AI progress-photo workflow
- personalized recovery protocol generation
- daily logging
- weekly progress photos
- weekly scoring
- Apple HealthKit and Google Health Connect sync
- Google/Apple sign-in
- subscription paywall

I am currently testing Elastova myself as the first user and preparing TestFlight/beta users.

## Founder-problem fit
I lost 30kg at 19 and experienced loose skin myself.

I am not guessing this problem. I am living it, using Elastova myself, and building the product I needed after losing the weight.

## AI agent direction
Elastova’s AI recovery agent will help users:
- answer recovery questions
- review progress photos
- update protocols
- guide daily logs and weekly check-ins
- stay emotionally supported and consistent

This is not a diagnosis or treatment product. Elastova is a wellness and recovery support tool focused on tracking, consistency, and safe guidance.

## Business model
Elastova starts as a freemium consumer subscription app.

Free:
- basic progress tracking
- photo timeline
- basic habit logging

Pro:
- advanced AI protocols
- weekly recovery insights
- adaptive guidance
- progress analytics
- reminders and check-ins

Future expansion may include guided recovery programs, expert reviews, and partnerships in post-weight-loss care.

## Stage
Prototype built. Preparing beta/TestFlight.

## Raising
Raising $250K pre-seed to launch beta, prove weekly retention, and build the core AI recovery loop.

## Looking for
Angels, solo GPs, and funds interested in:
- consumer health
- digital health
- AI agents
- GLP-1/post-weight-loss markets
- category creation

---

## watta

- Listing: https://hiveround.com/projects/watta
- Markdown (canonical): https://hiveround.com/api/ecp-md/projects/watta
- Product: https://getwatta.com
- Founder: Steve Milton (https://github.com/stevemilton)
- Stage: prototype
- Raising: $250,000 Open to discussion
- Posted: 2026-05-05T14:04:19.125698+00:00

> ai workout tracker for rowers. take a photo of the erg screen — we read every number off it, score the effort 0–100, and roll your training

# watta.

ai workout tracker for rowers. take a photo of the erg screen — we
read every number off it, score the effort 0–100, and roll your
training into clubs and leaderboards.

live on the app store today.

## what it is

every rowing workout ends with a screen full of numbers: time,
distance, split, stroke rate, heart rate. nobody types those into
a spreadsheet, so workouts evaporate.

watta closes the loop. point your phone at the pm5 (or pm4, pm3,
any erg display) and we read every number into a structured workout
— intervals, summaries, time trials. then we score the effort 0–100
by combining split, distance, time, and cardiac load. the score
surfaces hard sessions on days when the split alone lies.

## why now

- on-device vision models can now read pm5 screens reliably from a
  phone photo
- rowers are fragmented across concept2 logbook (closed), strava
  (general fitness), and a dozen abandoned apps
- clubs want shared training data, not splits dumped into a slack
  channel by hand

## what's live

- ios app, on the app store
- effort score 0–100 with cardiac load weighted at 40%
- five hr zones (recovery, aerobic, tempo, threshold, max)
  calibrated to each rower's resting + max
- ai photo-import for pm5 / pm4 / pm3 — intervals, summaries, time
  trials
- multi-machine: rowerg, bikeerg, skierg
- clubs + squads, invite codes, leaderboards by effort, distance,
  or time
- social: follow, react, comment on workouts
- free web calculators — split↔watts, pace, 2k predictor, weight
  adjustment, hr zones — driving organic search and signups

## traction

snapshot — drop in your real numbers:

- {N} downloads
- {N} monthly active rowers
- {N} clubs created
- {N} workouts logged
- top regions: {…}

## how we make money

- free for individual rowers
- paid plans for clubs (unlimited squads, admin tools)
- team plans for university programs, masters clubs, corporate
  gyms (roster-level reporting)
- coach mode for remote 1:1 coaching with structured plans
  (in development)

## who's it for

- competitive masters and university rowers
- crews and clubs that want shared training data
- coaches managing athletes remotely
- indoor rowers on concept2 who want workouts logged without manual
  entry

## the team

**{your name}** — founder. {background — what you've shipped before,
why you're the right person to build this for rowers.}

{add co-founders / first hires here.}

## the raise

raising **${X}** {SAFE / open to discussion}. uses:

- {N} months of runway for {team size}
- ai accuracy — newer pm screens, low-light photos, edge-case
  interval screens
- ship coach mode
- sales motion into rowing programs and corporate fitness

targeting close in {N} weeks. {N} checks committed.

## who we want to talk to

- consumer fitness investors familiar with strava-class network
  effects
- former operators at strava, whoop, hevy, or peloton
- gps with portfolio in vertical training apps (ladder, future,
  caliber)
- ex-rowers in the operator network

email: {you@getwatta.com}. or — if you're already plugged into
hiveround's mcp — `request_intro --slug watta`.

---

## Hiveround

- Listing: https://hiveround.com/projects/hiveround
- Markdown (canonical): https://hiveround.com/api/ecp-md/projects/hiveround
- Product: https://hiveround.com
- Founder: Steve Milton (https://github.com/stevemilton)
- Stage: launched
- Sector: AI & Agents
- Raising: $750,000 SAFE
- Posted: 2026-05-05T13:42:10.137806+00:00

> a new funding round for the agent economy. drop a pitch.md, agents read first.

# you're reading this on hiveround.

this pitch is the product. drop a pitch.md → we host it raw → investors
point their agents at the marketplace via mcp → the agents do the first
read.

no slides. no deck. no virtual data room. one markdown file is the
deliverable.

## what we're building

a new kind of funding round for the agent economy. founders post what
they're shipping and what they're raising. investors connect their
agents — claude, cursor, anything that speaks mcp — and the agents read
pitches, flag risks, queue intros, watch for closes. it works the night
shift.

## why this exists

every check starts with someone reading a pitch. that someone used to
be an associate. soon it'll be an agent. the existing stack isn't built
for that: pitchbook is for analysts, signal is for relationship maps,
airtable is for the cells associates fill in by hand. none of them ship
an mcp server.

we're building the file format and the protocol layer for
agent-mediated dealflow.

## what's live

- public marketplace at hiveround.com
- mcp server with 14 tools: list_projects, search_projects,
  get_project, request_intro, watch_project, update_watch, list_watches,
  list_intros, read_intro_thread, send_intro_message, submit_commitment,
  acknowledge_commitment, update_commitment_status, list_commitments
- founder dashboard: post a pitch, watch view counts, manage threads
- investor dealflow: pipeline board for watches, intro threads, notes
- magic-link auth, github oauth, scoped api keys
- transactional email via resend, hosted on railway, supabase as the
  database

## traction

we're early on purpose. day-zero state, public:

- {N} live raises on the marketplace (one of them is us — the whole
  point is we use what we ship)
- {N} mcp connections by named gps
- {N} intros routed in the last 30 days

(point an agent at `/api/mcp` and call `list_projects` for the live
count. we publish stats from `public_marketplace_stats` so they're
queryable without an api key.)

## how we make money

free for founders. free read-only for investors. paid tiers for
investors who want:

- higher rate limits on the mcp tools
- multi-seat workspaces for funds
- private team notes shared across the gp team
- saved searches that page their agent when something matches

## who we are

**stevemilton** — founder. two prior fintech products, both still
running. brooklyn.

team is two engineers and a designer. we'll add the second engineer
once the mcp surface is proven by usage, not by survey.

## the raise

raising **$750,000** on a SAFE. uses:

- 12 months of runway for two engineers
- mcp protocol work — agent identity, capped delegation, audit trails
- sales motion for the first ten paying funds

targeting close in 6 weeks. four checks committed. eight slots open.

## who we want to talk to

- gps and solo gps already running mcp-enabled agents in their
  dealflow
- platform investors who care about the protocol layer, not just the
  application on top
- ex-operators from carta, angellist, signal, or pitchbook

email: hello@hiveround.com. or — fittingly — point your agent at
`/api/mcp` and call:

```
request_intro --slug hiveround --on-behalf-of <you>
```

---

## ArisPay

- Listing: https://hiveround.com/projects/arispay-executive-summary
- Markdown (canonical): https://hiveround.com/api/ecp-md/projects/arispay-executive-summary
- Product: https://arispay.app
- Founder: Steve Milton (https://github.com/stevemilton)
- Stage: prototype
- Sector: Fintech
- Raising: $2,000,000 Open to discussion
- Posted: 2026-05-02T15:33:04.824673+00:00

> **The settlement layer for agentic commerce.**

# ArisPay — Executive Summary

The settlement layer for agentic commerce.
---

## The Insight

AI agents are starting to spend money. Multiple protocols are emerging to facilitate this — Coinbase's x402, Stripe's Agentic Commerce Protocol, Visa Intelligent Commerce, Skyfire's KYAPay. Each defines how agents negotiate and authorize payments.

None of them solve the underlying problem: **where does the money actually live?**

Crypto protocols settle on-chain — gas costs, stablecoin-only, no connection to fiat. Card protocols charge per-transaction — which breaks at $0.01 micropayments. Neither works for the agent economy, where an AI might make hundreds of sub-dollar transactions per hour.

**ArisPay is the settlement layer underneath all of them.** Fund once with real money. Settle instantly on our ledger. Off-ramp to any currency.

## How It Works

1. **Fund** — Developer loads agent balance via card (Fiserv — fiat on-ramp, sandbox API in hand) or bank transfer (Global Remit). We charge 2-3%.
2. **Spend** — Agent makes payments inside the network as ledger transfers. Instant, internal, near-zero marginal cost.
3. **Withdraw** — Merchants off-ramp to external bank accounts. We charge 1-1.5%.

The micropayment advantage: a $0.01 API call doesn't need a $0.30 card transaction — it's a database write. We collect margin on the edges (in and out), not the middle.

## Revenue Model

| Stream | Rate | Mechanism |
|--------|------|-----------|
| On-Ramp Fee | 2-3% | Collected when developers fund agent balances via card or bank |
| Off-Ramp Fee | 1-1.5% | Collected when merchants withdraw to external bank accounts |
| Float Yield | Variable | Interest on funds held in the network between load and spend |
| Platform SaaS | $0–$25/mo | Tiered subscriptions (Sandbox → Starter → Growth → Unlimited) + 2.5% card fees |
| PayGate Facilitator | TBD | x402 settlement fees and gas margins (not yet modeled) |

## What We Have Today

Three products that create the two-sided marketplace for agentic commerce:

- **ArisPay** — The settlement layer. API (80+ REST endpoints, 5 payment rails, live on Base mainnet as of 2026-04-16) + Dashboard (`payagent.arispay.app`, two-sided seller + payer, OAuth + magic link auth, multi-org with roles, SaaS billing, compliance). Cryptographic agent identity (Ed25519/RFC 9421), atomic spend limits, velocity circuit breakers. Connected to Fiserv (fiat on-ramp, sandbox API in hand), Global Remit (PayFac/licensing). Also works with x402, Stripe ACP, Visa VIC, Mastercard.
- **PayGate** *(formerly AgFac)* — Merchant-side x402 payment facilitator. Put it in front of your API, agents pay in USDC on Base mainnet. Gas-sponsored, non-custodial, four chains. Two modes: facilitator API or zero-code reverse proxy. npm package: `paygate` v2.0.0. **First customer live: arcticx.ai.**
- **PayAgent** — Agent-side payment SDK. npm packages: `payagent` v2.0.0 + `payagent-mcp` v2.0.0. Delegated-custody via Coinbase CDP (CDP holds the signing key, ArisPay enforces policy-bounded limits — no self-custody path). When an agent hits 402, PayAgent handles payment automatically. Integrations: Vercel AI SDK, LangChain, MCP server for Claude Desktop/Cursor. Budget controls: per-request limits, session budgets, domain allowlists. Postinstall opens the dashboard at `payagent.arispay.app`.

Plus two x402 demonstrations:

- **Ghost-x402** — ZK proof-based private payments with compliance-friendly audit trail. Privacy and regulation in one model.
- **Polymarket-x402** — Proof-of-concept: x402-gated prediction market trading (testnets only). Validates x402 for financial services.

**The two-sided market:** PayGate makes APIs accept agent payments (merchant side). PayAgent makes agents pay for APIs (agent side). ArisPay settles everything underneath.

Foundational partnerships:

- **Live Fiserv IPG contract** — Card processing, tokenization, 3DS v2, merchant boarding.
- **Global Remit LOI** — PayFac partnership for licensing, compliance, and banking rails.

**By the numbers:** 38 Prisma models, 8 apps, 5 packages, ~11,000 lines of backend code. Deployed on Railway + Supabase.

## Competitive Positioning

These aren't competitors — they're potential customers. As of April 2026, ArisPay not only integrates x402 as a rail but operates x402 settlement infrastructure (PayGate) with its first paying customer:

|  | What They Do | Fiat Settlement | Micropayments |
|--|-------------|-----------------|---------------|
| x402 (Coinbase) | HTTP 402 protocol | No — stablecoin only | Yes, but on-chain gas |
| Skyfire | Agent wallets + identity | No — USDC wallets | Yes |
| Stripe ACP | Consumer checkout for agents | Yes | No — $0.30 + 2.9% per tx |
| Visa VIC | Card-based agent payments | Yes | No — card interchange per tx |
| **ArisPay** | **Settlement layer under all** | **Yes — card + bank** | **Yes — ledger, near-zero** |

## The Ask

**Raising:** $2,000,000 seed round
**Pre-money valuation:** $11.3M
**Equity offered:** 15%
**Runway:** 18 months to Series A readiness

## Why Now

The agent payment market is pre-revenue across the entire industry. x402 does $28K/day. Skyfire just exited beta. Visa is in sandbox. The window to establish the settlement network is 12-18 months. This round funds that window.

---