Table of Contents

TrustedRouter Investment Report

TrustedRouter Investment Report

Category: AI infrastructure, multi-model routing, confidential computing

Company Stage: Seed; public beta

Founder or Founders: Joseph Perla

Headquarters: Miami, Florida, United States

Funding: $1.25 million seed round

Business Model: Usage-based AI infrastructure; provider cost plus routing markup

Product Hunt Launch Date: September 1, 2026

Report Date: September 4, 2026

Investment Metric	Assessment
Venture Potential	75/100
Unicorn Path	Conditional
Valuation Attractiveness	Not Assessable
Evidence Confidence	63/100
Final Decision	DD

Executive Summary

TrustedRouter provides one OpenAI-compatible API for accessing more than 600 AI models from over 80 infrastructure providers. It selects or fails over between providers while offering zero-data-retention, regional, bring-your-own-key, and confidential-computing routes. Its central differentiation is a publicly inspectable, hardware-attested gateway designed not to log prompts or outputs (official website, security architecture).

The product serves developers and enterprises that use multiple models but do not want to maintain separate provider integrations or expose sensitive prompts to another conventional routing intermediary. Customers pay provider cost plus a 5.5% markup for text and embedding traffic and 20% for video generation, with no required subscription (pricing).

The strongest investment signals are founder quality, recent funding, and early production usage. TrustedRouter raised a $1.25 million seed round from investors including Sam Lessin of Slow Ventures, Bill Tai, Linda Avey, and George Xing. Founder Joseph Perla told Axios that the platform had processed more than one billion tokens in a single day while still in public beta (Axios via Yahoo Finance).

The central concern is that token volume does not necessarily create substantial revenue at a 5.5% markup. Depending on model mix, one billion tokens may represent anywhere from hundreds to tens of thousands of dollars of underlying model expenditure—and only 5.5% of that amount becomes TrustedRouter’s gross revenue before infrastructure, payments, support, and engineering costs. Revenue, repeat usage, customer count, retention, gross margin, and customer concentration are not disclosed.

TrustedRouter is sufficiently differentiated and technically credible to justify DD. It is not yet an “Invest” because valuation and round terms are unknown, the public evidence includes only one detailed customer case, and the business must prove that privacy differentiation can overcome OpenRouter’s scale and support durable, high-margin economics.

Product Overview

AI developers increasingly use different models for different workloads. Maintaining integrations, billing systems, availability monitoring, privacy controls, and provider-specific APIs creates operational complexity. Conventional routers simplify integration but introduce another intermediary that may be able to inspect or log prompts.

TrustedRouter offers:

One OpenAI-compatible endpoint for hundreds of models.
Automatic provider fallback through the trustedrouter/auto route.
Zero-data-retention routing.
Confidential-computing and end-to-end encrypted routes.
EU-focused routing for geographically sensitive workloads.
Bring-your-own-provider-key support.
Public model pricing and provider-status information.
Prepaid credits, cards, bank payments, PayPal, and supported stablecoins.
Public source code and attestations connecting deployed gateway software to a published image.

The ordinary inference path is designed not to store prompts or outputs. Billing records retain metadata such as model, provider, token count, cost, latency, status, and region. The optional batch service temporarily retains enclave-encrypted artifacts for up to 30 days (privacy policy).

Pricing is transparent: text and embeddings are charged at provider cost plus 5.5%, subject to a $0.01-per-million-token floor for very cheap routes. Video generation carries a 20% markup. The company also offers migration credits for customers already spending more than $100 monthly on model APIs (pricing).

The product replaces direct provider integrations, self-hosted routing software, conventional AI gateways, and manual provider failover. Migration can require only changing the API base URL, reducing adoption friction.

Product Quality: Technically differentiated and unusually transparent, but reliability, latency, and privacy guarantees require independent enterprise validation.

Founder and Team Assessment

Joseph Perla founded TrustedRouter in 2026. His LinkedIn profile reports previous technical and growth roles at Facebook and Lyft, including work on Lyft’s autonomous-vehicle program. It also identifies him as a co-founder of Terminal.com, which was acquired by Udacity, and Zandigo, which became Zinch and was acquired by Chegg (Joseph Perla).

Perla describes himself as a four-time exited founder. Refresh Miami repeats that characterization, but not every claimed exit was independently verified in public acquisition records reviewed for this report. His documented experience in infrastructure, machine learning, engineering management, and high-scale consumer platforms nevertheless indicates strong technical founder-market fit (Refresh Miami).

Public team evidence is limited. LinkedIn categorizes the company as having 2–10 employees but displays only one associated profile. The company said it would use the seed financing to expand systems engineering and backend development, implying that the current team remains small (FinSMEs).

Perla’s profile also lists a continuing role as president of Turntable Labs. His full-time allocation and operational responsibilities across the two companies require verification. At TrustedRouter’s current stage, substantial dependence on one technical founder creates key-person and execution risk.

Founder Assessment: Strong technical and scaling experience, but team depth and full-time founder commitment require verification.

Market Opportunity

The initial customer is a software company spending at least several thousand dollars annually on multiple LLM providers and handling data sufficiently sensitive to justify privacy controls. Likely verticals include legal technology, healthcare, financial services, enterprise software, cybersecurity, and AI-agent developers.

A bottom-up scenario illustrates the opportunity:

20,000–50,000 addressable AI development teams globally;
$50,000–$250,000 of annual model expenditure per customer;
5.5% router markup;
Implied annual TrustedRouter revenue per customer of approximately $2,750–$13,750.

At 20,000 customers and $5,500 average annual routing revenue, the business would generate approximately $110 million in annual revenue. These figures are analyst assumptions, not company results.

The market can support a venture-scale business, as demonstrated by Stripe’s agreement to acquire OpenRouter for more than $8 billion after OpenRouter’s reported $1.3 billion valuation earlier in 2026 (Axios, Yahoo Finance mirror). That transaction validates strategic interest in aggregated inference demand, but it does not establish that a smaller router can reproduce OpenRouter’s liquidity, distribution, or transaction volume.

Adjacent opportunities include enterprise governance, compliance reporting, intelligent model selection, committed-spend optimization, observability, private deployments, regional routing, API security, and a marketplace for model access.

Traction and Growth Signals

TrustedRouter entered public beta in May 2026 and launched publicly alongside its seed announcement at the end of August. Available traction evidence includes:

More than one billion tokens processed in a single day, according to the founder’s statement to Axios.
More than 600 models from over 81 providers, reported by Axios at launch.
A $1.25 million seed round from several identifiable investors.
One detailed production case study involving legal-AI company Robot, Robot & Human.
Public source code, SDKs, status monitoring, and attestation tooling.
A sixth-place Product Hunt finish on September 1, 2026 (Product Hunt leaderboard).

The customer case study reports billions of tokens during the first month, 170,974 legal documents and 616,682 pages processed across 34 matters, and production-scale deployment within three weeks. Corpus counts came from the customer’s database, while token claims were checked against TrustedRouter’s metadata, according to the company (case study). This is meaningful workload evidence but remains a single, company-published reference.

The public GitHub repository documents the architecture, billing controls, deployment configuration, trust boundary, SDK integration, and tests. Importantly, it explicitly states that the current deployment has not reached its aspirational target of one trillion tokens per day (GitHub).

Missing metrics include active customers, paying customers, monthly routed expenditure, net revenue, revenue growth, cohort retention, customer concentration, and gross margin.

Traction Assessment: Technically credible early production usage, but commercially unverified and potentially concentrated in one customer.

Competitive Position

Direct competitors include:

OpenRouter, a large multi-model marketplace and unified API.
LiteLLM, an open-source gateway that can be self-hosted.
Portkey, which combines an AI gateway with observability and governance.
Cloudflare AI Gateway, which provides caching, logging, analytics, and provider management.
Direct model-provider APIs and internal routing layers.

TrustedRouter’s clearest differentiation is the combination of public source code, a hardware-attested hosted gateway, fail-closed behavior, and no prompt or output logging. Self-hosted software provides control but places deployment and verification burdens on the customer. Conventional hosted routers may be easier to operate but require greater trust in the vendor’s logging policy.

The product has low integration friction because it supports the OpenAI API format. That helps acquisition but also lowers switching costs: customers can move away by changing the base URL again. There is no demonstrated network effect comparable to a mature marketplace’s provider liquidity and demand aggregation.

If OpenRouter, Cloudflare, or a major cloud platform offered an equivalent attested no-log gateway within six months, customers might stay for TrustedRouter’s neutral multi-cloud posture, open implementation, or privacy reputation. Those advantages are credible but not yet durable.

Defensibility Assessment: Medium

Business Model and Economics

TrustedRouter’s model is primarily transactional. At a 5.5% markup, $100,000 of customer model expenditure produces only $5,500 of platform revenue. This aligns price with usage and reduces procurement friction, but it creates a demanding scale requirement.

Potential gross-profit costs include:

Confidential-computing infrastructure.
Multi-region gateways and monitoring.
Billing, fraud, and payment-processing fees.
Provider failures and refund exposure.
Technical support and enterprise security reviews.
Migration credits.
Data-egress and observability infrastructure.

Because the underlying provider charge is passed through, gross margin should be measured on TrustedRouter’s markup revenue—not total billings. Reporting routed spend as revenue would materially overstate the business’s economics.

BYOK may lower working-capital exposure and preserve customers’ provider discounts, but public information does not clearly disclose whether BYOK carries the same markup or a separate fee. Enterprise subscriptions, minimum commitments, or governance fees could improve revenue predictability and margin.

Unicorn Path

An 8× net-revenue multiple is assumed for a high-growth AI infrastructure company with strong retention and healthy gross margin on its routing fees.

Required annual net revenue = $1 billion ÷ 8 = approximately $125 million.

At a 5.5% markup:

Required annual underlying routed spend = $125 million ÷ 5.5% = approximately $2.27 billion.

Illustrative customer combinations include:

22,700 customers spending $100,000 annually on models.
4,550 customers spending $500,000 annually.
2,270 enterprise customers spending $1 million annually.

These figures exclude revenue from higher-margin video routing or future enterprise subscriptions. They also assume TrustedRouter can count the markup—not total provider pass-through—as economically meaningful revenue.

Reaching this scale requires repeatable enterprise sales, multiple referenceable regulated customers, materially larger engineering capacity, multi-region reliability, commercial security certifications, and a product moat extending beyond confidential routing.

Unicorn Path: Conditional

Valuation Assessment

TrustedRouter raised $1.25 million from Sam Lessin of Slow Ventures, Bill Tai, Linda Avey, George Xing, and other investors (Axios via Yahoo Finance). The round’s valuation, instrument, investor ownership, board rights, and liquidation terms were not disclosed.

OpenRouter’s proposed acquisition is a relevant strategic comparable, but its greater scale and marketplace position make direct valuation comparison inappropriate. TrustedRouter’s revenue is also unknown.

Valuation Attractiveness: Not Assessable

Required information includes current net revenue, routed provider spend, growth, gross margin on markup revenue, customer concentration, retention, burn, runway, post-money valuation, SAFE or preferred-stock terms, option pool, and liquidation preferences.

Key Risks

Economically thin markup: Very high token volume may generate limited net revenue.
Customer concentration: Public production evidence centers on one legal-AI customer.
Scale disadvantage: OpenRouter and large cloud providers have broader distribution and provider relationships.
Privacy differentiation may compress: Competitors can introduce no-log and confidential-computing routes.
Small team and key-person risk: Public evidence identifies only one employee profile.
Founder commitment: Perla’s profile lists a concurrent leadership role at Turntable Labs.
Infrastructure reliability: Confidential gateways, billing, routing, and provider fallback create a complex operational surface.
Upstream-provider risk: End-to-end privacy ultimately depends partly on each selected provider’s policies.
Security credibility: One breach or mismatch between attested code and operations could damage the company’s core trust proposition.
Working-capital and fraud exposure: Prepaid credits reduce but do not eliminate payment, refund, and provider-settlement risks.

Final Assessment

Venture Potential: 75/100

Category	Score
Market Size and Expansion Potential	18/20
Traction and Growth Evidence	12/20
Founder and Team	13/15
Product Strength	9/10
Distribution Potential	10/15
Business Model and Economics	6/10
Defensibility	7/10
Total	75/100

The strongest elements are founder quality, market timing, technical differentiation, and low-friction integration. The weakest are thin transaction economics, limited customer evidence, small-team execution risk, and an unproven distribution engine.

Evidence Confidence: 63/100

Funding, investors, pricing, legal entity, product architecture, founder identity, public-beta timing, and one customer deployment are documented. Token volume and customer results are primarily company-reported but repeated by reputable press. Revenue, retention, customer count, gross margin, team size, burn, cap table, and valuation remain unavailable.

Final Decision: DD

TrustedRouter warrants formal diligence because it addresses a large infrastructure problem with credible technical differentiation, relevant founder experience, and early production usage. Investment should depend on verifying net revenue—not merely token volume—customer concentration, security architecture, founder commitment, and financing terms.

Upgrade Conditions

At least $1 million of annualized net routing revenue.
Five or more referenceable production customers, with no customer exceeding 25% of revenue.
More than 120% net revenue retention from production cohorts.
Positive contribution margin after confidential-compute and payment costs.
Independent security audit and enterprise compliance roadmap.
Demonstrated multi-region reliability and measured uptime.
Full-time founder commitment and a credible systems-engineering team.
Financing terms consistent with verified revenue and concentration risk.

Downgrade Conditions

Token growth without corresponding net-revenue growth.
Loss or contraction of the principal legal-AI customer.
Material privacy, attestation, billing, or security failure.
A major router matching the confidential-computing functionality.
Provider restrictions that impair routing coverage or margins.
Founder attention remaining divided across companies.
Material discrepancies in reported funding, usage, or customer results.

Questions for Further Diligence

What are current monthly net revenue and total routed provider spend?
How many active, paying production customers use the platform?
What percentage of tokens, routed spend, and revenue comes from Robot, Robot & Human?
What are 30-, 90-, and 180-day customer retention and net revenue retention?
How is BYOK usage priced, and what margin does it generate?
What is gross margin on the 5.5% markup after infrastructure and payment costs?
What are customer acquisition cost, sales cycle, and primary acquisition channels?
Have the gateway, attestation implementation, and billing system received independent security audits?
How much of Joseph Perla’s working time is committed to TrustedRouter versus Turntable Labs?
What is the current team structure, and which infrastructure roles remain unfilled?
What are monthly burn, cash balance, and runway following the seed round?
What are the post-money valuation, fully diluted cap table, investor rights, and liquidation preferences?