Downlink
YC W24San Francisco, US · Founded 2024 · 1 employees · 2 known investors
Downlink provides an API platform that optimizes large language model performance by increasing throughput, reducing latency, and improving accuracy.
Also known as Titan · downlink.dev
Founders & leadership· Y Combinator alumni (W24)
Downlink was founded in 2024 by TJ Murphy.
Investors · 2
Company profile
researched Aug 2026Downlink is a San Francisco-based developer platform that sits between an application and large language model providers to improve inference performance. Through a single API, the company says it expands rate limits, lowers response times, reduces costs and raises accuracy, describing itself as a platform that "continuously improves your AI" and marketing a headline claim of making LLMs three times faster and boosting performance by 20% or more.
The product is designed as a drop-in replacement for existing model clients: developers point an OpenAI-compatible client at Downlink's base URL (https://app.downlink.dev/api/v1) and supply a Downlink API key, keeping their existing code and model identifiers such as gpt-4o. Documentation shows examples in Python, TypeScript, Go and curl. Behind the API, Downlink selects models and fine-tunes them for the customer's use case, positioning the service as a way to avoid building and maintaining that optimization layer in-house.
The company was founded in 2024 by TJ Murphy, who describes his background as building internet-scale data infrastructure, and went through Y Combinator's Winter 2024 batch with Nicolas Dessaigne as primary partner. Y Combinator lists the company in the artificial intelligence, developer tools, SaaS and API categories with a team size of one.
Founding story
Founded in 2024 by TJ Murphy, whose stated focus is building internet-scale data infrastructure. The company joined Y Combinator's Winter 2024 batch, working with primary partner Nicolas Dessaigne.
Business model
Downlink provides a hosted API service that developers integrate in place of calling model providers directly, handling model selection, fine-tuning and performance optimization on their behalf. Access is gated behind a "Request Access" flow on the company website; pricing terms are not disclosed publicly.
Traction
Y Combinator lists the company as active with a team size of one as of its Winter 2024 batch listing; the product is offered on a request-access basis. No customer counts, revenue or usage figures are disclosed.
▸Full profile — market position, technology, go-to-market, geography, history
Market position
An early-stage, single-person company in the LLM infrastructure and developer tooling segment, positioned as an optimization layer above commercial model APIs rather than as a model provider.
Emphasis on requiring no architectural change — customers keep their existing client and model calls while Downlink handles model selection, fine-tuning, rate-limit expansion and latency management, with quantified claims of 3x faster LLM performance and 20%-or-more improvement.
Technology
An OpenAI-compatible API endpoint (app.downlink.dev/api/v1) that accepts standard chat-completion requests with existing client libraries and model names. The platform performs model selection and use-case-specific fine-tuning, and manages rate limits and latency, with the stated aim of improving throughput and accuracy without added architectural complexity. Integration is supported from any language, with documented examples in Python, TypeScript, Go and curl.
Go-to-market
Developer-led distribution: a self-describing website with copy-paste code samples and a minimal integration path (change the base URL and API key), combined with an access-request waitlist. The company's Y Combinator listing provides additional visibility to the developer and investor audience.
AI engineers and software teams building LLM-backed applications that face rate limits, latency constraints, inference costs or accuracy shortfalls, and that want to avoid adding fine-tuning and routing infrastructure to their own stack.
Geography
Headquartered in San Francisco, California, with a service delivered over the internet as a hosted API.
History
Founded in 2024 and part of Y Combinator's Winter 2024 cohort. The company operates from San Francisco and, as listed by Y Combinator, remains active with a one-person team offering its API on a request-access basis.
Compiled by commissioned research from 8 cited public sources — announcements, filings, and press listed under research sources below.
Key figures
latest reportedCompany-reported or press-reported figures, each dated to when it was claimed — not independently audited.
Timeline · 1
launches, deals, and filingsDownlink is listed as a Y Combinator Winter 2024 batch company, based in San Francisco, with Nicolas Dessaigne as primary partner.
Dated company events from announcements, filings, and press; legal rows summarize public dockets and regulator releases.
In the news
▸Research sources · 8
primary sources listed
- Downlinkdownlink.dev · web
8 public sources were cited for this profile; the first-party ones are listed here.
Frequently asked questions
- What does Downlink do?
- Downlink offers a single API that routes and fine-tunes LLM calls to raise throughput and accuracy while cutting latency.
- Who founded Downlink?
- Downlink was founded by TJ Murphy in 2024.
- Who are Downlink's investors?
- Downlink's investors include Y Combinator, Everywhere Ventures.
- Where is Downlink headquartered?
- Downlink is headquartered in San Francisco, US.