The build directory opens once real customer apps are running through the network. No placeholder logos, no fake testimonials. Submissions are rolling — featured entries get published with model mix, routing strategy, and what they would do differently.
Sample preview
Conversational, devtools, docs, localization, content, analytics
Curated; not pay-to-list
Reviewed weekly during launch ramp
Opens once we have real builds to feature
Coming soon. The directory opens once we have builds to feature with real production traffic on the network — currently 20+ models on Google Vertex AI, with Google (Gemini, Imagen, Veo) — Anthropic, OpenAI, and AWS Bedrock shipping next.
The directory is editorial, not a marketplace. We will feature work in these six categories at launch — picked because they cover the dominant workloads we see teams routing through NemoRouter.
Chat assistants, support bots, internal-tool copilots — anything where the LLM is the primary surface a user talks to. We plan to feature builds that ship custom personas, retrieval, and multi-turn evaluations.
Code completion, refactor assistants, CLI helpers, IDE extensions, and CI agents. The category where latency and tool-use correctness matter most — and where routing strategy is the differentiator.
PDF extraction, contract analysis, knowledge-base ingestion, OCR-plus-LLM pipelines. Long-context, batch-friendly, and a strong fit for the cost-tier rankings.
Translation services, glossary-aware adapters, and locale-specific evaluation. Highest-throughput category we expect to feature — the one where one-key, multi-provider really earns its keep.
Marketing copy, blog generation, social, product descriptions, and editorial workflows. Tighter prompt templates and tone-matching are the wedge here — A/B tests are typically what teams ship next.
Turning unstructured text into structured JSON — entity extraction, relationship graphs, schema-validated outputs. Latency-tolerant, accuracy-critical, observability-heavy.
The directory only matters if every entry is worth reading. Here is the rubric editorial uses for every submission.
We feature builds that customers actually use, not landing pages. The simplest signal is consistent token volume on NemoRouter — pinned to the org for at least 30 days.
A one-paragraph description of what your build does and who it serves. Editorial-quality, not a sales pitch. We will ghost-write polish, not the substance.
Model mix, routing strategy, guardrails, and retrieval pattern. The directory exists to help other teams learn — vague entries get returned for revision.
Email us with three things — what you built, why it matters, and a contact for editorial follow-up. Decisions land within two weeks during the launch ramp.
Send the email
Email support@nemorouter.ai with subject line Submit a build.
Share the build
A link to the live product, your category, and a one-paragraph description of who uses it and why. Architecture notes optional but appreciated.
Editorial follow-up
We reply within two weeks during the launch ramp. Featured builds get a draft to review before publication; declined submissions get a one-paragraph reason.
Launch list · ~weekly drops
One email when the first batch publishes. Then nothing until something interesting happens.
Already shipping on NemoRouter? Tell us at support@nemorouter.ai.