Bring startup competitor research into your own AI agents.
The internal FSS CLI exposes the research stack behind Find Similar Startups: source-aware discovery, competitor evaluation, trend signals, crowdedness checks, and report generation. We are opening it as an AI market research API and CLI for teams building agentic research systems.
Debug loop
$ fss research list --pretty
$ fss discovery list <threadId> --pretty
$ fss tool serper "site:producthunt.com AI sales agent"
$ fss tool batch-evaluate --candidates candidates.json
$ fss report rerun <threadId>Why expose the API and CLI?
AI research agents need evidence loops, not screenshots of final reports.
The CLI lets an agent reproduce a source query, inspect raw candidates, score them against a startup, enrich the winners, and rerun the final artifact without guessing where quality changed.
Create custom reports
Run discovery, enrich selected companies, then regenerate founder-readable reports from your own agent workflow.
Search the right source first
Route by startup type across Product Hunt, YC, Crunchbase, BetaList, G2, app stores, AI directories, and smaller launch surfaces.
Evaluate competitor fit
Use production-faithful batch scoring to separate direct threats, adjacent players, false positives, and weak matches.
Read market movement
Pull Google Trends, autocomplete, forums, YouTube, ads transparency, and AI Mode evidence into your market read.
Measure crowdedness
Compare candidates, source density, duplicate rate, false positives, and signal quality before your agent commits to a claim.
Commands that already exist internally
One interface across search, scoring, traces, startups, and reports.
Inspect sessions and stage status.
Read raw candidates, annotate, rescore.
Run search, crawl, trends, and scoring tools.
Invoke FSS sub-agents directly.
Inspect enriched competitors by thread.
Rerun reports from existing evidence.
Trace prompts, tool calls, cost, latency.
Create benchmark profiles from the CLI.
Workflow
From raw source to measured report.
Find the thread or create a benchmark profile.
Search startup-native sources and raw web surfaces.
Score candidates with the same evaluator the FSS agent uses.
Enrich the winners with source-backed startup details.
Generate or rerun a report from the measured evidence.
Example commands
Scriptable pieces your agent can call.
The public beta will focus on stable primitives first: source search, similarity evaluation, trends, crowdedness signals, and report generation from saved evidence.
fss tool trends "Cursor,Replit,Lovable" --data-type TIMESERIES --date "today 12-m" --geo US --prettyfss agent run directory-search --platform producthunt --user-startup-description "AI support agent for Shopify stores"fss tool batch-evaluate --candidates '[{"name":"TestCo","description":"AI support automation"}]' --user-description "AI support agent for Shopify"fss discovery rescore <threadId> --prettyfss report rerun <threadId>Join the waitlist for CLI access.
Tell us where the CLI would sit in your agent flow. We will use that to prioritize the first beta paths: custom reports, source routing, evaluation, trends, and crowdedness analysis.
FAQ
What the beta is for.
Is the FSS CLI public today?
Not yet. The internal CLI exists and powers FSS debugging, source research, scoring, report reruns, and agent experiments. This page is for the public beta waitlist.
Who is this for?
Teams building agent workflows that need startup discovery, competitor scoring, market-source search, trend analysis, and report generation without rebuilding the FSS research stack.
What does the CLI output?
JSON by default, with Rich-formatted pretty output for humans. It is meant to be scriptable inside agent pipelines and inspectable by founders or operators.