BotFriendly
Directory for AI agents to discover bot-friendly services, APIs, and tools. The Yellow Pages for autonomous software.
facts
- role
- Founder
- year
- 2025–present
- discipline
- R&D · App
- stack
- Next.js, Supabase, Vercel, OpenAPI
brief
01 problem
AI agents need to interact with external services — payment processors, data providers, communication tools, storage — but there's no centralized way for them to discover which services are agent-friendly. Most APIs are designed for human developers reading docs, not for autonomous software making runtime decisions about which service to use.
02 constraints
The directory needs to serve two audiences simultaneously: human developers evaluating services and AI agents discovering them programmatically. The data model must be structured enough for machine consumption but rich enough for human browsing. Cold-start problem — the directory is only useful with sufficient listings.
03 solution
BotFriendly is a structured directory where every listing includes machine-readable metadata: authentication methods, rate limits, pricing models, capability descriptions, and agent compatibility scores. Services are categorized by what they do (payments, storage, communication, data) and how agent-friendly they are. Providers can self-list through a submission portal. The directory is queryable both through a web UI and a programmatic API that agents can call directly.
highlights
- 01Curated directory of agent-compatible services and APIs
- 02Machine-readable metadata enabling autonomous service discovery
- 03Self-serve listing portal for service providers
- 04Categorization by capability, pricing model, and agent compatibility level
- 05API endpoint for programmatic directory queries
notes
Why this exists
As agents become more autonomous, they'll need to make runtime decisions about which services to call. Today, a developer hardcodes those choices. Tomorrow, an agent should be able to discover that it needs a payment processor, query BotFriendly for options, evaluate them against its constraints (cost, latency, supported currencies), and integrate — all without human intervention.
The data model
Each listing is a structured record: name, description, category, authentication method (API key, OAuth, none), pricing model (free, usage-based, subscription), rate limits, supported formats, and a compatibility score based on how well the service supports automated interaction. This isn't just a link directory — it's a machine-readable service registry.
Part of the stack
BotFriendly sits alongside AgentFinance (agent wallets) and an unannounced distribution layer in Spectral's agent infrastructure thesis. Together, they form the primitives for an agent economy: discover services, deploy agents, and give them the financial tools to transact.
results
- Structured metadata schema covering auth, pricing, rate limits, and capabilities
- Dual-interface: human-browsable web UI and machine-queryable API
- Self-serve listing flow for service providers
- Categorization taxonomy for agent service discovery