okaihub

AI Agents & Infrastructure tools — page 65

AI agent tools are software that lets an LLM plan and execute multi-step tasks by calling tools, reading context, and acting on external systems without a human prompting each step.

This is the broadest and fastest-moving category in the directory, and it covers three distinct layers that are often conflated. At the bottom is agent runtime and orchestration — the frameworks and hosts that give a model a loop, a tool interface and memory. In the middle are the infrastructure services agents depend on: model routers and gateways, vector stores and retrieval layers, sandboxed execution environments, evaluation harnesses and observability. At the top are finished agent products that ship a particular workflow to end users.

The distinction matters when comparing tools, because they are not really substitutes. A framework is something you build on and own the result of; a gateway is something you put in front of models you already use; a finished agent is something you adopt as-is. Many products straddle two of these layers, and the pricing models differ sharply between them — usage-based token or seat pricing at the infrastructure layer, subscription pricing at the product layer.

A large share of this category is self-hostable or offered as open source, though this directory only records licensing for a minority of listings, so treat that as a characteristic of the space rather than a measured statistic. Where it does apply, deployment model — hosted, self-hosted, or embedded — is often a more important differentiator than the feature list, because it determines where your data and your compute live.

At a glance

Products listed3,214
Free or freemium1,678of 2,840 with published pricing
Paid875
With traffic data1,768top: 56M/mo
Sub-types tracked12

Top AI Agents & Infrastructure tools by traffic

3,201–3,214 of 3,214
  1. View Papers by Ouroboros details
    #3201
    Papers by Ouroboros

    Ten plugins that plug into ChatGPT, Claude, Grok, and Cursor and keep your AI on the facts you saved: Claim, Continuity, Deposit, Desk, Invoice, Milestone,…

  2. View Zendesk AI details
    #3202
    Zendesk AI

    ZendeskのAIは、顧客や従業員を案内するだけで終わらせず、問い合わせを最後まで解決し、サポートの品質・スピード・対応規模を引き上げます。 無料で試す. Listed in AI Agents & Infrastructure.

  3. View uvx details
    #3203
    uvx

    Astral builds high-performance developer tools for the Python ecosystem, starting with Ruff, uv, and ty. Listed in AI Agents & Infrastructure.

  4. View Arcmira: YouTube Transcript Search details
    #3204
    Arcmira: YouTube Transcript Search

    Arcmira makes YouTube transcript search and monitoring for your AI. It is an SF-based AI company building the search engine for the spoken web.

  5. View Enconvo x MCP details
    #3205
    Enconvo x MCP

    AI that works across your Mac: ask from any app, talk instead of type, and let Agents do the busywork. Listed in AI Agents & Infrastructure.

  6. View Zenable details
    #3206
    Zenable

    AI code review that catches and fixes bugs before commit. Listed in AI Agents & Infrastructure.

  7. View WRDO details
    #3207
    WRDO

    WRDO gets things done in your neighbourhood, on WhatsApp or the app. Ask for a quote, a booking, a table or dinner and he sorts it. Cape Town and the Winelands.

  8. View Crypto Mcp details
    #3208
    Crypto Mcp

    Live crypto market data for AI agents, with the source and age of every answer attached. Listed in AI Agents & Infrastructure.

  9. View Statiko details
    #3209
    Statiko

    Real-time Telegram monitoring and analytics. Track every channel. Listed in AI Agents & Infrastructure.

  10. View GitHub Spark details
    #3210
    GitHub Spark

    A team of researchers and engineers at GitHub prototyping the tools, workflows, and ideas that will shape the future of software development.

  11. View Void Editor details
    #3211
    Void Editor

    Void is an open source AI IDE with full privacy. Listed in AI Agents & Infrastructure.

  12. View Paradigm AI details
    #3212
    Paradigm AI

    Paradigm is a team of expert research agents built for businesses to create and maintain thousands of up-to-date profiles on companies, people, and markets…

  13. View Agent Council details
    #3213
    Agent Council

    Eight live x402 APIs for AI agents: web access, telegram search api, token safety, chain data, search, video. Pay per call in USDC on Base, no API key.

  14. View RiverScript details
    #3214
    RiverScript

    Your all-in-one space to turn any audio, video, and live recordings into text. Supports capturing live audio directly from your computer and files up to 50 GB.

Frequently asked questions

What are AI Agents & Infrastructure tools?
AI agent tools are software that lets an LLM plan and execute multi-step tasks by calling tools, reading context, and acting on external systems without a human prompting each step. This directory lists 3,214 of them, 2,547 of which have enough detail to be indexed, each confirmed live and serving its own product site as of the date shown on its listing.
Are AI Agents & Infrastructure tools free?
Most have some free access. Of the 2,840 AI Agents & Infrastructure tools with a published pricing model (88% of those listed), 59% offer a free or freemium tier, 31% are paid, 21% include a free trial and 30% do not publish pricing and require contacting the vendor. These categories overlap — a tool can be both freemium and paid — so the figures total more than 100%. Pricing for the remaining 374 was not published on their sites, so it is not counted here. Note that "freemium" means a usable free tier with limits on volume, seats or features rather than the full product, and a "free trial" is time- or usage-limited rather than free indefinitely.
What are the most popular AI Agents & Infrastructure tools?
By estimated monthly traffic, the most-visited AI Agents & Infrastructure tools in this directory are PolyBuzz, Jotform AI Agents, OpenRouter, Manus and Thomson Reuters. This ranking covers the 1,768 tools (55% of those listed) that have published third-party traffic estimates; tools without traffic data are excluded rather than assumed small. Traffic is one signal of adoption, not a quality judgement, and estimates vary in accuracy between providers.
What types of AI Agents & Infrastructure tools are there?
The most common sub-types in this category are AI, Productivity, Automation, Development and Chatbots. These labels come from the source directories the products were imported from, so they reflect how those sources categorise the space rather than a taxonomy we imposed. A single tool often spans several of them.
How are these tools ranked and verified?
Each entry points at the product's own website, which was checked to confirm it is live and still serving that product; sites that have shut down or no longer offer it are left out, and two listings for the same product are merged into one. Within a category, products are ordered by estimated monthly traffic where that figure exists, and alphabetically where it does not. Traffic and rating figures are third-party estimates and may be incomplete. Listings are re-checked periodically and each entry carries the date it was last verified.
How do I choose an AI Agents & Infrastructure tool?
The criteria that matter most in this category are: Which layer you actually need: framework, infrastructure service, or finished agent. Buying the wrong layer is the most common mistake in this category; How tool-calling and state are handled across long runs, and what happens on failure partway through — retry, checkpoint, or silent truncation; Whether context and memory are yours to export. Lock-in here is expensive because it is the accumulated state, not the code, that has value; Cost behaviour under real load. Per-step pricing looks cheap in a demo and expensive in a loop that retries. Start from the specific job you need done rather than the longest feature list, and trial against your own data or content — demo material in this category is usually chosen to flatter the product.

What to look for

  • Which layer you actually need: framework, infrastructure service, or finished agent. Buying the wrong layer is the most common mistake in this category.
  • How tool-calling and state are handled across long runs, and what happens on failure partway through — retry, checkpoint, or silent truncation.
  • Whether context and memory are yours to export. Lock-in here is expensive because it is the accumulated state, not the code, that has value.
  • Cost behaviour under real load. Per-step pricing looks cheap in a demo and expensive in a loop that retries.
  • Observability: can you see why an agent did something, after the fact, without reproducing the run?