Methodology

What this directory is — and isn't

This is a working catalog of products built specifically for connecting software: iPaaS platforms, workflow tools, unified APIs, embedded integration platforms, AI agent tooling, RPA tools, ETL platforms, and traditional enterprise service buses. It exists because we (Left Hook) work in this space every day, and we wanted a single page where we could see the landscape, track what's active vs. acquired vs. dead, and make informed recommendations for clients.

It is: a snapshot, opinionated about which products belong, biased toward tools that show up in our client conversations or that we want to be aware of.

It is not:a Gartner Magic Quadrant, a paid placement directory, or an exhaustive market census. We don't accept money to add or feature tools.

Categories

Every tool gets exactly one primary category. Boundaries blur (most workflow tools have AI features now; some unified APIs ship embedded SDKs) so we pick the category that best describes what the product fundamentally is, not every feature it has.

iPaaS

e.g., Zapier, Make, Workato, Boomi, MuleSoft, Power Automate

General-purpose integration platforms-as-a-service. Visual workflow builders with a large pre-built connector library, aimed at business users or ops teams.

Unified API

e.g., Merge, Finch, Codat, Apideck, Knit

One API that normalizes data across many providers in a vertical (HRIS, accounting, CRM, ATS, etc.). You write to one schema, they handle the per-provider mapping.

Embedded

e.g., Paragon, Pandium, Ampersand, Membrane

Integration platforms designed to ship inside someone else's SaaS product. The end-user-visible UI runs in your app, you own the customer relationship, the platform runs the connectors.

Workflow

e.g., n8n, Pipedream, Windmill, Trigger.dev, Gumloop

Developer-oriented automation engines, often open-source or self-hostable. More code-friendly and lower-level than iPaaS — think "Zapier you can fork."

AI Tooling

e.g., Composio, Arcade, Superface, MCP servers

Integration infrastructure built for AI agents and LLMs to take actions: function-calling, MCP, tool-use frameworks, agent-friendly auth.

Data ETL

e.g., Fivetran, Segment, Talend, Informatica

Bulk data movement and transformation. Focused on getting data from operational systems into warehouses, lakes, or analytics tools — not on driving real-time workflows.

RPA

e.g., UiPath, Nintex, Magical

Robotic process automation. Tools that drive UIs (clicks, keystrokes, screen scraping) when there's no API, often via desktop bots or browser extensions.

ESB / Enterprise

e.g., TIBCO, IBM App Connect, SAP Integration Suite

Traditional enterprise service buses and integration suites — message-oriented middleware, ESB patterns, on-prem and hybrid deployments at large-enterprise scale.

Fields we capture

description

One neutral sentence describing what the product does. We try to avoid marketing copy verbatim — we paraphrase to keep entries comparable.

founded

Year the company was founded (not when the current product launched). Sourced from Crunchbase, LinkedIn, or About pages. Where the year is ambiguous, the history field notes the uncertainty.

history

A one-line note: acquisitions, pivots, notable funding, or product-line changes that affect how we'd recommend the tool.

status

"Active" (default), "New" (founded ≥2023, still proving the model), "Acquired" (bought by a larger company — relevant because behavior often changes), "Pivoted" (pivoted away from integration), "Discontinued" (dead or sunset).

deployment

How you can run it: "saas" (vendor-hosted), "self-hosted" (you run it), "embedded" (ships inside your app), "hybrid" (mix). Many tools list multiple.

pricingModel

High-level shape only — "Freemium," "Usage-based," "Subscription (from $X/mo)," or "Contact for pricing." Pricing pages change frequently; we don't try to track every tier.

targetMarket

Who the product is built for. E.g., "AI Engineers, Enterprise," "B2B SaaS, Product Teams," "Individuals, Sales Teams."

openSource

True only when source is published under an OSI license or a well-known fair-code license (BUSL, Elastic License, n8n's Sustainable Use License). "Open core" products with a small OSS shell don't count.

keyFeatures

3–5 short bullets capturing the technically distinguishing features — not generic claims like "easy to use."

integrationCount

Vendor-claimed number of pre-built integrations or connectors. For tools where the concept doesn't apply (browser-based RPA, dynamic-API tools), we use a descriptor like "Browser-based" or "Dynamic (any API)."

experience

Our familiarity with the tool (see scale below) — not a quality rating of the tool itself.

tags

A flat, lower-case vocabulary used for cross-cutting filters (e.g., open-source, mcp, hris, ai-agents, enterprise, no-code). Tags overlap with category by design.

featured

Reserved for a small set of tools we want to surface prominently — usually because we have hands-on experience and frequently recommend them. Not a quality ranking.

The "experience" scale

This describes ourfamiliarity with the tool — how confidently Left Hook can recommend, scope, or build with it. It is not a rating of the tool's quality.

Expert

Multiple production deployments, deep familiarity with edge cases, partnerships or certifications where they exist.

Advanced

Hands-on production work, comfortable scoping and shipping integrations on it.

Intermediate

We've built with it in non-trivial ways but wouldn't claim full coverage of its surface area.

Beginner

We've evaluated the product or used it lightly — enough to know what category it's in, but we'd want to do a discovery sprint before committing to it for a client.

How tools get added

Tools enter the directory through three channels:

1. Internal radar. The Left Hook team flags products we encounter — in client conversations, on Hacker News, on launches, in competitive research. This is the biggest source. Newly-flagged tools default to experience: "Beginner" until we've actually built something with them.
2. Client work.When a client uses or evaluates a tool we don't already track, we add it. The work itself promotes our experience rating over time.
3. Suggestions. Anyone can send us a tool to consider. We don't accept payment for inclusion, but we do read every suggestion.

New entries go through a light due-diligence pass — confirming the company is real and active, identifying the right category, pulling the key feature list from the product itself (not the marketing site's tagline), and noting any acquisition or pivot history.

Citing this directory

If you reference this directory in research, journalism, an analyst report, a blog post, a podcast, a market map, or any other public work, please link back. Backlinks help us keep the directory free, ad-free, and editorially independent — and they help your readers find the source.

Suggested attribution:

Left Hook Integration Tools Directory. https://lefthook.com/integration-tools. Methodology at /methodology.

For markdown or BibTeX-style citations, or if you'd like a CSV or JSON export of the data for a specific project, just get in touch— we're happy to help.

Limitations & biases

• Snapshot in time. Pricing, status, and integration counts change. We update entries opportunistically, not on a fixed schedule.
• English-language & North-America-leaning.We're likely under-representing strong tools from the EU and APAC.
• Vendor-claimed numbers. Integration counts are taken from vendor sites and not independently verified.
• Category overlap is real. A lot of products straddle two categories. We pick one primary category and use tags to capture the rest.
• We have skin in the game.Left Hook builds integrations professionally and partners with some of the products listed. Featured status is editorial, not paid, but it's not neutral either.