Apideck is one of the more developer-friendly unified API platforms in 2026 - and if you are evaluating it, you have probably already noticed that. The documentation is excellent. The SDK quality is high. The pass-through architecture is better for compliance than cache-heavy alternatives. And the consumer-based pricing model rewards companies whose customers adopt multiple integrations.
But good developer experience does not answer every evaluation question. This review covers the dimensions that G2 ratings and vendor comparison pages do not: the 24-hour polling fallback hiding behind "webhook support," the payroll depth gap that matters if HRIS data is central to your product, the category ceiling at eight verticals, and the pricing tier restrictions that gate enterprise-critical features. If you are trying to decide whether Apideck is the right fit - or whether you should look at Merge, Finch, Knit, or Nango instead - this evaluation gives you the specific, verifiable data to make that call.
Bottom line: Apideck is an excellent choice for teams building accounting and CRM integrations who value developer ergonomics and want a pass-through architecture. It is a poor fit for payroll-intensive products, event-driven systems with real-time requirements, or companies that need integrations across email, calendar, meetings, e-signature, or assessment categories.
What Apideck Does
Apideck Unify is a unified API platform that normalizes access to third-party SaaS APIs behind a single, standardized interface. Rather than building and maintaining individual integrations to Salesforce, QuickBooks, BambooHR, Shopify, and dozens of others, a SaaS engineering team integrates once with Apideck and gains access to all of them through a consistent schema and authentication flow.
The platform is built on three interlocking products:
- Unify: The unified API layer - normalized schemas across categories, a single integration point for all connectors, and data model consistency that insulates your codebase from upstream API changes
- Vault: The authentication and connection manager - handles OAuth flows, token refresh, and end-user connection setup through an embeddable white-label UI
- Proxy: A direct passthrough tool for accessing vendor-specific endpoints that fall outside Apideck's normalized schema (useful for niche fields and advanced use cases)
This structure means your team owns the product experience while Apideck absorbs the ongoing maintenance cost of keeping connectors current as upstream APIs evolve.
Architecture Deep Dive
Real-Time Pass-Through: The Foundation
Apideck's most consequential design decision is its architecture: the platform operates as a real-time proxy, not a data cache. When your application calls the Apideck API for customer records from a CRM, Apideck proxies that request to the downstream provider in real time, normalizes the response, and returns it. Nothing is written to Apideck's infrastructure. The source system remains the authoritative record.
Apideck calls this zero data retention, and it is the default behavior - not a premium add-on. The practical effect:
- No Apideck data lake means no third-party data sub-processor for your customer's records
- GDPR right-to-erasure requests do not require any action on Apideck's side for application data
- SOC 2 audits are scoped more narrowly - there is no data-at-rest control to account for on Apideck's infrastructure
- Enterprise security reviews that ask "where does our data live?" get a clean answer
The trade-off is equally real: Apideck has no buffer layer. If QuickBooks or Salesforce has a degraded API, that degradation propagates directly to your users. Cache-based platforms like Merge can serve stale-but-available data during provider incidents; Apideck cannot. Whether this works for your product depends on your tolerance for provider downtime and whether data freshness or resilience is the higher priority.
The Webhook Polling Reality
Native Webhooks vs. Virtual Webhooks
Apideck supports webhooks, but the implementation varies by connector - and this is not prominently documented. For integrations where the downstream SaaS provider has native webhook support, Apideck can relay those events in near-real-time. For integrations without native webhook support, Apideck falls back to virtual webhooks: a polling mechanism that detects changes and emits them as events.
The 24-Hour Default
The default polling interval for virtual webhooks is 24 hours for many integrations. That is a 24-hour lag between a change occurring in the upstream system and your application receiving the event notification.
In concrete terms:
- An employee termination in an HRIS system might not trigger an offboarding workflow for up to a day
- A contract signed via an unsupported e-signature integration would not notify your system until the next polling cycle
- A ticket status change in a project management tool with no native webhooks would be invisible to your application for up to 24 hours
Teams building event-driven products need to confirm, for each specific connector in their target list, whether native webhook support is available. Do not assume "webhook support" means sub-minute delivery across the board.
Platforms built webhook-first - where native event delivery is the baseline, not an upgrade path - handle this differently. Knit, for example, is architected around real-time event delivery as the default, with no polling fallback for primary workflows.
What Apideck Does Store
Two data types live on Apideck's infrastructure:
- OAuth tokens and API credentials - stored encrypted in Vault, with standard key management practices
- Request and response logs - retained for audit and debugging; retention period is configurable, with longer retention available at higher plan tiers
The credentials storage is necessary and expected. The logging is useful for debugging. Neither represents the application data that compliance reviews typically focus on.
Integration Coverage
Eight Categories, 200+ Connectors
Apideck Unify covers eight integration categories as of 2026:
The total catalog spans 200+ connectors. Accounting and HRIS carry the most depth by connector count.
Where Apideck Excels
Accounting
The 48-connector accounting category is Apideck's strongest. Accounting software is notoriously fragmented by geography and company size - QuickBooks dominates the US SMB market, Xero owns Australian and UK small business, Sage serves mid-market UK and EU, and dozens of regional platforms sit below these. Apideck's accounting coverage is among the broadest available in a unified API platform, making it a strong choice for vertical SaaS products in finance, procurement, or any category that needs to read or write accounting data across a diverse customer base.
CRM
The 22-connector CRM category covers the major platforms across enterprise and mid-market. Combined with strong documentation and a clean normalized schema, this is a well-executed category.
Where Coverage Is Insufficient
HRIS and Payroll Depth
The 56-connector HRIS count looks strong until you examine what the connectors actually expose. Apideck's HRIS layer does not support:
- Payroll line-item detail (earnings, taxes, deductions, contributions by pay period)
- Write-back operations for payroll deductions or benefit contributions
- The kind of structured employment data that benefits platforms and compensation analytics tools require
Finch, purpose-built for employment data or Merge or Knit which are broader in their coverage offer better for HRIS and Payroll. For products where payroll data is central, Apideck is a broad generalist competing against a specialist. The specialist wins on depth.
Missing Verticals
Apideck does not cover:
- Email (Gmail, Outlook)
- Calendar (Google Calendar, Outlook Calendar)
- Meetings (Zoom, Teams, Google Meet)
- E-Signature (DocuSign, Adobe Sign, HelloSign)
- Assessment Tools
- Expense Management
- Subscription Management
- Communication Platforms (Slack, Teams messaging)
SaaS products that need integrations in any of these categories - productivity tools, HR technology, document automation, financial operations - will find Apideck's scope insufficient and need supplemental infrastructure. Platforms like Knit cover 13+ categories, including all of the above.
Custom Field Access Gated
Custom field mapping - the ability to surface non-standard, user-defined fields from upstream systems - is locked to the Scale tier ($1,299/month). This matters because enterprise customers almost universally customize their CRM and HRIS instances. A Salesforce deployment at a 500-person company typically has dozens of custom fields on the Account and Contact objects. An integration that cannot expose those fields is limited in its enterprise utility.
Pricing
Consumer-Based Pricing: How It Works
Apideck's pricing unit is the consumer - an end user (typically one of your customers) who has authenticated one or more integrations through your product. The key distinction from per-linked-account pricing: one consumer using three integrations counts as one consumer, not three.
Example at scale:
If 300 of your customers each have three integrations connected:
- Per-account model at $50/account: 300 x 3 x $50 = $45,000/month
- Apideck Scale tier: $1,299/month + any overage
The math favors consumer-based pricing heavily in multi-integration scenarios. The advantage narrows when customers typically connect only one integration. An additional thing to validate though is that for accounting integrations which might generally have multiple organizations within the same accounting instance - its not clear if Apideck would count them as multiple customers and hence multiple the price effectively.
The Three Tiers
Launch - $599/month
Designed for early-stage SaaS teams validating integration use cases.
- 25-50 active consumers
- 1 Unified API category (one of the eight available)
- Passthrough API, webhooks, MCP server
- Standard email support
- Not included: Custom field mapping, data scopes, SSO, white-label Vault
The single-category restriction is the most significant constraint at this tier. A product needing both CRM and accounting integrations cannot operate on the Launch plan.
Scale - $1,299/month
The practical tier for production integration use cases.
- 100-500 active consumers
- 3 Unified API categories
- Everything in Launch, plus custom field mapping and data scopes
- Not included: SSO, white-label Vault, configurable log retention
Custom field mapping is the critical addition at this tier. The jump from $599 to $1,299 is steep, and for teams whose primary need is custom field access, it is a binary gate.
Enterprise - Custom pricing
- Unlimited consumers
- All categories
- All features including SSO, white-label Vault, configurable log retention
- Dedicated CSM, premium support, SLA, connector prioritization
- Annual billing
All plans include unlimited API calls and connections - no per-call metering. A 30-day free trial covers Launch and Scale. Annual billing provides approximately 10% savings.
Developer Experience
Documentation and tooling is where Apideck has invested more than most of its direct competitors, and it shows in the day-to-day integration experience.
Documentation
The developer portal at developers.apideck.com is thorough and actively maintained. API reference docs include complete parameter descriptions, request/response schemas, and error code documentation. The coverage documentation - which maps exactly which fields are supported per connector - is particularly useful for pre-integration scoping. Most platforms leave this level of specificity to support tickets; Apideck publishes it openly. Changelogs are current, and migration guides are available when schemas change.
SDKs
Teams working in Go will need to use the REST API directly; a Go SDK is not publicly available. Ruby is also absent.
Developer Tooling
Beyond SDKs, the platform provides:
- API Explorer: Interactive browser-based tool for testing API calls without code
- CLI: Command-line interface for querying 200+ SaaS APIs from the terminal
- Vault embed: A configurable connection UI embeddable in your product, handling OAuth flows for your customers
- Portman: Open-source API test generation tool for contract testing
- Code samples: Pre-built examples and open-source building blocks on GitHub
The Vault embedding experience is particularly well-executed. End users see an OAuth connection flow inside your product - either Apideck-branded on lower tiers or your brand's on Enterprise - without your team owning authentication implementation.
MCP Server for AI Agents
Apideck has shipped an MCP server that connects AI agents to its connector catalog. Key implementation details:
- Progressive discovery: Agents start with four meta-tools that dynamically load only the specific tools needed for a task, reducing context window usage by approximately 64x compared to loading all available tools upfront
- Scope controls: Administrators can restrict agent operations by type (read/write/delete) and filter API responses at the field level
- Multi-framework support: Works with Claude, Cursor, Windsurf, OpenAI Agents SDK, and LangChain
The MCP server covers the same eight categories as the core platform. Teams building AI agents that need to interact with email, calendar, meetings, or e-signature systems will find the MCP server's scope limited. Platforms with broader category coverage - Knit exposes 13+ categories through its MCP server including email, calendar, meetings, e-sign, and assessment - offer AI agents a wider operational surface without additional integration work.
Onboarding and Support
G2 reviewers (4.9/5 across 61 reviews) consistently rate support quality highly. The team is noted for responsiveness, willingness to engage on feature requests, and direct engineering access during implementation. Several reviews mention initial onboarding complexity around the product's three-layer model (Unify, Vault, Proxy), but note this is resolved quickly with support engagement.
Compliance and Security
SOC 2 Type II
Apideck achieved SOC 2 Type II certification in 2022, with annual renewal through independent third-party audits covering security, availability, processing integrity, confidentiality, and privacy. The certification is current.
Data Minimization and GDPR
The pass-through architecture is the primary compliance benefit. Because Apideck does not store application data, it sits outside the scope of data-at-rest controls for that information. GDPR deletion requests do not require any action from Apideck for the application data itself - only for any logs or tokens in Vault, which Apideck manages.
For products selling to European enterprises with strict data residency requirements, Apideck's pass-through model is a meaningful sales asset.
HIPAA
Apideck does not publicly list HIPAA compliance or BAA availability. Teams building in healthcare-adjacent categories that process protected health information should confirm BAA availability directly with the Apideck sales team before making purchasing decisions.
Encryption and Credential Security
OAuth tokens and API credentials are encrypted at rest in Vault. Request and response logs are retained with configurable retention periods (Enterprise tier). Standard TLS is in place for data in transit.
Honest Limitations
1. HRIS Payroll Depth Is Insufficient for Payroll-Adjacent Products
Apideck's HRIS category covers 56 providers but does not expose payroll line-item detail - earnings, taxes, deductions, contributions - that benefits platforms, compensation tools, and financial wellness products need. Write operations for payroll are also absent. This is not a minor gap; it is a category mismatch. If payroll data is central to your product, Finch - with 250+ providers, line-item payroll models, and write-back operations - is a better starting point.
2. Virtual Webhooks Rely on Polling
The 24-hour polling fallback for integrations without native webhook support is a real constraint for event-driven systems. "Webhook support" in Apideck's documentation does not uniformly mean near-real-time event delivery. Confirm native webhook status for each target connector before assuming latency characteristics.
3. Eight Categories Leaves Significant Verticals Uncovered
Email, calendar, meetings, e-signature, assessment, expense management, subscription management, and communication platforms are all outside Apideck's current scope. The platform is well-suited to a focused integration problem; it is not the right choice for a product that needs broad, multi-category integration coverage.
4. Custom Field Mapping Is Scale-Only
Gating custom field mapping at $1,299/month means teams that need to support enterprise customers' customized system fields face a pricing step-change before their usage volume justifies Scale. This affects enterprise readiness at lower plan tiers.
5. No Buffer Against Provider Downtime
The pass-through architecture's integrity has a cost: no data buffer during provider outages. Teams with high availability requirements should account for the upstream availability dependencies that Apideck's architecture inherits.
6. Evaluate Apideck's Own Comparison Content Critically
Apideck publishes a significant volume of comparison content against competitors, including dedicated "Merge alternative," "Finch alternative," and "Kombo alternative" pages. This content is written by Apideck and should be read as marketing. Independent reviews on G2, Gartner Peer Insights, and analyst sources provide more balanced signal.
Who Apideck Is Best For
Accounting or CRM integrations at mid-market scale: The 48-connector accounting category and 22-connector CRM coverage, strong documentation, and consumer-based pricing make this a well-matched solution for vertical SaaS products that need to integrate with accounting or CRM software across a diverse customer base.
Developer-led teams that prioritize documentation quality: If your team's velocity is constrained by integration complexity and you need a platform where the documentation is good enough to move without opening support tickets, Apideck's documentation investment pays off quickly.
Compliance-sensitive products that benefit from data minimization: The pass-through architecture simplifies GDPR obligations and enterprise security reviews. If your sales process includes detailed data handling questionnaires, the ability to say "your data does not reside on our infrastructure" is a real differentiator.
Products with multi-connector adoption per customer: Consumer-based pricing rewards breadth. A customer using three integrations costs the same as a customer using one. This structure works well for products where integration adoption per customer is a growth metric.
Who Should Evaluate Alternatives
Payroll-intensive products: If earnings, deductions, tax data, or benefit contribution write-back are core to your product, Apideck's HRIS layer will not meet your data requirements. Start with Finch and also look at Merge.dev and getknit.dev
Event-driven products that require real-time event delivery: If your product logic depends on detecting system events within seconds or minutes, validate that your target connectors have native webhook support in Apideck before committing. For webhook-first architectures where real-time event delivery is the baseline, evaluate platforms built around that model from the start.
Products needing 3+ integration categories: If your integration roadmap includes email, calendar, meetings, e-signature, expense management, or subscription management alongside HRIS and CRM, Apideck's eight-category ceiling will require supplemental infrastructure. Platforms like Knit, covering 13+ categories including all of the above, are better suited to wide-category integration requirements.
Startups with fewer than 25 connected customers: The $599/month floor is a real cost for teams with minimal integration traffic.
Enterprise sales requiring custom field access below Scale: If your enterprise customers have customized their upstream systems and expect those custom fields to be available, the Scale tier ($1,299/month) is the minimum. Factor this into your integration cost modeling early.
How to Choose
Choose Apideck if your primary integration category is accounting, you value developer experience as a first-order concern, and your customer profile does not require payroll depth or categories outside the eight Apideck covers.
Choose Knit if you need integration coverage across a broader set of categories - particularly HRIS, ATS, email, calendar, meetings, e-signature, assessment, or communication - or if you are building AI agent-driven products that need webhook-first event delivery, zero data storage, and MCP support across 13+ categories. Knit is purpose-built for products where real-time event handling, wide category coverage, and security-first architecture are non-negotiable. Customers including Multiplier, Payoneer, and HackerEarth use Knit for this profile of requirements. For teams that need to add new connectors quickly, Knit's AI connector builder delivers new integrations in two business days.
Choose Merge if you are building HRIS or ATS integrations at enterprise scale and need the broadest normalized integration catalog, or if cache-based availability resilience during provider outages matters for your product.
Choose Finch if payroll line-item data - earnings, deductions, taxes, contributions - is central to your product, or if you need write-back operations against payroll systems.
Choose Nango if you want full ownership and transparency over the integration layer and prefer an open-source foundation with the option to self-host.
Choose Kombo if your primary market is European HR teams and GDPR-native architecture is a sales requirement or compliance mandate.
FAQ
What is Apideck's G2 rating in 2026?
Apideck Unify holds a 4.9 out of 5.0 on G2, based on 61 reviews as of 2026. Top praise covers documentation quality, support responsiveness, and the time savings from normalized integration. Common criticisms include limited coverage of niche fields within certain connectors, pricing accessibility for early-stage startups, and an initial learning curve around the platform's three-product structure.
Does Apideck store my customers' data?
For application data - CRM records, accounting transactions, employee information - no. Apideck operates as a real-time pass-through proxy: data is processed in memory and never written to Apideck's infrastructure. This is the default architecture, not an optional add-on. OAuth credentials are stored encrypted in Vault, and request/response logs are retained with configurable retention periods. Apideck calls this zero data retention, and it is accurate for application data.
How does Apideck's per-consumer pricing compare to Merge's per-account pricing?
Apideck charges based on consumers - one of your customers connecting one or more integrations counts as a single billable unit regardless of how many integrations they have connected. Merge charges per linked account, meaning each connected integration counts separately. In practice: if 200 of your customers each use three integrations, Merge would count 600 linked accounts while Apideck counts 200 consumers. Apideck's model is more cost-efficient for multi-integration adoption; Merge's model can be more predictable at lower integration density per customer.
Is Apideck suitable for HRIS and payroll integrations?
Apideck's HRIS category covers 56 providers and handles standard employee data well. However, it does not support payroll line-item detail (earnings, taxes, deductions, contributions by pay period) or write operations for payroll deductions and benefit contributions. For products where structured payroll data is central - benefits platforms, compensation analytics, financial wellness tools - Finch, which covers 250+ employment providers with line-item payroll data and write-back operations, is the more appropriate platform.
Does Apideck support AI agents?
Yes. Apideck has shipped an MCP server that connects AI agents to its 200+ connector catalog with progressive tool discovery (which significantly reduces context window usage) and administrative scope controls restricting agent permissions. The server is compatible with Claude, Cursor, Windsurf, OpenAI Agents SDK, and LangChain. The MCP server covers the same eight categories as the core platform, so teams building agents that need to interact with email, calendar, or e-signature systems will need to supplement or choose a platform with broader category coverage.
What are the most important limitations to evaluate before choosing Apideck?
Three limitations carry the most weight: First, if payroll data is part of your product's value proposition, Apideck's HRIS layer will not meet your depth requirements. Second, if your product depends on real-time event detection, confirm native webhook support for each target connector - the virtual webhook fallback uses polling cycles that can delay event delivery by up to 24 hours. Third, if your integration roadmap extends beyond Apideck's eight categories into email, calendar, meetings, or e-signature, you will hit a category ceiling that requires additional infrastructure or a platform change.
Knit is a unified API platform covering 13+ integration categories with a webhook-first, zero-storage architecture and MCP server support for AI agent workflows. SOC 2 Type II and ISO 27001 certified. G2 rating: 4.9/5 for Ease of Integration.
