Superblocks
Also known as: Clark, Clark AI, Superblocks Clark, Superblocks 3.0
Enterprise internal app platform whose Clark AI agent generates production apps from natural language inside existing user permissions, with a knowledge graph, user preference memory and three deployment models including full cloud prem.
Superblocks is an enterprise platform for building internal applications, with Clark as its AI coding agent. Clark takes a natural language prompt and generates a production grade internal app: connected to real enterprise data, carrying the organization design system, integrations, permissions, single sign on and audit logging automatically. It runs in a plan mode and a build mode, supports checkpoints and rollbacks, and produces React and TypeScript the customer owns outright and can sync to their own git repository.
The governance model is the differentiator and it is unusually concrete. Clark works on top of an organization knowledge graph made of data schemas, API documentation, agent memory, existing app patterns and integration configurations. More importantly it operates inside existing user permissions across systems like Postgres, Snowflake, Salesforce and Databricks, so it can only read and generate code against data the individual builder is already authorized to see. That is permission bounded generation rather than a governance layer applied after the fact.
Memory is explicitly tiered and named: Clark User Memory, Clark Integration Memory and Clark Organization Memory. Deployment is equally explicit, with three published models. Cloud runs data and AI inference in the Superblocks cloud. Hybrid keeps production data in the customer virtual private cloud while only non production data reaches Clark. Cloud-Prem places the entire platform including AI inference inside the customer AWS, GCP or Azure account while Superblocks continues to manage it.
The honest limits are packaging rather than capability. The two upper memory tiers, single sign on, audit logs, source control, secrets management, observability pipelines and VPC deployment are all Enterprise only, so the self serve tier is a genuinely smaller product. Agent consumption bills in Governed Agent Units on top of the base price, and the consumption rate per unit of work is not published.
Vendor details
Canonical URL
https://www.superblocks.com
Category
Agent builder
Subcategory
AI generated internal applications with enterprise governance
Funding status
Private, 60M USD raised in total including a 23M USD round from Spark Capital, Kleiner Perkins, Meritech Capital and Greenoaks announced alongside the Clark launch. New York based, CEO Brad Menezes.
Company status
independent
Use cases & customers
Primary use cases
Target customers
Deployment options
Integrations
50+ integrations across databases, APIs, SaaS and cloud services including Postgres, Snowflake, Salesforce, Databricks and Slack. Enterprise adds an Admin MCP surface, source control through GitHub, GitLab, Bitbucket and Azure DevOps, secrets management via AWS Secret Manager, Azure Key Vault, Google Cloud Secret Manager and HashiCorp Vault, embedded applications and a private package registry. Two way local editing with VS Code, Cursor or the Superblocks CLI, and generated React and TypeScript that the customer fully owns and can push to their own git repository.
In practice
An operations lead describes a Salesforce performance dashboard in plain language and Clark generates a working app wired to live CRM data, already carrying the company design system, permissions and audit logging.
A regulated enterprise runs Cloud-Prem so the entire platform including AI inference sits inside its own cloud account, letting business users build on production data without it ever leaving the environment.
A platform team lets analysts build their own back office tools while Clark stays inside each builder existing database permissions, so nobody can generate an app against data they could not already query.
Sources & related URLs
Research sources
Agentic Index coverage score
10.5 / 14 capabilities · 75%
| Integrations & Tool Calling | Full |
|---|---|
|
Clark builds apps wired to live data over 50+ integrations on every plan, with integration knowledge describing each connected system. Apps and workflows act on those systems with centrally managed credentials and permissions. Sourcesuperblocks.com/pricingread 2026-09-29 |
|
| Workflow Orchestration | Full |
|
Clark works in a Plan mode, where the user edits the plan, then a Build mode that creates the application across many steps, with checkpoints between generations. Workflows for the current product are documented but still marked coming soon. Sourcedocs.superblocks.com/building-with-clark/plan-and-build-modesread 2026-09-29 |
|
| Knowledge Grounding & RAG | Full |
|
Clark builds against maintained Knowledge at app, organization and integration level (data schemas, API documentation, existing app patterns, integration configurations). It is managed on a Knowledge page and written for LLM retrieval. Sourcedocs.superblocks.com/building-with-clark/knowledgeread 2026-09-29 |
|
| Human Oversight & Guardrails | Full |
|
In Plan mode Clark proposes an approach, and the user edits it or clicks 'Build it' before Clark builds. Generations are saved as checkpoints that can be restored before publishing, and publish gates and policies govern what ships. Sourcedocs.superblocks.com/building-with-clark/plan-and-build-modesread 2026-09-29 |
|
| Security, Identity & Governance | Full |
|
Teams includes role based access control, and Enterprise adds SSO with SAML or OIDC (Okta, Entra ID, Google) and audit logs. The trust center lists SOC 2 Type 2 and HIPAA, with penetration testing and access log controls. SourceSuperblocks pricing page and trust.superblocks.comread 2026-09-29 |
|
| Observability & Auditability | Full |
|
Audit logs record builder activity (edits, deploys, permission changes), end user activity and platform changes. Each Clark generation is kept as a checkpoint with commit history, and Clark's conversation history per application can be retrieved through the Admin MCP. Enterprise adds observability pipelines. Sourcedocs.superblocks.com/admin/audit-logsread 2026-09-29 |
|
| Memory & State Persistence | Partial |
|
User knowledge keeps 'Personal preferences for how Clark works with you' (code style, workflow preferences, app types). Clark drafts it from the user's input, it is private to the user and used across all their Clark conversations, and entries can be viewed and refined. No lifetime is stated, and preferences Clark applies are guidance rather than a memory of past work. Sourcedocs.superblocks.com/building-with-clark/knowledge/user-knowledgeread 2026-09-29 |
|
| Deployment & Data Residency | Full |
|
Superblocks documents three deployment models: Cloud, Hybrid and Cloud-Prem. In Hybrid, 'Production data remains in the customer VPC for deployed apps'. In Cloud-Prem, 'The full Superblocks platform, including AI inference, is deployed within your AWS environment', with guides for ECS Fargate, Cloud Run, Kubernetes, Azure, VMs and Docker. Sourcedocs.superblocks.com/enterprise/deployment-overviewread 2026-09-29 |
|
| Prebuilt Agents, Templates & Packs | Partial |
|
Existing app patterns captured as app and organization knowledge shape new generations, and 50+ integrations ship on every plan. These are assets the customer assembles, and no library of prebuilt agents or starter templates is documented. Sourcesuperblocks.com/pricingread 2026-09-29 |
|
| Triggers & Channel Coverage | Partial |
|
Workflows triggered by webhooks, cron schedules, Slack messages and inbound email are documented for the current Clark product but marked coming soon, with an early access list joined by contacting the vendor. Scheduled Jobs and webhook callable Workflows are documented only on the legacy documentation site for the earlier low code application builder. In the current product, those triggers are not yet generally available. Sourcedocs.superblocks.com Workflows and docs-legacy.superblocks.com core conceptsread 2026-09-30 |
|
| Model Flexibility & Routing | Partial |
|
Enterprise lists 'Smart Model Routing', which is Superblocks routing between models, and Cloud-Prem runs AI inference inside the customer's AWS account. No page documents the customer choosing a provider or bringing its own model, so this is vendor internal routing. Sourcedocs.superblocks.com/enterprise/deployment-overviewread 2026-09-29 |
|
| APIs, SDKs & MCP Extensibility | Full |
|
Superblocks' own Admin MCP server (app.superblocks.com/mcp, bearer token) lists, deploys and undeploys apps, creates, updates and deletes integrations, manages permissions, roles and folders, and queries audit events. Generated code syncs to the customer's git repository and edits locally through the CLI. Sourcedocs.superblocks.com/admin/mcp-serverread 2026-09-29 |
|
| Testing, Debugging & Optimization | Partial |
|
Policies with run history, findings and publish gates, and SBOM and CVE checks, validate the apps Clark generates before they ship. No evaluation of Clark's own generations with scored tests or a comparison loop is documented. Sourcedocs.superblocks.com/admin/mcp-serverread 2026-09-29 |
|
| Browser & Computer Use | Not documented |
|
Clark acts through integrations, generated code and the platform's APIs. No browser, desktop or computer control is documented. Sourcedocs.superblocks.com/building-with-clarkread 2026-09-29 |
|
The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded
Pricing
100 USD per month billed annually, or 125 USD billed monthly
base subscription plus Governed Agent Units plus hosted apps
Included quota
Teams includes 100 Governed Agent Units per month, one hosted app, staging and production environments, 50+ integrations, Clark User Memory and up to 15 builders. Enterprise adds organization level GAU top ups shared across the team.
What is public
Both tiers, the monthly and annual base prices, included Governed Agent Units, builder ceiling, hosted app allowance, the additional app rate and a full feature comparison table are published openly on the pricing page. Only Enterprise pricing and GAU consumption rates are withheld.
Billing mechanics
A flat base price per team rather than per seat, with the Teams tier scaling to 15 builders at no additional seat cost. Annual billing saves 20 percent against monthly. Agent consumption meters separately in Governed Agent Units. Enterprise is quoted per organization with custom terms, dedicated account team and uptime SLAs.
Cost watchouts
The Teams tier is gated well below enterprise readiness: SSO and SAML, audit logs, source control, secrets management, observability pipelines, embedded applications, VPC deployment, smart model routing and both upper memory tiers are Enterprise only. Any security review that asks for SSO or audit logs forces a jump to custom pricing. Agent work consumes Governed Agent Units beyond the included 100 per month, and hosted apps beyond the first cost 10 USD each per month, so both the agent and the app count scale independently of the base price.
Variable cost rationale
The base price is public and exact and the per app rate is a published flat 10 USD, so most of the bill is predictable. The genuine variable is Governed Agent Unit consumption, which is bounded by a defined included allowance and topped up through defined packs rather than metered without limit. Graded medium rather than high on that basis, and deliberately one step below retool, whose hourly agent billing at model dependent rates has no comparable ceiling.
Additional watchouts
Do not read the 100 USD base as the cost of running Superblocks. It buys the platform, not the agent work, and it excludes every enterprise control.
Overage / add-ons
Agent usage beyond the included 100 Governed Agent Units per month is bought through GAU packs as an add on, with organization level top ups shared across the team on Enterprise. Hosted apps beyond the first are billed at 10 USD per app per month.
Sales call required
Mixed (some tiers require a call)
Free / trial
14 day free trial on Teams; no permanent free tier
Lowest paid plan
Teams
Commercial notes
Unusual in this category for not charging per builder seat inside the Teams tier, which makes it cheap for a small team to give many people building access. The trade is that everything an enterprise security review asks for sits behind the custom tier.
Key ambiguities
The consumption rate is the gap. Superblocks does not publish how much real agent work one Governed Agent Unit buys, nor the price of a GAU pack, so the 100 included units cannot be translated into a number of app builds. Superblocks has an AI credit use policy and a GAU documentation page; ask how many units a typical app build consumes.
Missing data
Governed Agent Unit consumption rate per unit of work, GAU pack pricing, and Enterprise base pricing.
Related vendors
- Lindy — AI teammate that lives in Slack, answers from your tools and…
- Activepieces — Open source, MIT licensed automation platform where every one of 700…
- AgentX — No code platform to build, evaluate, and deploy multi agent…
- AI Library — No-code platform for building and running AI agents, with a…
- Aigensei — Enterprise agentic workflow platform that runs business processes…
- Airia — Enterprise platform that unifies agent orchestration, security, and…
Alternatives to Superblocks
The closest documented capability profiles to Superblocks among agent builders tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.
- Joget11.0 / 14Fuller documented coverage on Model Flexibility & Routing
- Activepieces11.5 / 14Fuller documented coverage on Triggers & Channel Coverage and Model Flexibility & Routing
- Cohere North11.0 / 14Fuller documented coverage on Triggers & Channel Coverage and Model Flexibility & Routing
- Dify12.0 / 14Fuller documented coverage on Prebuilt Agents, Templates & Packs and Triggers & Channel Coverage
- Minded12.0 / 14Adds documented Browser & Computer Use
- Vellum11.0 / 14Fuller documented coverage on Model Flexibility & Routing and Testing, Debugging & Optimization
Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded