ooligo

Superblocks

internal-tool-builder low-code · ai-app-generation · agent-governance
AI-NATIVE MCP API
RevOpsLegal OpsRecruiting & TACustomer Success
7.8 /10

What it is

Superblocks is an internal-app builder with an AI agent, Clark, sitting where the drag-and-drop canvas used to be. You describe the app in a sentence, Clark generates it as React, Python, JavaScript or SQL, and the platform holds the result inside the access rules IT already set. The closest category leader is Retool — same job, different entry point. Retool assumes an engineer who writes SQL. Superblocks assumes a domain expert who does not, and puts the engineering controls behind that person instead of in front of them.

Clark is not one model answering a prompt. Superblocks runs specialist agents for design, IT, engineering, security and QA, each applying the standards your team registered, so a generated app arrives with PII masking, audit logging and secrets handling already applied rather than retrofitted. Superblocks raised $23M in May 2025 to build it, taking total funding to $60M from Kleiner Perkins, Spark Capital, Greenoaks and Meritech, with Box CEO Aaron Levie and Workday founder Aneel Bhusri on the cap table. Instacart, Cvent, Carrier, Virgin Voyages and Benchling run on it.

Why it shows up in ops stacks

Because the alternative is already happening. Ops people are building apps in Lovable, Replit and Claude right now, and those apps want production data. Superblocks imports prototypes from all three and re-hosts them against governed connections — the specific reason it lands in a stack next to Zenity rather than competing with it.

The governance layer is the product. Superblocks sits as an integration control layer that enforces authentication, brokers permissions down to row level, abstracts secrets away from app code, and audits every call into a connected system. Superblocks 3.0, shipped 3 August 2026, four months after 2.0, added security agent swarms, custom policy agents, static analysis, continuous CVE monitoring and a private package registry. There is a Superblocks MCP server, plus an Admin MCP on Enterprise, so Claude or Cursor can drive the workspace under the same permission scoping.

AWS listed the platform in the AI Agents and Tools category of AWS Marketplace in July 2025. That matters for procurement more than for engineering: it is the path that spends committed AWS budget instead of opening a new vendor.

A use case worth buying it for

Commercial-real-estate brokerage Matthews used Clark to generate offering memorandums, cutting a 3-5 day process to 12 hours (vendor-reported). Generalize it: the recurring document or approval workflow that a domain expert understands end to end, that reads from two or three systems of record, and that no engineer will ever prioritize. Deal-desk quote approvals, vendor-security intake, candidate-scorecard rollups. The person who knows the rules builds the app, and the security agents keep it from reaching data that person cannot already see.

Pricing

Teams is $125/month month-to-month or $100/month on annual — a flat team rate, not per seat. That covers up to 15 builders, 100 Governed Agent Units a month, and one hosted app, with additional apps at $10/app/month. There is a 14-day trial and no free tier.

GAUs are the meter, and they pool at the org level rather than charge per builder seat. Consumption scales with request complexity: a header is 1 GAU, a navigation colour change 2, a Salesforce table with a chart 4, a full sales dashboard 8. So 100 GAUs is roughly twelve dashboards a month for the whole team. Top-up packs run $625 for 500 GAUs and $1,250 for 1,000, and the pool resets monthly.

Enterprise is quote-only, and the real number is wide. Vendr’s tracked Superblocks purchases put the median at $20,448/year, with the observed range running $4,362 to $169,630. The Standard Marketplace Package on AWS Marketplace lists at $149,999 for a 12-month contract, with consumption past the base package billed at $1.00 per unit. Treat $20K as the mid-market anchor and the Marketplace figure as the ceiling for a full VPC deployment.

Best for

IT, RevOps and ops leaders at 500-plus-employee companies who have already found staff prototyping internal tools in consumer AI builders and need those tools to run against production data under policy. The ROI band is best where you have many small app requests and a backlog engineering will never clear.

Do not buy it if the app faces customers — this is internal software, and the hosting and permission model assume it. Do not buy it if your builders are engineers who already write React and SQL comfortably; Retool’s canvas is cheaper per builder and the agents go unused. And if the thing you want has no interface at all, a SaaS-to-SaaS sync on a schedule, that is n8n or Zapier for a fraction of the price.

Alternatives, and when to pick them instead

  • Retool — the segment leader, $120M ARR as of October 2025 on $190M raised at a $3.2B valuation. Pick it when engineers are the builders and you want the canvas, three separate meters (seats, workflow runs, agent-hours) and the deeper component library. Retool’s Team tier at $10/builder undercuts Superblocks badly at small builder counts; the gap closes once governance enters the requirement list.
  • Microsoft Power Apps — the incumbent by installed base, because it arrives inside the M365 licensing your firm already pays for. Pick it when the app is a form over SharePoint or Dataverse and the buyer is IT rather than engineering. The licensing economics beat Superblocks; the generation quality does not.
  • Lovable — the fastest-growing entrant in the segment, a $330M Series B at a $6.6B valuation in December 2025 after passing $200M ARR inside a year. Pick it when the app is a prototype or a customer-facing site and nothing needs to touch production systems. It is also the tool your team is using without telling you, which is an argument for Superblocks rather than against it.

Watch-outs

  • 100 GAUs is about twelve dashboards, shared across all 15 builders. Open Teams to 15 people and the monthly pool is gone in a week. Guard: pilot with 2-3 builders for a full billing cycle, read actual GAU burn per app type, then size top-up packs at $1.25/GAU before widening access — and set the auto-refill threshold deliberately instead of leaving the default, because auto-top-up charges the card on file.
  • Everything you are buying it for sits on Enterprise. Source control (GitHub, GitLab, Bitbucket, Azure DevOps), secrets management, VPC and Cloud Prem deployment, SSO/SAML/OIDC, RBAC, audit logs, observability pipelines and the Admin MCP are all above the Teams line. Guard: if governance is the reason for the evaluation, price Enterprise from the first call. The $1,200/year Teams tier is a trial, not a step on a path.
  • The quoted range spans nearly 40x. Vendr’s $4,362 low and $169,630 high describe the same product. Guard: anchor the first counter on the $20,448 median rather than the Marketplace list price, and decide separately whether routing through AWS Marketplace, which spends committed AWS budget, is worth the delta over a direct contract.
  • A different company also called Superblocks had its assets acquired by Banzai International on 7 November 2025. That one is an AI website and SEO hosting builder; this one is not, and search results conflate the two. Guard: confirm the vendor is superblocks.com before security review or procurement, and check the legal entity name on the paperwork.