Recurse
Recurse is a serverless platform for defining, testing, and deploying custom AI agents through a declarative agent.yaml manifest. Engineers specify agent inputs, outputs, tools, and verification logic in YAML; Recurse runs agent candidates serverlessly, measures their outputs against a quality bar, and deploys only the proven specialist. Agents can invoke Codex and Claude, register custom Python functions as tools, and export as tools, MCP servers, or bots. Billing is usage-based: $0.013 USD per minute of execution plus token costs that vary by model.
Recurse is a serverless platform for defining, integrating with Codex and Claude. InnovaAI rates it 4.2 of 10 for agency adoption, best for Founder, Tech Lead and Project Manager roles.
Agency Audit
Recurse is a serverless platform for engineering teams to build, test, and deploy custom AI agents defined through a declarative agent.yaml manifest. Agencies with coding-heavy workflows benefit most: teams building specialized agents for content generation, design validation, or quality assurance can define measurable verification bars, iterate on agent candidates until they pass those bars, then deploy as tools or MCP servers. Best suited for engineering-led agencies where a Founder or Tech Lead owns agent development and wants to avoid the overhead of managing agent infrastructure.
3recommended
36/mo
No paid plan published
High
Illustrative scenario. Not a guarantee. Net capacity needs a verified paid base plan, and none is published for this service, so it is not modeled. Hours saved come from the service estimate; implementation, taxes, and unprovided usage charges are excluded.
- Founder handling custom agent design and iteration
- Tech Lead handling agent quality verification and testing
- Project Manager handling agent deployment and integration
- Your team has no in-house engineers or coding expertise; Recurse's value is locked behind manifest authoring and tool registration, both requiring Python and schema knowledge.
- You use only off-the-shelf generalist agents like ChatGPT or Claude without custom tooling; Recurse adds overhead if you never need to deploy specialized agent variants.
- Your agency operates on a tight per-project budget and cannot absorb token costs for iterative agent testing; Recurse's usage-based pricing compounds during the development phase.
Internal Adoption Path
No paid plan published
36 hr/mo
3 seats × 12 hr each
$2,700/mo
modeled at $75/hr labor rate
No paid plan published
Illustrative scenario. Not a guarantee. No verified paid base plan is published for this service, so subscription cost and net capacity are not modeled. Implementation, taxes, and unprovided usage charges are excluded.
Platform Features
Core capabilities of Recurse
Declarative agent.yaml manifest
Defines agent identity, input/output schemas, and tool registration in a single YAML file. Allows Founders and Tech Leads to version-control agent specs alongside code and ensures every agent run validates against a pinned contract before execution.
Serverless agent execution with per-minute billing
Runs agents without managing infrastructure; charges $0.013 per minute of secure execution. Eliminates DevOps overhead for engineers who would otherwise provision containers or manage API rate limits.
Candidate testing and measurement
Iteratively runs multiple agent variants and scores their outputs against a quality bar. Lets your Tech Lead or Strategist refine prompts and tool sets until an agent passes verification, then deploy only the proven specialist.
Schema validation and contractual outputs
Enforces input and output schemas at runtime; agents cannot return malformed results or accept invalid parameters. Reduces downstream debugging for Project Managers integrating agent outputs into client workflows.
Custom Python tool registration
Register functions from tools.py as agent capabilities alongside built-ins. Lets your engineers compose domain-specific logic (e.g., design validators, content filters) without rewriting agent prompts.
Deploy as tools, MCP servers, or bots
Once an agent passes verification, export it as a reusable tool, MCP server, or bot. Enables Account Executives and Project Managers to integrate proven agents into client deliverables or internal workflows without re-engineering.
What Makes Recurse Different
Unique advantages vs similar tools in this niche
Manifest acts as a hard boundary on what the model may accept, call, or claim
vs Generalist agents with unconstrained tool accessThe engine refuses a manifest it does not recognize and validates every run against the pinned schema.
Coding agent owns the design loop while Recurse runs and measures candidates
vs Manual agent iterationThe parent agent changes the specialist rather than supplying an answer when the search stalls.
Usage-based billing with no markup on model token rates
vs Subscription agent platforms with idle chargesOnly active requests are billed and no markup is added to model token rates.
Value Equation
Outcome-likelihood-time-effort assessment for Recurse
Limited agency channel
Recurse scored below the agency-resellability threshold (agency_fit_score < 50). The Value Equation projects agency-side outcomes, which don't apply to tools without a clear resell pathway.
Contact RecursePricing
Recurse platform cost to your agency
Pay as you go
- No monthly subscription required
- Pay only for what you use — see per-unit rates below
- Cancel anytime, no contract lock-in
How usage-based pricing works
Recurse charges per consumption unit (per minute (secure execution / serverless)). Below are the component rates the vendor publishes. Each row is a separate charge: your total cost combines them based on your configuration and volume. Component rates range from $0.013 per minute (secure execution / serverless).
Final agency cost = (sum of selected component rates) × client usage volume. Confirm a usage estimate with each client before quoting.
Component Rates
Cost per unit: total depends on your configuration and volume
Add-ons
Optional extras priced on top of any main plan
No verified white-label program for Recurse: client-facing delivery runs under the platform's native branding.
Market Intelligence
Offer + scale economics for Recurse
Limited agency channel
Recurse scored below the agency-resellability threshold (agency_fit_score < 50). It's a useful tool but not designed for white-labeled or retainer-based reselling, so we don't publish productized offer economics for it.
Contact RecurseInvestment Decision Framework
Strategic vetting analysis for Recurse
Situational Fit
Fit depends on your client mix
Buy If
4Your Founder or Tech Lead spends 6+ hours per week manually testing and refining AI agent outputs before deployment, and you need a structured way to automate that iteration loop.
Your team builds custom agents for repeatable tasks like design validation, content quality checks, or code review, and you need a way to pin input/output schemas so agents never return malformed results.
You deploy agents as internal tools or MCP servers and currently lack a serverless harness, forcing your engineers to manage infrastructure or rely on generic LLM APIs without verification gates.
Your Project Managers or Strategists define agent requirements but your engineers lack a declarative format to translate those requirements into testable, measurable agent specs.
Skip If
4You use only off-the-shelf generalist agents like ChatGPT or Claude without custom tooling; Recurse adds overhead if you never need to deploy specialized agent variants.
Your team has no in-house engineers or coding expertise; Recurse's value is locked behind manifest authoring and tool registration, both requiring Python and schema knowledge.
Your agency operates on a tight per-project budget and cannot absorb token costs for iterative agent testing; Recurse's usage-based pricing compounds during the development phase.
Your workflows do not have measurable quality bars or verification logic; Recurse's core strength is enforcing contracts and measuring candidate outputs, which is wasted if you cannot define what 'good' looks like.
Bottom Line
Recurse is a serverless platform for engineering teams to build, test, and deploy custom AI agents defined through a declarative agent.yaml manifest. Agencies with coding-heavy workflows benefit most: teams building specialized agents for content generation, design validation, or quality assurance can define measurable verification bars, iterate on agent candidates until they pass those bars, then deploy as tools or MCP servers. Best suited for engineering-led agencies where a Founder or Tech Lead owns agent development and wants to avoid the overhead of managing agent infrastructure.
Reality Check
Recurse requires hands-on coding expertise to define agent manifests, schemas, and verification logic. Agencies without in-house engineers or those treating AI agents as a one-off feature rather than a core capability will see minimal ROI. Token-based pricing scales with agent usage, so high-volume deployments need cost monitoring.
High effort: requires technical configuration and team training
Academy for Recurse
Work through it in order: the course for this service first, then the modules behind it.
Course for this service
Recurse Agency Implementation, Building AI Agent Services
Learn how to architect, test, and deploy custom AI agents using Recurse's declarative agent.yaml manifest for your clients. This course teaches agencies how to build productized agent services, manage candidate testing workflows, and structure pricing around per-minute execution costs to maximize margins on automation projects.
Open the courseNo Academy modules are published for this service yet. Browse the full Academy
Why this category matters
The commercial case before the tooling.
Core concepts
The mental model you need to price and scope the work.
- Wiring Over WidgetsConcept
The AI agent itself is a commodity, but the value for agencies lies in the integration layer: connecting a pre-built agent to a client's CRM, calendar, and review cycle. This framework shifts focus from selecting the 'best' agent to mastering the wiring process. For example, an agency using Vendasta's white-label AI receptionist for a local business must configure it to match the client's booking rules and follow-up cadence, turning a generic tool into a tailored service. As agentic AI adoption grows (77% of decision-makers now run agents in production), clients expect this customization. Agencies that treat agents as components and invest in repeatable wiring processes can charge retainers for ongoing optimization, rather than one-off setup fees.
- Wiring Over WidgetsConcept
The AI agent market sells finished workers, but the strategic value for agencies lies not in the agent itself, which is increasingly a commodity, but in the wiring that connects it to a specific client's CRM, calendar, and review cycle. This framework, 'Wiring Over Widgets,' argues that agencies that treat agents as components rather than products win. The agent is the widget; the wiring is the integration, customization, and ongoing optimization that turns a generic tool into a tailored solution. For example, a white-label platform like Vendasta provides AI employees, but the agency's role is to configure them for each local business's unique lead flow and follow-up process. This wiring is where retainer pricing originates, as it requires ongoing maintenance and adjustment. Recent research shows that 88% of B2B marketers face foundational gaps, meaning clients need help not just deploying agents, but ensuring their operations can support them. Agencies that master the wiring can charge a premium for the irreducible value they add.
- Wiring Premium Over Agent CommodityConcept
Pre-built agents are converging on the same underlying model capability, so the agent itself prices toward zero. What holds value is the wiring: the mapping of a specific client's CRM fields, calendar rules, escalation paths, and review cadence into the agent's loop. Agencies that sell the agent as the product compete on seat price against every reseller of the same worker. Agencies that sell the wiring charge for discovery, field mapping, exception handling, and monthly tuning, which is retainer work. Vendasta's white-label AI workforce and Relevance AI's pre-built sales agents both arrive configured out of the box, which means the configuration is not the moat; the client-specific plumbing is. A practical test: if a competitor could swap your agent vendor next quarter without the client noticing, you sold a commodity. If the swap would break their CRM hygiene, approval chain, or reporting, you sold wiring.
Decision and risk
How to judge the fit, and the ways it goes wrong.
- AI Agents Rule: Wire the Agent, Not the ProductEvaluation Rule
Treat the AI agent as a commodity component and focus your value on the integration into the client's specific workflows, systems, and review processes.
- AI Agents Rule: Wire the Agent, Not the ProductEvaluation Rule
Treat the AI agent as a commodity component and charge for the integration into the client's specific systems and workflows.
- AI Agents Decision: Resell a Finished Agent vs Wire Agents Into Client SystemsDecision Framework
IF your client roster shares one repeatable function (lead capture, review requests, inbox triage) and you can resell a white-label agent without touching their stack, THEN package the finished agent as a low-configuration productized service. IF each client's value sits in the wiring between the agent and their CRM, calendar, and review cycle, THEN sell the integration as the deliverable and treat the agent itself as a replaceable component.
- The Productized Agent Trap: Why AI Agent Services Stall Without Client-Specific WiringFailure Pattern
- The Demo-to-Retainer Gap: Why AI Agents Stall After the Pilot CallFailure Pattern
Delivery system
Blueprints and procedures for running it as a service.
- Agentic Intake Triage Offer (10-14 days)Implementation Blueprint
A fixed-scope deployment that puts a pre-built agent in front of inbound leads, support tickets, or document queues, then wires its output into the client's CRM, calendar, and review cycle. Priced as a setup fee plus a monthly retainer for monitoring and tuning.
- Agent Scope Contract (Onboarding)Operating Procedure
- Agent Access Provisioning and Data Boundary Check (Onboarding)Operating Procedure
12 modules selected for Recurse
Frequently Asked Questions
Answers about pricing, setup
Recurse is a serverless harness for building and deploying custom AI agents. Your engineers define an agent's inputs, outputs, tools, and verification logic in a declarative agent.yaml manifest. Recurse runs agent candidates, measures their outputs against a quality bar, and deploys only the proven specialist as a tool, MCP server, or bot. It integrates with Codex and Claude, allowing agents to invoke specialized models for specific tasks.
Recurse has a free plan; its paid prices are not published.
Founders and Tech Leads own agent design and deployment, saving time on infrastructure and iteration loops. Strategists and Project Managers define measurable quality bars and agent requirements, which Recurse translates into testable specs. Account Executives benefit indirectly by deploying proven agents as client deliverables or internal tools without re-engineering.
A Tech Lead or Founder iterating on a custom agent candidate typically spends 4-6 hours per week on manual testing, prompt refinement, and infrastructure setup. Recurse compresses that to 1-2 hours per week by automating candidate runs and enforcing verification gates, saving 3-4 hours per week per engineer. Payback depends on token volume during development; high-iteration projects see faster ROI.
Yes. Manifests require YAML authoring, tool registration requires Python, and verification logic requires schema design. Agencies without in-house engineers should not adopt Recurse. If your team has one or more full-stack engineers or backend developers, they can own manifest authoring and tool composition.
Once an agent passes your verification bar, Recurse exports it as a tool, MCP server, or bot. Your team can then integrate it into internal workflows, client deliverables, or third-party platforms. No additional infrastructure or DevOps work is required after deployment.
Recurse validates every output against your pinned schema before returning it. If an agent violates the schema, the run fails and returns an error instead of a malformed result. This prevents downstream bugs and ensures your Project Managers and Account Executives always receive valid data.
Yes. You register custom Python functions as tools in your agent.yaml manifest. Those functions can call APIs, query databases, or invoke any external service. Recurse makes those tools available to the agent at runtime.