Technical Documentation

Welcome to the reference manual for the robound.io pipeline stack. We engineer, compile, and configure type-safe backend automated workflows powered by the Weft compiler and Restate's durable execution runtime.

Note: All pipelines are compiled into standalone Rust binaries. You keep ownership of the compiled executable, minimizing vendor lock-in risk.

How It Works

Traditional low-code systems like Zapier or Make fail silently when third-party APIs update their JSON schemes, leading to lost events or corrupt datasets. robound addresses this by using Weft—a type-checked domain specific language (DSL) that defines pipeline topologies before compilation.

Durable Execution

By compiling mappers to a Restate-backed binary, the execution engine maintains intermediate state variables during third-party service degradation. If Slack, Salesforce, or your custom CRM encounters an outage, the pipeline pauses, holds values safely, and resumes cleanly once the API recovers.

Pricing Models

We decouple development retainer fees from processing and infrastructure compute costs:

Core Retainer: $750/month per active configuration. This fee includes custom mapping configuration, initial graph compilation, production deployment, prompt versioning, and continuous daily monitoring.

Variable API Processing: ~$10-$90/month. These costs comprise individual token fees from LLM endpoints (Anthropic/OpenAI) and data search/enrichment calls. Clients can either use their own credentials ($0 pass-through) or let us manage billing at cost + 20% infrastructure markup.

Pipeline Specifications

Below is a code representation of how standard pipeline graphs are declared in Weft (.wft) syntax before compiler checks are executed:

Lead Qualification Engine Spec

# ── Header ──
Project: "Lead Qualification Engine"
Version: 1.0.0

# ── Triggers ──
weekly_crawl = Cron {
  label: "Weekly Lead Crawl"
  schedule: "0 9 * * 1"
}

# ── Research & Enrichment Nodes ──
company_research = TavilySearch {
  label: "Company Research"
  queryPerPage: 5
}

contact_enrich = ApolloEnrichment {
  label: "Contact Enrichment"
  enrichFields: ["name", "email", "company_size"]
}

# ── AI Scoring Phase ──
qual_llm = LlmInference -> (response: JsonDict) {
  label: "Qualify & Draft"
  model: "anthropic/claude-3-5-sonnet"
  temperature: "0.3"
  parseJson: true
}
        

Customer Support Triage Spec

# ── Inbound Channels ──
inbound_email = EmailReceive {
  label: "Email Support"
  address: "support@client.com"
}

# ── Intent Classifier ──
classifier_llm = LlmInference -> (classification: JsonDict) {
  label: "Intent Classifier"
  model: "anthropic/claude-3-5-sonnet"
  temperature: "0.1"
}

# ── Conditional Routing Gate ──
resolution_gate = Gate {
  label: "Auto-resolve or Escalate"
  mode: "route"
}
        

Frequently Asked Questions

What happens to our pipeline if we decide to cancel? ▼

We require a standard 30-day notice. Upon cancellation, maintenance support terminates (we will no longer execute prompt updates, error log trace diagnostics, or third-party schema alignment fixes). The standalone binary is yours to host, but future API changes in external platforms may degrade its stability.

Is my customer data secure under your environment? ▼

Yes. Pipelines execute directly within your dedicated virtual private servers or isolated Wevemind cloud configurations. All credentials and transaction data remain under your environment, meaning we do not store or access your customer records.

What is the average pipeline deployment SLA? ▼

Standard configurations utilizing pre-built API integration blueprints (Slack, HubSpot, Zendesk, PostgreSQL) are deployed within 48 hours of design lock. Pipelines requiring legacy databases or undocumented architectures follow our Custom Track delivery schedule.

Can we write custom logic steps in standard scripting languages? ▼

Yes. Weft pipelines support custom logic execution using isolated script runners (Python or JavaScript engines) directly inside ExecCode nodes for quick formatting or custom parsing tasks.

Getting Started Sequence

1. Discovery Consultation: We review your tech stack and requirements to assign your Standard or Custom delivery track.
2. Design Scope Lock: You approve the pipeline node graph layout, and the 48-hour SLA period begins.
3. Draft Validation: We run test tokens through your configurations and send sample outputs to your test Slack channel.
4. Production Deployment: We deploy the compiled binary and activate live listeners.
5. Operational Check-ins: We execute monthly schema checks and tune LLM prompts based on diagnostic error traces.