/work
Direct examples of the systems companies actually ask for, structured around workflow cost, operational control, and implementation reality.
projects
11
pattern
4
mode
build
Eleven Project Types
These are examples of the work itself, not abstract categories. Each one can be scoped as a real implementation engagement.
01
What It Is
A system for lead qualification, CRM updates, follow-up drafting, pipeline summaries, and meeting preparation.
How It Gets Built
Map the sales workflow, connect CRM, forms, inbox, calendar, and enrichment tools, then use AI for summarization and drafting while keeping routing and stage logic deterministic.
02
What It Is
A support layer that answers common tickets, looks up order or account context, drafts replies, and escalates risky cases.
How It Gets Built
Cluster historical tickets, build retrieval over policies and help content, connect support tools and commerce systems, then add confidence thresholds and human handoff.
03
What It Is
A pipeline that turns invoices, forms, contracts, claims, and attachments into structured data and downstream actions.
How It Gets Built
Ingest from upload, inbox, or drive, apply OCR and extraction, validate required fields, route exceptions to review, and push clean outputs into the system of record.
04
What It Is
Operational systems for prior auth, claims intake, denial triage, appeals drafts, and revenue-cycle follow-up.
How It Gets Built
Start with one narrow workflow, connect source documents and status data, use AI for extraction and drafting, then keep submission and final decisions inside strict review controls.
05
What It Is
AP, AR, collections, close support, vendor onboarding, and approval routing for finance teams handling repetitive workflow volume.
How It Gets Built
Map the approval logic, connect inbox, ERP, accounting systems, and payment tooling, then use AI for summarization and exception support while keeping all financial actions controlled.
06
What It Is
An internal assistant over SOPs, docs, tickets, Slack, contracts, and wikis that answers questions with source grounding.
How It Gets Built
Fix permissions first, index the right sources, build retrieval and reranking, then return cited answers with freshness and confidence signals.
07
What It Is
A workflow that helps analysts review suspicious activity, assemble evidence, prioritize cases, and move faster without losing control.
How It Gets Built
Connect transaction data, identity signals, payment tools, and prior case history, then use AI for case summaries and evidence extraction while humans own final decisions.
08
What It Is
Internal AI support for ticket triage, test generation, incident summaries, PR context, and engineering runbooks.
How It Gets Built
Connect repo, ticketing, CI, docs, and logs, start in draft mode, then add bounded actions only after the context layer and review path are reliable.
09
What It Is
A system for campaign briefs, asset routing, reporting summaries, lifecycle workflows, and performance-linked creative iteration.
How It Gets Built
Map the campaign workflow, connect analytics, ad platforms, CRM, and asset systems, then use AI for briefs, drafts, and summaries while approvals stay structured.
10
What It Is
A company-specific orchestration layer where multiple agents handle intake, retrieval, planning, QA, and reporting across one operational process.
How It Gets Built
Define the workflow stages first, give each agent a narrow role and explicit tools, then manage state, handoffs, logging, and fallback through a controlled workflow engine.
11
What It Is
Installation options for the project layer, including OpenClaw, Hermes, Codex, and Claude Code setups alongside the core build.
How It Gets Built
Package the project add-on as a full installation: OpenClaw for the interface layer, Hermes for a lighter operational layer, Codex for terminal-first implementation, and Claude Code for coding support.
Build Pattern
Most high-value AI implementation work follows the same structure. The workflow changes. The underlying build pattern does not.
Map
The first step is not prompting. It is identifying the manual steps, handoffs, exceptions, and business rules that make the work expensive.
Connect
The build usually sits across the systems the team already runs on: CRM, support, docs, spreadsheets, APIs, inboxes, and internal tools.
Structure
AI handles extraction, summarization, drafting, retrieval, and decision support. Deterministic logic handles routing, permissions, state, and controls.
Control
Production systems need human review paths, confidence thresholds, audit trails, and visible failure handling. Otherwise it is a demo.
Next Step
The brief is for scoped work. The call is for fit. If the workflow is expensive, repetitive, or fragile, it is usually possible to tell quickly whether the system should be built.