- contact@verticalserve.com
Author Apps — small, versioned bundles that pair a form, a workflow, and a plan. Run them in chat, in a browser, in your editor, or headlessly from a script. Your models, your data.
Connect your systems, choose your AI model, and let the worker handle the rest
Integrate with email, documents, databases, cloud services, and enterprise apps. Okta SSO and role-based access out of the box.
Use OpenAI, Anthropic Claude, Google Gemini, or your own custom models. Your data stays with you — run models on-premise or in your own cloud.
Read emails, process documents, generate reports, monitor systems, fix issues, and reduce manual errors — all through natural language.
Deploy domain-specific skills for insurance, healthcare, retail, finance, and DevOps. Extend with your own custom skills and apps.
One platform, five surfaces. Same apps, same models, same audit trail — whichever way your team likes to work.
Drive apps from your shell. Wire into CI, cron, runbooks, and scripts. Versioned apps, playbooks, and guards built in.
~ % insightworker --help InsightWorker — Enterprise AI Worker by VerticalServe Three flavors, one install: insightworker Interactive REPL (terminal) insightworker --web Browser UI: chat + apps + files insightworker /app create "<name>" --ui Widget form for one app ~ % insightworker insightworker> /help InsightWorker Commands: /init Set up workspace (.insightworker/) /help Show this help /config Show configuration /tools List available tools /skills List domain skills /memory Show long-term memory /web [port] Launch browser UI (chat + apps) /clear Clear session quit | exit Exit Apps: /app create "what you want to do" Create an app /app list List all apps /app run <name> Execute next task /app done Mark current task done /app show <name> Show app details Playbooks (project context): /playbooks List loaded playbooks Use @playbook-name in message to load on-demand playbooks Guards (quality gates): /guard init Create example guard /guards List all guards /guard run <name> Run a guard manually
Run insightworker --web on your laptop or any EC2 box. Open a browser to chat with the agent, run declarative apps from a sidebar, and manage server-side files in a split-pane explorer. Perfect for demos and non-engineer teammates.
~ % insightworker --web Web mode: http://127.0.0.1:8765 Workspace: ~/.insightworker Renderer: bundled (web bundle) Chat + apps + file manager — same agent as the REPL. Stop with: /web stop (or just quit the REPL) In the browser: • Chat with the agent (streaming, cancelable) • Click an app → fill its widget form → live SSE panels • File manager: preview text / images / PDFs inline • Upload, create folders, download — all on the server side
Chat with the agent next to your code — every tool call visible inline with a one-click copy on code blocks. Type @ or # to reference any App in the workspace; right-click an App to launch its widget UI in your browser.
One-click apps for underwriters, claims, support, and ops. The same skills your engineers use — wrapped in a chat UI anyone can drive.
SSO, per-team policy, model whitelists, token quotas, audit logs, skills marketplace, usage analytics. See who's using what, on which model, for how much — across your whole org.
Every InsightWorker App is a small folder of five files. Versioned, signable, and portable across CLI, browser, editor, and headless runs.
Product metadata — name, version, owner, visibility, category. The face your team and marketplace see.
The form your users fill in — file uploads, text fields, dropdowns — plus the output panels they see when the run finishes.
The steps. Each step is either an LLM prompt: (judgement work) or action: shell (deterministic — runs a script, no model in the loop).
The author's how-to — markdown context loaded into every step's prompt. Plus manifest.yaml with the cryptographic signature for shared apps.
Open a terminal in any project. The agent scaffolds the bundle, enriches it from your description, and runs it — in chat, in a widget UI, or headless from a script.
Ready-to-deploy skill packs for business, development, and operations teams across enterprise domains
Policy intake & comparison, claims file preparation, underwriting review, loss run analysis, document OCR for certificates and forms, compliance apps.
Medical form processing, HIPAA-compliant document apps, patient data extraction, inventory management, order processing, and customer communication automation.
DAG review & pipeline triage, stored procedure analysis, backfill planning, database schema review, code review, incident response, and infrastructure monitoring.
You control the models, the data stays with you, and skills are extensible for any domain.
SSO, per-team model whitelists, token quotas, a private skills/playbooks marketplace, audit logging — without giving up data sovereignty. Self-hosted or SaaS.
Everything your AI Worker needs to automate enterprise apps and reduce manual effort
Read, search, and manage emails. Download attachments, draft responses, and send with human approval
Extract text, tables, and form fields from scanned documents, PDFs, and images automatically
Query databases, explore schemas, analyze data, and generate insights — all with natural language
Choose OpenAI, Anthropic, Bedrock, Gemini, or a private host. Switch per task to balance cost and quality.
Business teams, developers, and operations — InsightWorker adapts to how your team works
Automate document processing, email management, and report generation across any industry
Pre-built skills for insurance, healthcare, retail, finance — ready to deploy with domain expertise baked in
Help developers write, review, and debug code. Manage data pipelines and database operations with AI assistance
Automated monitoring, incident response, and infrastructure management with full audit trails
Four real apps InsightWorker handles end-to-end — drawn from current customer engagements.
Promote an app from your laptop to a managed production worker — with pinned skill versions, service-account credentials, per-environment policy, and full audit. Manage every deployment from one Enterprise control plane.
k8s CronJob, Airflow, or cron runs the worker on a schedule. Executes one app per tick, reports usage, exits.
Webhook, queue message, or file drop spins up a worker, runs one app, posts back, and exits. Fast-reaction automation.
A persistent worker attached to a chat channel or queue. Handles many requests over a session — conversational agents.
Run on your infrastructure, choose your AI model, and keep complete control over your data and apps.
Use OpenAI, Anthropic Claude, Google Gemini, AWS Bedrock, or your own custom-trained models. Switch models per task. No vendor lock-in.
Deploy on-premise or in your private cloud. The AI Worker runs in your network, processes data locally, and never sends information to external services.
Okta SSO, role-based access control, full audit trails, encryption at rest and in transit. Meet HIPAA, SOC 2, and CCPA requirements.
InsightWorker is the engine — CLI, VS Code, Desktop. InsightStudio is the SaaS marketplace on top of the same bundle format: business users discover apps, fill a form, click Run, watch live results in their browser. Your authors publish once with iw app publish; Studio surfaces them automatically.
Workers stay on your boxes — secrets never leave the daemon. Studio just coordinates.
iw app publish uploads your bundle to S3; Studio's marketplace surfaces it.insightworker --worker --studio <url> joins your InsightWorker to the farm.iw app pull <slug> brings any published bundle to your laptop.Reduce manual work, eliminate errors, and boost productivity with pre-built skills for your industry. Bring your own model and keep your data secure.
Bring your own model • On-premise deployment • Pre-built domain skills • Data never leaves your network