Why eClips

The model is
the easy part.

Calling a model API takes an afternoon. Getting documents in from email and WhatsApp, holding risky actions for a person, logging each decision and writing the result back to your ERP is where the months go. eClips Workforce is that layer, already built.

What's underneath

Workforce reads documents with Google's Gemini and runs most of its agents on Anthropic's Claude models. Both companies are named as sub-processors in our data processing agreement.

The honest comparison

Building vs. buying.

The work you take on when you build agents on a model API yourself, next to what Workforce already does.

CapabilityBuilding it yourselfeClips Workforce
Getting startedBuilding it yourself
An engineering project
eClips Workforce
Self-serve sign-up, with a free starter plan
Domain agentsBuilding it yourself
Written and tested by your team
eClips Workforce
Agents for finance, procurement, HR, legal, sales and operations
ERP integrationsBuilding it yourself
One integration at a time
eClips Workforce
Write-back to Business Central and other ERPs through their APIs
Company memoryBuilding it yourself
Build from scratch
eClips Workforce
A Context Book with your people, approval limits and company facts
Graduated autonomyBuilding it yourself
Your own rules engine
eClips Workforce
Trust levels 1 to 5, earned from each agent's own track record
Workflow builderBuilding it yourself
Build the graph engine, canvas and run orchestration
eClips Workforce
A workflow canvas with conditional steps and live run streaming
Browser automationBuilding it yourself
A Playwright wrapper and your own safety model
eClips Workforce
eClips Reach, a browser agent with its own safety checks
Human approvalBuilding it yourself
Build the approval screens and workflow
eClips Workforce
A triage queue in the dashboard, with approvals from Slack
Audit loggingBuilding it yourself
Roll your own
eClips Workforce
A log of agent actions and approval decisions
Multi-tenant isolationBuilding it yourself
Row-level security, key rotation and org scoping, built by you
eClips Workforce
Row-level security in the database
GuardrailsBuilding it yourself
Prompt engineering
eClips Workforce
A written constitution every agent's output is checked against
SupportBuilding it yourself
Public docs and forums
eClips Workforce
The team that built it, and a public changelog

The architecture

The model reads.
Workforce does the rest.

A Gemini model reads each document and pulls out the fields. A Claude model in the right domain agent decides what to do with them and scores its own confidence. Workforce handles everything around those calls. It takes documents in from each channel, routes them on your rules, holds anything risky for approval, writes back to your ERP and keeps the audit trail.

Every agent's output is checked against a written constitution before anything happens. The first rule on that list: never take an irreversible action without explicit human approval in the triage queue.

How Workforce works
architecture
// Your documents arrive
email / WhatsApp / Slack / API
// Workforce takes them in
store → pick the right agent
// The models work here
extract → classify → decide
defaults: gemini-2.5-flash extracts, claude-sonnet-4-6 decides
// Workforce acts here
check rules → hold for approval → write → log
// Your ERP receives the result
Business Central and other ERPs, over their APIs

Common objections

What buyers ask us.

We already have engineers. Why not build it ourselves?

You can. The question is what else they could be building. Intake from email and chat, extraction, approval screens, audit logs and tenant isolation are the same work at every company. We have built that part. Your engineers can spend their time on the work only your company has.

We already pay for a model API. What does Workforce add to the bill?

Workforce is billed in credits. Each agent action draws credits based on the work it does and the model it uses, so a short invoice on a fast model costs less than a long contract on a larger one. The plans, including a free starter plan, are on the Workforce pricing page.

What about agent frameworks like LangGraph?

We use LangGraph ourselves, inside Workforce. A framework gives you building blocks. Workforce is what we built with them: the intake channels, the agents, the approval queue, the audit log and the dashboard your operations team works in. If you want to design your own chains, use a framework. If you want agents working on your documents without a build project first, that is what Workforce is for.

What if we need to customize it?

Agent types, outcome contracts, routing rules and the contents of your Context Book are all yours to configure. If you need an agent we don't have, we can build it with you.

Is this just a wrapper around a model?

No. The model reads the document and proposes a decision. Workforce does everything around that call. It takes documents in from email, WhatsApp and Slack, scores its own confidence, routes on your rules, holds risky actions for a person, writes the result back to your ERP and logs every step.

Ready to see it on your documents?

Bring a few real invoices, purchase orders or contracts. We'll run them through Workforce on the call, and you'll see what it pulled out and how confident it was.

Talk to us

Start with
one problem.

Tell us where the work gets stuck, whether it's a spreadsheet at month-end or an ISO 9001 audit on the calendar. We'll show you the product that fits, or tell you if none of ours does.

A 30-minute live walkthrough. No slides.