Published · Updated
What is an AI Work Unit?
An AI Work Unit is FernLex's unit of AI processing. Different operations consume different amounts depending on how much work the model does: classifying an email is cheap; analyzing a stack of documents for risk is expensive. Work Units let us price AI by the job it finishes — not by an opaque token counter your firm can't predict.
The practical benefit is budgeting. Nobody budgets their practice management software by database queries, and you shouldn't have to budget AI by tokens. Work Units are fixed per operation, so on any given day you know exactly what an action will cost before it runs.
What operations cost
Typical consumption per operation:
- Email classification: 100 units
- Document summary: 400 units
- Drafting task: 800 units
- Risk analysis: 1,200 units
- Multi-document analysis: 2,000 units
What each plan includes
Starter ($49/user/mo) is core practice management without AI — 0 Work Units. Professional ($99/user/mo) includes 1,500,000 Work Units per user each month, pooled across the firm, for document summaries, basic extraction and writing assistance. Business ($149/user/mo) includes 5,000,000 Work Units per user for the full AI workforce — matter intelligence, evidence processing, inbox triage, intake, forms and drafting.
Firms that need more can buy packs starting at $15 for 500K credits, any time, without changing plans. Every plan includes a 7-day free trial with no credit card required.
A worked example: one busy month of usage
Numbers are easier to judge in context. Here's a heavy month for a single Professional user:
- 3,000 email classifications × 100 units = 300,000 units
- 400 document summaries × 400 units = 160,000 units
- 80 drafting tasks × 800 units = 64,000 units
- 40 risk analyses × 1,200 units = 48,000 units
- 20 multi-document analyses × 2,000 units = 40,000 units
- Total: 612,000 units — well inside the 1,500,000 included with Professional
Do unused Work Units roll over?
No. Work Units reset each billing cycle to keep pricing simple and predictable. If your usage is consistently near the ceiling, that's a signal to move up a plan or add a pack — not to ration AI work mid-month. Metered AI that teams hesitate to use defeats the purpose of having it.
Why FernLex prices by work, not tokens
Token pricing sounds precise but pushes the complexity onto you: costs shift with model, prompt length and context size, and no one can predict a monthly invoice from a token count. Work Units invert that. Each operation has a fixed price, so a document summary costs the same in week one of the trial as it does at month twelve.
- Predictable: fixed unit costs per operation, published and consistent.
- Comparable: you can weigh a unit price against the paralegal hours the work would otherwise take.
- Pooled: units are shared across your firm, so light users subsidize heavy matters automatically.
- Expandable: when a litigation-heavy month arrives, add a pack instead of downgrading your workflow.
Frequently asked questions
What is an AI Work Unit in FernLex?
It's FernLex's unit of AI processing. Operations consume fixed amounts — email classification costs 100 units, document summary 400, drafting 800, risk analysis 1,200, multi-document analysis 2,000 — so firms can predict AI usage like any other utility.
Do AI Work Units roll over?
No. They reset each billing cycle to keep billing simple and predictable. If you consistently run near your limit, upgrade the plan or add a pack rather than rationing AI work.
Which plans include AI Work Units?
Professional includes 1,500,000 units per user per month, pooled across the firm; Business includes 5,000,000 per user for the full AI workforce. Starter is $49/user/mo without AI. Extra packs start at $15 for 500K credits.
How much work do the included units buy?
On Professional's 1,500,000 units: about 15,000 email classifications, or about 3,750 document summaries, or roughly 1,250 risk analyses — or a mix of all three. Business's 5,000,000 units cover over three times that volume across the full AI workforce.
Are Work Units the same as AI tokens?
No. Tokens are a raw model-cost measure that varies with every request. Work Units are a fixed price per completed operation — one number you know before the work runs, regardless of how the underlying model processed it.
What happens if we run out of Work Units?
You can purchase additional packs at any time — starting at $15 for 500K credits — without changing plans, or move up a tier if the usage is ongoing. Running near the ceiling consistently is a planning signal, not a wall.