Pricing
Priced like a hire.
Not a sticker on software.
Every seat is one employee working a 50-hour week — so the honest comparison is what a person costs. Here’s what you save.
- A full-time hire, loaded
- ~$7,500/mo
- A Praxa employee
- $2,500/mo
- You keep
- ~$5,000/mo
and months to ramp
working within days
≈ $60,000 a year, per seat
Seats, not tiers
Add one, add three, add any number — each seat is one employee working a 50-hour week.
The plan stays flat
$2,500/mo per seat on your own AI key, or +$300/mo per seat for fully-managed compute. No per-task fee, no cost you didn't approve.
Capacity grows with seats
Every seat adds another 50-hour week of work you can point where it matters. Running more seats earns quieter volume rates — talk to us and we'll scope it.
Or start a free 3-day trial — no charge during your trial, cancel anytime.
Enterprise runs the same seat model with an isolated tenant, SSO, a dedicated audit trail, and self-hosting — every hour of delivered work shows on your dashboard, and if a role keeps maxing out its week, that's the signal to add a seat.
What it gives back
Measured against the hire it replaces.
The honest comparison isn’t a per-task fee — it’s a person. So we show each piece of work as the time it would have taken someone, and price the role against what that someone costs.
- Triage and reply to a support ticket~8 min
- Qualify an inbound lead and book the meeting~15 min
- Chase a batch of overdue invoices~30 min
- Draft a blog post from a product update~45 min
- Write the weekly competitor brief~1 hr
These are rounded estimates using a standard AI model. Inference runs on your own key, or on managed compute for +$300/mo. Harder work takes a bit more; a quick, simple task takes almost none. The dashboard shows exactly what each task did, every time.
Compared honestly
Weigh it like any hire.
You have three other ways to get this work done. Here’s the honest comparison — including the row where a person still wins.
| A Praxa employee | AI chat tools | A freelancer | A full-time hire | |
|---|---|---|---|---|
| Who does the work | Does the job end to end, on its own | You do the work — it suggests | They do, while the contract lasts | They do, once they’ve ramped |
| Proof the work landed | Every outcome verified, with a receipt you can open | None — you check everything yourself | You spot-check what you can | Reviews, eventually |
| Time to productive | Days — it proves itself first | Instant, but it never ramps | Weeks to find and brief | Months |
| A hard judgment call | Holds the work and brings it to you | Answers anyway, confidently | Depends on who you found | The reason you hired them |
| What it costs | from $2,500/mo, does the work | $20–60/mo per person, you do the work | $3–8k/mo while it lasts | ~$7,500/mo loaded |
The judgment-call row is the design, not a gap: when the call is genuinely hard, a Praxa employee holds the work and brings it to you with the context gathered — it doesn’t guess on your behalf.
01
How much does a role cost?
A Praxa employee starts at $2,500/mo — versus about $7,500/mo for a loaded full-time hire, so you save roughly $5,000 a month per role. Start on your own with your own AI key, add fully-managed compute for +$300/mo, or talk to us and we'll scope the tightest role and the sharpest price. Want to see it first? Try it free for 3 days.
02
Why not just post a price?
We do — a Praxa employee starts at $2,500/mo, and you can start today. We just don't stop there: every seat is a working employee, so talking to us gets the role scoped to your actual work — often a sharper, cheaper fit than a one-size number. Either way, the total is shown before you commit.
03
What's included?
Everything the role needs to do the job: the proof step before it goes live, the full record of what it did, and your approval gates on anything risky. Inference runs on your own AI key (BYOK) — or add managed compute + inference for +$300/mo and there's nothing to set up. One flat monthly price for the working role, no per-task fee that creeps.
04
Can I use my own AI keys?
Yes — bringing your own key is the standard way to start at $2,500/mo, so inference lands on your existing cloud bill and is never resold. Don't want to manage keys? Add fully-managed compute + inference for +$300/mo and there's nothing to set up. Enterprise brings keys against a dedicated cloud account.
05
How much work does one employee actually do?
Each seat is one employee working a 50-hour week — more time on the job than a full-time hire — and you see every hour on your dashboard. On your own key (BYOK) inference lands on your own cloud bill; on managed compute (+$300/mo) it's included. If a role keeps maxing out its week, that's the signal to add a seat — talk to us and we'll right-size it.
06
Will the price jump around?
No. It's a flat monthly price per seat. On BYOK, inference is billed to your own cloud account at cost and never marked up; on managed compute it's a flat +$300/mo. You only pay more if you decide to add another seat — no per-task fee, no cost you didn't approve.
07
Can I add a seat later?
Yes — add one, add three, add any number, whenever you like. Each seat is one more employee working a 50-hour week, and before you confirm, the dashboard shows exactly what it adds for the rest of the month and your new total — so there are no surprises.
08
Where does my data live?
On the standard plan, your data is kept separate from other customers, hosted in the US. On Enterprise, you get your own dedicated space in the region you choose, and you can host it yourself. See our security page for the details.
09
How does an employee pause itself?
Every role has a quality bar it has to clear. If its recent work slips below that bar, it stops on its own and lets you know. Nothing more happens until a person reviews it and decides to retrain it or retire it.
Ready when you are
Start yourself, or let us scope it.
A working employee for a fraction of a hire — from $2,500/mo on your own key, +$300/mo for managed compute, or tell us the role and we’ll scope the sharpest price. Want to see it work first? Try it free for 3 days.