Invoice arrives
Agent reads it
Checks your ERP
A person signs off
Booked, on record
With company memory
Knows which suppliers your controller always checks twice, and why, from the threads where it was decided.
Back-office workflows we build agents for
Signed contract arrives
Agent prepares the checklist
Updates your HR system
The manager signs off
Onboarded, on record
With company memory
Knows the onboarding steps that never made it into the checklist.
Back-office workflows we build agents for
Request arrives
Agent classifies it
Checks the planning
The planner signs off
Scheduled, on record
With company memory
Knows which jobs your planner never books on a Friday afternoon.
Back-office workflows we build agents for
Customer question arrives
Agent finds the case
Checks your CRM
A colleague signs off
Answered, on record
With company memory
Knows the exceptions your team grants, and who approved each rule.
Back-office workflows we build agents for
Purchase request arrives
Agent checks the budget
Matches your suppliers
The buyer signs off
Ordered, on record
With company memory
Knows which suppliers are preferred in practice, not only on the list.
Back-office workflows we build agents for
Access request arrives
Agent checks the role
Checks your directory
The owner signs off
Access granted, on record
With company memory
Knows which access requests always need the owner's call.
Back-office workflows we build agents for
Most teams already use AI, each person on their own: their own prompts, their own pasted files, their own idea of how the firm does things. Company memory gives everyone the same starting point. Praxeum learns how your company works from where the work happens, a person approves it, and the Claude, ChatGPT or Copilot each person already uses works from it.
On your own
Each person's AI knows only what that person pasted in. Two colleagues ask the same question and get two different answers.
As a team
The team's AI works from the same context and the same approved way of working, so a new colleague gets the answer your most experienced one would give.
Company-wide
Every team's context in one place, and each person sees only what they can already open. When someone leaves, what they knew stays.
From that context come skills: how your firm does one piece of work, written down from how your best people actually do it. A person approves each skill, and their name stays on it. Share a skill with the whole company, or only with the teams and roles that should use it, and set what it may do on its own.
| Skill | What it may do | |
|---|---|---|
| leave-requestsThe whole company | The whole company | sign-off |
| invoicingFinance | Finance | execute |
| refund-exceptionsFinance team leads | Finance team leads | sign-off |
| lead-qualificationSales | Sales | suggest |
| ticket-triageSupport and operations | Support and operations | execute |
Bundle the skills one job needs and you have an AI co-worker, on the Claude or ChatGPT you already run: a named set of approved skills with a job. Each skill names the person who approved it, and a person signs off where it matters.
Four examples of what a co-worker can be. Yours will look like your team.
refund-exceptions
Refunds handled the way your team actually handles them, including the edge cases that never made it into the help doc.
Finish linethe refund got approved.
See it in the demoApprove a $240 refund on an annual plan booked 41 days ago?
refund-exceptions
billing-exceptions
AR follow-ups and the billing exceptions your team makes by feel, with the money-touching step gated behind a human.
Finish linethe dunning sequence paused for a known-good account.
Pause dunning for Acme (net-60 agreed in Slack)?
billing-exceptions
renewal-followup
Lead scores reach a rep as suggestions to confirm, never as decisions. Renewal follow-ups go out on their own, with the offer your reps actually use, and the judgment calls flagged, not guessed.
Finish linethe renewal nudge went out with the right offer.
pricing-exceptions
Deal-desk knowledge is pure Slack exhaust. Praxeum turns "when can we discount?" into a cited rule, with pricing exceptions routed for sign-off.
Finish linethe non-standard discount was approved on the record.
Approve a 22% discount (above the 15% desk limit)?
pricing-exceptions
onboarding-runbook
Where documentation is most confidently wrong. Your real onboarding steps, cited, and the gaps between the doc and the practice surfaced on day one.
Finish linethe new hire got the correct, current checklist.
See it in the demoleave-requests
The leave and approval rules that live in a manager’s head, applied the same way every time, with the sign-off where it belongs.
Finish linethe leave request was approved with cover in place.
Approve Priya’s leave request for the last week of the quarter, with cover arranged?
leave-requests
escalations
Escalations handled the way your team actually handles them, with the judgment calls flagged for a person, not guessed.
Finish linethe escalation reached the right person with the context attached.
Escalate the Acme outage ticket to the on-call lead?
escalations
incident-runbook
Your postmortems already live in Notion. Praxeum keeps the runbook current as reality drifts, and flags when a step stops matching what actually happened.
Finish linethe on-call followed the current runbook, not a stale one.