The organizations that get the most out of PANTA OS share a pattern: they roll out in phases, name champions, and curate ruthlessly. Here’s the playbook.Documentation Index
Fetch the complete documentation index at: https://help.pantaos.com/llms.txt
Use this file to discover all available pages before exploring further.
The four phases
Phase 1 — Pilot (weeks 1–4)
Goal
Prove value with one team, one use case, one assistant.
Scope
5–15 people from a single team — usually sales, HR, or support.
What to build
One excellent assistant grounded in real organizational knowledge.
Success looks like
Pilot users self-report time saved or quality improved. Word spreads to other teams.
Phase 2 — Expand (weeks 5–12)
Goal
Multiply the pilot — three more teams, three more assistants.
Scope
Each new team starts the same way: one assistant, real data, daily check-ins.
Critical
Name a champion in every team. They’re the ones who actually drive adoption.
Watch out for
“Build everything at once” energy. It produces a messy Library and frustrated users.
Phase 3 — Embed (months 3–6)
Goal
Make PANTA OS part of how work gets done — not a side tool.
Signals
Teams reference assistants in their normal workflow. Onboarding starts with “and here’s our PANTA OS workspace.”
What to add
Connected tools (Outlook, Notion, etc.) once basic adoption is real.
What to retire
Assistants no one uses. Be ruthless — a clean Library compounds.
Phase 4 — Optimize (month 6+)
Goal
Get more value per token. Higher-quality output for the same spend.
Tactics
Right-size models. Trim system prompts. Cap expensive assistants. Use Auto Mode aggressively.
Reporting
Quarterly business review with leadership: spend, value, what’s next.
Build the flywheel
Best assistants in one team get adopted by others. Wins in Community spread organically.
Roles to set up early
Executive sponsor
A VP-level voice who owns the rollout outcome at leadership reviews.
Workspace admin(s)
Two or three people for platform stewardship — branding, identity, budgets.
Champion per team
One person in each pilot/expansion team who drives adoption locally.
Library curator
Often the same as a champion — owns the catalog quality for their team.
Common failure modes
Big-bang rollout
Big-bang rollout
Inviting everyone in the company on day one. Adoption thins out and the Library fills with mediocre experiments.
No champion
No champion
Tools without owners die. Each team needs a person who cares.
Free-for-all building
Free-for-all building
No naming convention, no tag taxonomy, no quality bar. The Library becomes unusable.
No measurement
No measurement
“AI is good for us” is not a metric. Decide upfront what you’re measuring and check at month 3.