Platform
AI Agents
Agents built for enterprise operations, not demos.
Purpose-specific AI agents grounded in your data, operating across every Maaya product — lead gen, support triage, forecasting, formulation, daily planning, strategy cadence.
How it works
What makes ai agents different
Data-grounded
Agents answer from your real data via RAG + structured queries — not from training-time guesses.
Action-capable
Agents draft POs, reply to emails, raise tickets, and book calendars with human-in-the-loop review.
Auditable
Every agent action writes an audit-log entry tied to a user, a dataset, and a decision path.
Customer-customisable
Prompts, tools, and policies are version-controlled per customer — not buried in a vendor’s SaaS.
Grounded in your customisation
Code-level customisation applies to agents too: your rules, your vocabulary, your workflows.
Cross-product agents
StratoPi agents consume signals from MaayaERP, PoultryCare, FeedPi, RestoPi, and LivMatrix.
The agent architecture
Every Maaya agent is built on a shared backbone: a tool layer (what the agent can do), a data layer (what the agent can see), a policy layer (what the agent can’t do), and an audit layer (what the agent did). Customers can review and edit the first three; the fourth is append-only.
Agents in MaayaERP
- Lead gen & qualification — drafts replies, scores against ICP, books meetings.
- Support triage — classifies incoming tickets, drafts responses, escalates when policy says so.
- Demand forecasting — item, location, week; writes the weekly replenishment plan for human sign-off.
- Anomaly watchdog — flags unusual GL entries before month-end close.
Agents in StratoPi
- BMC agent — drafts a Business Model Canvas from interview notes and desk research.
- OKR cascade agent — proposes key results that roll up cleanly.
- Cadence agent — drafts review agendas from live data and flags OKR drift.
- Daily planner — writes each person’s top 3 for tomorrow.
Agents in the vertical products
- PoultryCare — early-warning agent reads LivMatrix signals and raises disease risk alerts 24–72 hours ahead.
- RestoPi — menu-engineering agent drafts the weekly item-profitability report and recommends drops/promotes/re-prices.
- FeedPi — formulation agent re-runs affected recipes as ingredient prices move.
Grounded in your data
Agents don’t guess. Every response is either (a) produced from a structured query against your data, or (b) synthesised from retrieved documents that the audit log can show you. If the agent doesn’t know, it says so — and offers to escalate.
Ready when you are
Want ai agents in action?
One call. Bring the question you care about most. We’ll show you the real thing, not a pitch.