AISquare Documentation

Every decision, explained.

Govern any agent, run the Trust Loop on every decision it makes, and turn governed intelligence into the Studios your customers ship. From impressive demo to deployable product, one narrative, one platform.

Govern -> Understand -> Prevent -> Fix -> RememberAI + BI = CI5-minute setup

01 The narrative spine

Three surfaces, one layer beneath them.

AISquare reads as a single through-line. You govern an agent, run the Trust Loop on what it does, and ship the result as Studios. Beneath all three sits Collective Intelligence: every surface writes what it learns into one reasoning and memory graph, and every surface is powered by what that graph already holds.

Step 01

Govern

Connect any agent, SDK or proxy. It keeps running as-is.

Step 02

The Trust Loop

Understand it, prevent the bad call, fix it in place.

UnderstandPreventFixRemember
Step 03

Studios

Ship it as experiences: docs for repos, reports, expert agents.

the layer, not a step

Collective Intelligence

AI + BI = CI. The reasoning-aware, auditable, governable layer with the memory and context of your agents and your humans.

everything above writes ineverything above draws on it
built on the reasoning graph, the base of the platform

02 The Trust Loop

Govern is four capabilities working as one loop.

Understand what an agent did, prevent the call it should not make, fix it in place, and remember the outcome. Each capability is a glass card here and a section in the Govern docs.

Understand

reasoning graph

See every decision as a reasoning graph: the claims, the evidence behind them, and the inference chain that led to each step.

Prevent

policy gates

Enforce policy at runtime. Each action passes or is blocked against the exact rule, with a citation you can audit.

Fix

correction loop

Correct an agent in place. Replay the trace, adjust the policy, and verify the new behaviour before it ships.

Remember

enterprise memory

Persist every governed decision to enterprise memory, so your agents and your team get smarter together over time.

03 Why AISquare

The layer the rest of the stack leaves out.

Observability shows you what happened. Eval platforms score it. Guardrails block the obvious. None of them tell you why a decision was made or carry the context forward. That missing layer is the whole point of AISquare.

Observabilitymissing: why
Eval platformsmissing: cause
Guardrailsmissing: context
AISquarethe missing layer

04 Quickstart

From impressive to deployable in five minutes.

Wrap the agent you already have. Nothing about how it runs changes; governance and traces come along automatically. This is the default treatment for every snippet in the docs.

quickstart.py5-minute setup
1from aisquare import GovernedAgent2 3# wrap the agent you already have4agent = GovernedAgent.from_agno(my_agent)5 6# run it exactly as before, traces ship automatically7agent.run("Issue a refund for order #4821")
Trusted · Governed · Auditable · Explainable

05 Proof

Built for teams who ship.

Dramatyze stood up a governed, customer-facing experience on AISquare in a single sprint. Read the full case study for how the spine held end to end.

30
hours to build
569
commits
3
people
Read the case study