Σ1διαλογος ·dialogue
Lives Everywhere
video slot
Telegram, Discord, Slack, WhatsApp, Signal, email, terminal — and more. One agent, one memory, every surface.
Universitas
Discover the Universe. An agent built to verify intelligence.
Install desktop app
Install via terminal
SOCRATES.AGENT// ΣΩΚΡΑΤΗΣ
# agi is not a singleton
video slot — patch in the product walkthrough here
[ interface preview — placeholder container, ready for hero-demo.mp4 ]
The stack
For teams
Includes Universitas's proprietary data integration, tooling, and federated memory — built into the vertical stack, not exposed anywhere else.
For developers
Composable building blocks. Bring your own integrations and infrastructure — this track is deliberately unbundled.
macOS, Windows, and Linux — or build directly against the SDK.
For developers
Call Socrates directly from your own code. Sessions hold state across calls — no context to re-send, no history to manage yourself. The SDK wraps our Stateful API; call either one.
$ pip install socrates-sdk
from socrates import Agent
agent = Agent(model="socrates-1")
session = agent.session() # state persists across calls
session.ask("what do we actually know about this?")
session.ask("and what haven't we verified yet?")
# ^ remembers the prior turn — no history to resend
$ npm install @universitas/socrates
import { Agent } from "@universitas/socrates";
const agent = new Agent({ model: "socrates-1" });
const session = agent.session(); // state persists across calls
await session.ask("what do we actually know about this?");
await session.ask("and what haven't we verified yet?");
For developers
A coding agent that lives in your terminal. Same reasoning core as the desktop app, tuned for reading, editing, testing, and shipping inside your repo.
Reasoning API
Pure quantitative reasoning. Post-trained by Universitas for mathematics, the sciences, and competition-grade problem sets — AIME-class and beyond. Call it directly, independent of the agent.
Every answer carries a chain of provenance back to its inputs — you can see the steps, not just the result.
$ pip install socrates-sdk
from socrates import Sigma
result = Sigma.solve(
problem="prove that sqrt(2) is irrational",
show_work=True,
)
print(result.answer)
print(result.provenance) # step-by-step chain of provenance
Speaks the standards, not just its own API.
Agent-to-agent
Socrates doesn't work alone. Agent-to-Agent (A2A) research networks let multiple specialized agents collaborate in real time — sharing context, cross-validating findings, and synthesizing insights across disciplines.
Deploy coordinated agent teams on your most complex research challenges.
video slot — patch in the A2A network visualization
Σ — the sum of its parts
Σ1διαλογος ·dialogue
video slot
Telegram, Discord, Slack, WhatsApp, Signal, email, terminal — and more. One agent, one memory, every surface.
Σ2αναμνησις ·memory
video slot
Learns your projects, writes its own notes, and syncs across sessions and surfaces. It never re-derives what it already worked out, and it never resets.
Σ3πραξις ·practice
video slot
Plain-language scheduling for reports, backups, and briefings. Runs unattended, stays on task.
Σ4μαιευτικη ·method
video slot
Isolated subagents with their own conversations and tools, so one line of inquiry never derails another.
Σ5ζητησις ·inquiry
video slot
Web search, browser control, vision, image generation, and reasoning across multiple models — with citations attached.
Σ6απορια ·testing
video slot
Five sandboxed backends — local, Docker, SSH, Singularity, Modal — so it can test an idea without risking your system.