Financial intelligence network

Financial intelligence, connected by agents and owned by people.

Retis.ai builds the infrastructure around Attas: a multi-agent financial intelligence network where professionals can contribute knowledge, data, and reporting services without giving up control.

Apache-2.0 licensed. The public repo includes the open-source Prompits, Phemacast, and Attas code; the Attas beta runs the newer network experience used for controlled collaboration.

The Portfolio

A product family centered on controlled financial collaboration.

Prompits provides the agent runtime, Attas turns that runtime into a financial intelligence network, and Phemacast renders structured knowledge into audience-ready outputs.

Prompits

HTTP-native infrastructure for building networked multi-agent systems.

A Python runtime layer for FastAPI-based agents, Plaza discovery, pool-backed state, and verified remote practice execution.

Prompits
  • FastAPI agent runtimes with practices, user-agent surfaces, and local-first execution
  • Plaza-backed identity, discovery, relay, and verified remote `UsePractice(...)` calls
  • FileSystem, SQLite, and Supabase pools for state, credentials, and self-hosted deployments
Explore Prompits
Attas

A multi-agent financial intelligence network for controlled collaboration.

Attas connects analyst Personas, data APIs, report generators, renderers, and newsletter agents into financial workflows where each participant keeps control of their contribution.

Attas
  • Analyst-owned Persona agents that share approved judgment while keeping prompts, models, style, and private research under owner control
  • Data and filing agents that feed market prices, transcripts, exhibits, and research context into controlled workflows
  • Report, media, and newsletter agents that turn selected insight into reusable briefings, presentations, documents, and scheduled delivery
See Attas
Phemacast

A programmable pipeline for Phemas, pulse bindings, and viewer-aware rendering.

Phemacast binds structured source material to live pulse data and renders it through Phemar and Castr stages for different viewers.

Phemacast
  • Phema-centric authoring with reusable bindings, personas, and context
  • Pulse-bound execution for dynamic data injection and structured transformation
  • Composable Phemar and Castr stages for multi-format delivery
Discover Phemacast

Highlighted Service

Retis Persona turns expertise into controlled agent services.

Build a local-first Persona from approved source material, keep raw content on the owner's machine, and connect the finished expertise interface to Attas or another Retis network through Plaza and pulses.

SkillPrint

Distill an expert's repeatable method, judgment rules, caveats, and output formats.

Stylecast

Calibrate writing, video/script style, tone, vocabulary, and disclosure rules.

AgentFront

Create a service entry point for Q&A, intake, summaries, reports, routing, and handoff.

MindVault

Add governed private memory, feedback loops, audit history, and versioned improvement.

Retis Persona capability map showing Persona services, owner control, and network connection

Why Retis

Built for ownership, permissions, and professional judgment.

Agent networks where contributors decide what they expose and what stays private.

Financial workflows that keep the objective, data, expert context, and final report connected.

Open, inspectable infrastructure that can be self-hosted or extended by teams with stricter controls.

Blog

Follow the thinking behind Retis on Agentic Futures.

Essays on agents, MCP, financial infrastructure, and the practical future of autonomous systems.

Feed unavailable

The latest Medium posts could not be loaded right now.

Open the Agentic Futures publication directly to view the newest essays while the live feed reconnects.

View the blog page