Everything under the hood

A complete AI platform — not a chat wrapper.

AIDEN combines agent orchestration, retrieval, model routing and multi-modal IO into one coherent workspace. Here's what that looks like under the hood.

AI Agents

Agents that plan, reason, and execute.

Every AIDEN agent runs a loop of planning, tool use, reasoning and synthesis. You stay in control — edit the plan, inspect every step, or let it go end-to-end.

  • Planner

    Breaks tasks into atomic, verifiable steps.

  • Reasoner

    Thinks across your docs, tools, and the web.

  • Executor

    Calls tools, files, APIs — anything you connect.

1

Planner

Break down the task

2

Tool: Search

Query the live web

3

Tool: Retriever

Grounded in your docs

4

Reasoner

Synthesize with citations

5

Response

Stream tokens to the user

Retrieval-Augmented Generation

Your data. Cited. Always.

Upload PDFs, docs, audio, video — AIDEN chunks, embeds, and retrieves them so every answer is grounded. Hover any claim to jump to the source.

  • Hybrid retrieval

    Semantic + keyword + recency ranking.

  • Citations

    Page-level references to your original files.

  • Knowledge graph

    Automatic topic clustering across docs.

research.pdf

notes.md

meeting.mp4

q3-plan.docx

slides.pptx

transcript.vtt

Answer

“Revenue grew 24% QoQ [q3-plan.docx · p.4], driven by enterprise expansion [meeting.mp4 · 12:03].”

Model Routing

The right model for every task.

AIDEN picks fast, cheap models for routine chat and the strongest reasoning models for hard problems. You pay only for what each step actually uses.

  • Auto-route

    Dynamically picks GPT, Claude, Gemini, Llama.

  • Cost-aware

    Every step is priced by tokens, not subscriptions.

  • Fallbacks

    Graceful retries across providers if one fails.

GPT-4oauto-route weight
Claude 4 Opusauto-route weight
Gemini 2 Proauto-route weight
Llama 3.2 70Bauto-route weight
Groq Fastauto-route weight
Multi-Modal

Text, images, audio, video — all native.

Ingest a YouTube lecture, transcribe a meeting, generate an image, or chat with a diagram. AIDEN handles every modality you throw at it.

  • Vision

    Chat with screenshots, diagrams, and slides.

  • Voice

    Hands-free chat with streaming STT + TTS.

  • Video

    Summarize YouTube, Loom, and long-form videos.

Text

native ingestion + generation

Images

native ingestion + generation

Audio

native ingestion + generation

Video

native ingestion + generation

Credit-Based Compute

One currency. Every model. Total transparency.

No confusing plans. Every model, tool and modality consumes credits at published rates. See every step's cost, live, in your thread.

  • Pay for compute

    Not per-seat, not per-message.

  • Live meter

    See credit spend on each step in real-time.

  • Detailed usage

    Drill down by agent, model, or modality.

Thread · Research Assistant

live
Plannergpt-4o-mini1.2c
Web Searchserp · tool0.8c
Retrieveraiden-embed0.3c
Reasonerclaude-opus8.4c
Synthesisgpt-4o2.1c

Total this run

12.8 credits

Performance & Scale

Built for production from day one.

Streaming responses, parallel tool calls, horizontal workers and queued background jobs. Your agents are responsive today — and stay that way at scale.

  • Streaming

    Tokens and tool events stream as they happen.

  • Parallel tools

    Runs independent tools concurrently.

  • Resilient

    Queued jobs, retries, and graceful timeouts.

~200ms

first token

99.95%

uptime

8x

parallel

Ready to see it in action?

Spin up your first agent in under 2 minutes.