Skip to Content

Application — Peeler

Orchestrate teams of AIs
on any task.

Peeler coordinates multiple AI providers in deterministic, multi-stage pipelines. Decompose complex tasks, run parallel analyses, cross-validate between models, and iteratively refine results — all with complete reproducibility.

Not a Chatbot

What if you could hire a team of AIs — and they actually worked together?

Peeler doesn't send your document to one AI and hope for the best. It orchestrates a team: one model extracts structure, another validates accuracy, a third generates the final report. Each stage has explicit dependencies, deterministic caching, and real-time progress streaming.

Other tools let you talk to one AI at a time. Peeler coordinates multiple AIs on the same task — using Claude for nuanced analysis, GPT-4 for structured extraction, and a local model for privacy-sensitive stages — in a single, deterministic pipeline.

1
Extract — Claude analyzes structure and key points
2
Validate — GPT-4 cross-checks accuracy and completeness
3
Enforce — CLIPS rule engine applies compliance policies
4
Report — Local model generates final structured output
5
Review — Grok gets snarky with feigned praise
app-split-pipeline-visual-b

Any Input, Any AI

Every format.* Every provider.* One pipeline.

PDFs, Word documents, images, CSVs, JSON — Peeler handles them all. Route each stage to the best provider for the job: Claude for nuanced analysis, a local model for privacy-sensitive data, a rule engine for policy checks.

Peeler leverages nxusKit capabilities to employ non-LLM AIs — rule engines, constraint solvers, Bayesian networks — alongside language models in the same pipeline. Because not every step in your workflow needs to generate text.

*Almost every. We do not currently support magnetic tape, floppy discs, paper tape, or punched cards. We also do not support voice calls to 411, smoke signals, or carrier pigeon. If your pipeline requires any of these, we'd love to hear about it.

app-split-formats-visual-a
Input formats
PDF · DOCX · TXT · MD · RTF · CSV · JSON · XML · Images
Analysis types
Summary · Comparative · Key Points · Legal Review · Technical Analysis · Custom
AI providers
Claude · OpenAI · Ollama · + nxusKit rule engines and solvers
Export
Markdown · Text · JSON · Clipboard
Interfaces
Desktop GUI + headless CLI
app-split-formats-visual-b

Architecture

Deterministic by design.

Multi-stage DAG execution

Explicit dependencies, parallel and sequential provider modes, real-time event streaming. Each stage targets its own AI provider and model.

SHA256-based caching

Deterministic input hashing guarantees reproducible results across runs. Rerun a pipeline on the same input and get the same output. Always.

Session persistence

25-result history with full session persistence. Return to any prior analysis. Export at any time.

Ready to orchestrate your first pipeline?

Explore plans, start a free trial, or request beta access.

View Plans & Pricing Request Beta Access

Enable Saved Interests?

If you continue, we will save this item and enable Saved Interests and My Recommendations for your account or current session.

We may use what you save, along with your account, purchase, and support context, to help you find relevant products, docs, and support guidance.

Reminder emails are optional and are not turned on by this step.

Learn more about how we handle your data

Database neutralized for testing: no emails sent, etc.
Neutralized