Intelligence Pipeline

Your intelligence
pipeline.

Design the workflow. Assign the models. Drop in your data. Each stage hands its output to the next — and you see everything.

Start building See how it works

You define the workflow

Any model, any stage

Full cost visibility

Bring your own API keys

Swap prompts without redeploy

Edit at any stage

Reusable lenses

Complete run history

You define the workflow

Any model, any stage

Full cost visibility

Bring your own API keys

Swap prompts without redeploy

Edit at any stage

Reusable lenses

Complete run history

How it works

Build once.
Run forever.

Define the stages. Assign the models. Write the prompts. From that point on, it’s one input and a run button.

Design the pipeline

Add stages in order. Pick a model for each. Write the prompt. Save it with a name that means something.

Drop in the input

A document. A URL. A data file. A transcript. Each stage picks up where the last one left off.

Review every step

Read, edit, or download the output at any stage. Token cost tracked automatically — no surprises.

Pipeline Builder

Building

Market Intelligence Pipeline

Web Research

perplexity-sonar

—

Analysis

claude-sonnet-4

—

Draft

claude-sonnet-4

—

Verify

claude-haiku-4

—

Total run cost

Your API provider · direct billing

$0.000

You define the workflow

Any model, any stage

Full cost visibility

Bring your own API keys

Swap prompts without redeploy

Edit at any stage

Reusable lenses

Complete run history

You define the workflow

Any model, any stage

Full cost visibility

Bring your own API keys

Swap prompts without redeploy

Edit at any stage

Reusable lenses

Complete run history

The old way

Every message costs
more than you see.

Chat carries the full conversation on every reply. Research, analysis, writing, and verification — all running through one premium model, at premium rates, every single time.

AI Chat Session

claude-sonnet-4-5

The hidden cost

One model for everything

Every reply re-reads the entire conversation. Research, a quick edit, a fact-check — all billed at the same top-tier rate, on a growing context window.

$0.00

per 4-stage run via chat

The time cost

Back and forth, by hand.

Six prompts to get one structured output. Copy the result. Start the next prompt. Repeat for every stage. Foundry runs them all — you come back to results.

6×

more prompts for one equivalent pipeline run

* Approximate costs based on published API pricing as of May 2026. Actual costs vary. Foundry does not mark up API costs — your provider bills you directly.

The Foundry way

Build the workflow.
AI does the work.

Define each stage once. Assign the right model to each job. Every run hands output from one stage to the next automatically — the right model, at the right cost, doing exactly what it’s built for.

Stages run in sequence. Each one gets the output of the last.

Each stage uses the model built for that job — not the most expensive one available.

Every output is visible, editable, and downloadable. No black box.

Save the pipeline. Run it again with new input in one click.

Use cases

One engine.
Any workflow.

The pipeline structure stays the same. What changes are the stages, the models, and the prompts. Your methodology, encoded once, runs on anything.

Research & Intelligence

Pull live data, synthesize findings, produce a structured report. Run the same brief through two pipeline configs and compare what comes back.

Web Search

Synthesis

Report

Verify

Legal Document Prep

Review source documents, flag risks, produce structured summaries. Every stage output is visible and downloadable before the next stage runs.

Review

Risk Flag

Summary

Draft

Investment Analysis

Pull market data, build a thesis, stress test assumptions, write the memo. Save named pipeline versions for different asset classes or deal types.

Data Pull

Thesis

Stress Test

Memo

Content & Publishing

Research a topic, write the narrative, apply editorial standards, ship. Your process, running while you do something else.

Research

Narrative

Editorial

Cleanup

Product Specifications

Turn discovery sessions into structured requirements, specs, and technical docs. Update the input and re-run — the methodology stays locked.

Discovery

Requirements

Spec

Review

Market Analysis

Research competitors, segment the landscape, extract insights, produce the brief. A named pipeline is a reusable methodology — run it on a new market in seconds.

Research

Segmentation

Insights

Brief

Built different

Your keys. Your models.
Your process.

Foundry runs on your API keys. Configure every stage independently — the pipeline runs exactly as designed, every cost visible, nothing marked up.

BYOK

Your keys. Their rates. No middleman.

Connect your Anthropic and Perplexity keys. Your provider bills you at published rates. Foundry charges a flat subscription and nothing else.

Anthropic

Perplexity

Named Pipelines

Your methodology, saved.

Name pipelines after what they do. Run the same workflow on new inputs without rebuilding it. Change one stage — everything else stays intact.

∞

Stage combinations

Models

Per-stage, on purpose.

claude-sonnet-4Analysis

perplexity-sonarResearch

claude-haiku-4Verify

Lenses

Context that travels with every run.

A lens carries your methodology, audience, or framework into every stage automatically — no copy-paste, no repetition.

Analytics

Know exactly what you’re spending.

Token count, cost per stage, total per run — for every pipeline you’ve built, so you can optimize based on real data.

Documentation

Up and running
in minutes.

Guides for every part of Foundry — from your first pipeline to model configuration and lens design.

Getting Started

Your first pipeline in 5 minutes

Connect your keys, add stages, write a prompt, and run. That's the whole setup.

Read guide

Pipelines

How stages pass output to each other

Stage sequencing, context flow, and how to design a pipeline for any kind of work.

Read guide

Models

Picking the right model for each stage

Which model fits research, analysis, writing, and verification — and what each choice costs.

Read guide

Lenses

Reusable context across every run

Build a lens once — your methodology, your audience — and it travels with every pipeline that uses it.

Read guide

BYOK

Connect your API keys

How to add your Anthropic and Perplexity keys, and how billing works when Foundry is in the middle.

Read guide

Analytics

Cost tracking and run history

Token usage, stage-by-stage cost, full run history. Download outputs at any point in the pipeline.

Read guide

Pricing

Flat subscription.
Zero markup.

Pay for Foundry once a month. Your AI providers bill you separately at their published rates. We never touch that number.

Starter

^$79 / mo

For individual professionals

✓

Unlimited runs

✓

3 custom pipelines

✓

100 run history

✓

Lens library

✓

BYOK — bring your own keys

Get started