Define a team of specialists. Describe your goal. Watch 20+ AI agents deliver documents, analysis, and decisions — in minutes, not weeks.
Rolling out to a whole team? See the enterprise path →
Specialists run in parallel — each hands off to the next layer. You just write the brief.
Three steps from blank brief to polished deliverables.
Pick from ready-made templates — DevSecOps, Bid Response, Business Plan, Due Diligence — or build a custom team from 100+ specialist roles.
21 templates · 100+ roles
Write a brief, or attach a PDF, Word doc, spreadsheet, or image — agents read your actual documents and even describe charts and screenshots via vision. The Brief Enhancer scores your input and flags missing context before agents start.
Multi-modal briefs · Brief scoring · Webhook trigger
Specialists run in sequence, each building on the previous — asking each other when something is missing, and revising weak work before you see it. A live timeline shows every step. Then reply to the result and the team revises it.
Live progress · Ask for changes · .docx / .pptx export
A single AI prompt gives you one perspective. Forge gives you a coordinated team of specialists — each checking and building on the previous.
Every output carries a confidence score and a critic pass at the end. Per-template run times are on each card below.
Four capabilities that make Forge different from every other AI tool.
Deployable on your own cloud. No vendor lock-in.
Real-time timeline as each specialist completes their work. Confidence scores, decisions, and reasoning all visible. No black box — total transparency into every step.
Describe your goal and have a team assembled for you, or pick from 100+ agent roles yourself — then reply to what comes back and the team revises it, or chain multiple teams into a Workflow Chain that runs end-to-end. Connect web search, Jira, Slack, and webhooks. Set per-agent goals, with conditional routing and auto-chaining on outcomes.
Agents ask each other when something is missing instead of assuming it, weak work is sent back for revision before you see it, and the team can bring in a specialist the brief turned out to need. Attach PDFs, spreadsheets and images to any brief — agents read them directly.
Every completed run teaches the system about your organisation, and you can upload your own material — methodologies, policies, past proposals — for agents to work from. Retrieval understands meaning, so a passage is found when it answers the brief, even in your own vocabulary.
Each template is a pre-built team of specialist agents, pre-wired to hand off work in sequence and produce a complete deliverable.
Business & Finance
Transform your idea into an investor-ready business plan. Agents analyse the market, map competition, model financials, write strategy, and produce a complete plan document.
Legal & Compliance
Review any contract for risk and negotiation opportunities. Agents parse clauses, flag risks, analyse legal exposure, build negotiation strategy, and produce a legal review report.
Strategy
Research a company, assess financial health, identify risks, review legal exposure, and produce an executive recommendation.
Strategy
Analyse the market, build user personas, map competitors, prioritise features with RICE scoring, and produce a product requirements document.
Tech
Pull a feature's acceptance criteria and Definition of Done from Jira or GitHub, locate the related code, audit code quality, assess how much of the DoD the code actually meets, and deliver a Code Quality report plus a DoD Assessment.
Tech
End-to-end autonomous delivery: requirements → architecture → security → IaC → implementation → QA → FinOps → release. Produces documents, code, decisions, and a full CI/CD delivery pack.
Tech
Threat model your system, audit OWASP Top 10, scan dependencies, document pentest findings, and produce a prioritised remediation roadmap.
Tech
Analyse your codebase or API, write full API docs, end-user guides, a structured changelog, and a quality review.
Content & Marketing
End-to-end RFI/RFP/RFQ response pipeline: intake & discovery → document analysis → requirements → capability matching → solution design → commercial → compliance → response writing → quality review. Produces a complete, submission-ready bid document.
All 11 templates included free. No credit card required.
Start for freeDescribe what you need and Forge assembles the team, or pick from 100+ specialist agent roles yourself — and admins can create entirely new agent roles from the UI. Ground runs in your own documents, and an AI advisor checks your team composition before you run.
Don't build your own AI platform — run on Forge. REST API, TypeScript SDK, and a command-line tool. Define agent teams in code, stream into your UI, embed live run views, and give agents persistent memory.
import { ForgeClient } from '@forge/sdk'
const forge = new ForgeClient({
apiKey: process.env.FORGE_APP_KEY,
})
// Submit a run and stream results
const run = await forge.createRun({
team_id: 'devsecops-team-id',
title: 'Q3 Security Audit',
brief: 'Review our Node.js API...',
})
for await (const frame of forge.streamRun(run.id)) {
if (frame.type === 'event')
console.log(frame.data.agent_role, '→', frame.data.title)
if (frame.type === 'done')
console.log('Status:', frame.data.status)
}
const { outputs } = await forge.getRunOutputs(run.id)
// outputs: risk register, remediation plan, report...Create runs, deploy teams, stream events, manage memory, query usage. Scoped API keys, full OpenAPI 3.0 spec, interactive reference.
View OpenAPI spec →Zero-dependency. Node, Deno, Bun, browser. Typed runs, teams, memory, SSE streaming, stream tokens, and a runAndWait() helper.
SDK quick start →npm install -g @forge/cli. Create runs and watch agents work live in your terminal, deploy teams, manage memory. Ideal for CI/CD pipelines.
CLI docs →Define teams in a version-controlled forge.yaml and reconcile with forge iac plan / apply — idempotent, reviewable in PRs, and safe to run from CI.
IaC docs →Consumer apps get:
Everything you need to ship AI features
Issue keys with granular scopes — runs, outputs, teams, memory, usage. Self-serve from the developer portal.
Deploy a full agent team from a JSON manifest via POST /teams. No dashboard — version teams in your repo, deploy in CI.
Short-lived stream tokens let your frontend open an EventSource and show live agent progress — without exposing your key.
Drop a live run timeline into any app with a single <iframe>. Status, events, outputs, and usage — Forge-hosted, zero CSS.
Store per-user and app-wide facts. They are auto-injected into future run briefs — persistent context with no vector DB to run.
forge run create streams agents live in your terminal. Deploy teams, manage memory, script anything. Perfect for CI/CD.
Start free. Upgrade when you need more runs or seats. No hidden fees.
Trial
Get started and explore the platform.
Starter
For small teams running regular pipelines.
Pro
For growing teams that run at scale.
Enterprise
Tailored contracts, SLAs, and volume pricing.
All prices in EUR, billed monthly. No setup fees. Cancel anytime from your billing portal. Limits reset on the 1st of each month.
How is Forge different from ChatGPT or Copilot?
ChatGPT is a single assistant. Forge runs a coordinated pipeline of specialist agents — each with a defined role, context from prior agents, and structured output. The result is a complete deliverable (a 30-page business plan, a full bid response), not a chat reply.
What LLMs does Forge use?
Forge runs on Azure AI Foundry, currently using GPT-4 class models. All inference runs within the EU by default. Enterprise plans can request specific model versions or bring their own Azure subscription.
Is my data secure and private?
Yes. All data is encrypted in transit (TLS 1.3) and at rest (AES-256). Row-level security means no user can access another organisation's data. Agent credentials are stored in a hardware-backed vault, never in code or logs. See our Security page for the full picture.
Can I self-host or deploy on my own cloud?
Yes — the Enterprise plan includes an on-premise or private cloud deployment option. You bring your Azure subscription and we handle the deployment. Your data never leaves your infrastructure.
How long does a run actually take?
Most templates complete in 15–25 minutes. Larger pipelines (DevSecOps with 9 agents, Bid Response with 9 agents) take up to 30 minutes. Independent agents run in parallel to minimise total time.
Can I build custom teams beyond the ready-made templates?
Yes. Pick any combination of 100+ specialist agent roles, set per-agent goals, connect tools (web search, Jira, Slack), and define hand-off conditions. An AI advisor reviews your team composition before you run.
Can I build my own product on top of Forge?
Yes — Forge is designed to be the AI infrastructure layer for your app. Use the REST API, TypeScript SDK, or CLI to create runs, deploy teams from a JSON manifest, stream live progress into your UI, embed a run timeline with a single iframe, and store persistent memory that's auto-injected into runs. Self-serve a scoped API key from the developer portal — no sales call needed.
Tell us what you need — SSO, custom teams, volume pricing, or a security review — and we'll get your organisation set up.
A deep-dive into how leading organisations are deploying AI agent teams to compress weeks of knowledge work into minutes — with real ROI data, architecture details, and a security & compliance overview.