Poetic
by Poetic (formerly Forge)
AI that learns like AI but runs with the precision of code, for regulated back-office processes.
Poetic automates complex, multi-hour back-office processes such as fraud investigations, KYC, disputes and insurance claims by learning a company's SOPs, training videos and expert feedback, then running them deterministically. It targets banks, fintechs and insurers whose compliance-heavy workflows have resisted conventional automation for years.
Poetic, which previously operated as Forge, is a San Francisco company founded by CEO Markie Wagner, a former Google and Waymo machine-learning engineer who earlier ran the AI consultancy Delphi Labs. It emerged publicly in June 2026 with a $50 million Series A at a $500 million valuation, led by Kleiner Perkins with Founders Fund, First Harmonic and OpenAI participating. Its product ingests a customer's standard operating procedures, training videos, historical examples and expert feedback, drafts an automation, and refines it through operator feedback until it is production-ready. Under the hood Poetic uses a proprietary programming language that lets operators describe complex workflows in natural language and compiles them into deterministic execution, which the company summarises as software that 'learns like AI but runs like code'. Automations are described as self-healing: they are maintained as underlying enterprise systems and data change. Poetic says data stays in the customer's environment and is processed in transit with zero retention, including for AI, and it cites SOC 2 Type II, PCI DSS, ISO 27001, HIPAA, GDPR and CCPA compliance, immutable audit logs and role-based access control. Named customers include SoFi, which it says reached 99%+ quality on fraud investigations within five weeks, AIG, at 99%+ accuracy on multi-hour claims processes, and Chime. The company says it hit an eight-figure run rate in 2025 with four employees and converted every pilot to production. Trade press notes that disclosure about the mechanics is thin.
A COO or head of operations at a bank, fintech or insurer with large manual investigation, KYC or claims teams working to strict procedures.
End-to-end automation of multi-hour regulated processes at a claimed 99%+ quality with full procedure adherence and audit logs.
At a Glance
- Category
- Automation & Workflows
- Pricing
- Contact for pricing
- Target Market
- COOs, Chief Compliance Officers, CIOs, Operations Leaders
- Deployment
- Hybrid
- Headquarters
- San Francisco, USA
Key Features
- ✓Procedure learning
Ingests SOPs, training videos, materials and historical examples to draft an automation of an existing business process.
- ✓Expert feedback loop
Operators review and correct outputs, and the system refines the automation until it meets production quality thresholds.
- ✓Deterministic execution language
A proprietary language turns natural-language workflow definitions into code-like deterministic runs, giving consistent, auditable outcomes.
- ✓Self-healing automations
Maintains automations as underlying enterprise systems and data change, reducing breakage compared with brittle scripted automation.
- ✓Zero data retention
Processes data in transit inside the customer's environment with no retention, including for AI, to satisfy regulated-industry requirements.
- ✓Audit and access controls
Immutable audit logs and role-based access controls give compliance teams evidence of how each case was decided.
Capabilities
Use Cases
- •Fraud investigations
SoFi used Poetic to automate fraud investigations, reporting 99%+ quality within five weeks and 100% compliance adherence.
- •Insurance claims processing
AIG automated complex multi-hour claims processes, with Poetic reporting 99%+ accuracy on those end-to-end workflows.
- •KYC reviews
Banks run know-your-customer procedures through Poetic to achieve what the company reports as 100% process adherence across cases.
- •Dispute handling at scale
Financial services teams automate transaction dispute investigations; Poetic cites more than 500,000 dispute cases handled on its platform.
- •Transaction monitoring
Compliance teams automate alert investigations from transaction-monitoring systems, clearing backlogs while keeping a complete audit trail per case.
Ideal For
Best For
- ✓Fraud investigation and auto-decisioning at banks and fintechs
- ✓KYC and transaction-monitoring reviews that must follow procedure exactly
- ✓Card and payment dispute handling at high volume
- ✓Insurance claims and underwriting processes that take hours per case
- ✓Regulated firms that require data to stay in their own environment
Not Ideal For
- ✗Simple, low-variance tasks already handled well by RPA or off-the-shelf workflow tools, where Poetic's enterprise engagement is overkill
- ✗Buyers who need transparent pricing, published documentation and independent reviews; Poetic discloses little about its mechanics and has no public pricing
- ✗Organisations without documented SOPs or subject-matter experts to provide feedback, since the system learns from both
Deployment
Market Analysis
Pros
- ✓Named, demanding references in SoFi, AIG and Chime
- ✓Strong compliance posture for regulated industries, with no data retention
- ✓Deterministic execution suits auditors better than free-form agent output
- ✓Backed by Kleiner Perkins, Founders Fund and OpenAI
Cons
- ✗Trade press flags thin disclosure about how the system actually works
- ✗The 99% accuracy claims are vendor-reported, and TNW questions whether they survive regulator and auditor scrutiny and messy edge cases
- ✗No public pricing, documentation, integrations list or independent user reviews (nothing on Hacker News)
- ✗Very small team relative to its enterprise commitments; the company reported four employees at its 2025 run rate
Pricing
Enterprise
Contact for pricing
- ✓Procedure learning and expert feedback loop
- ✓Self-healing automations
- ✓In-environment processing with zero data retention
- ✓Audit logs and RBAC
Poetic publishes no pricing; engagement starts through hello@poetic.com. Coverage points to large enterprise deals with Fortune 500 financial firms (one claimed $200M in annual fraud savings), so expect contracts scoped per automated process.
Security & Compliance
Connect
Sources
This page was written from 4 sources, 3 on domains other than poetic.com.
Stay Ahead of the Curve
Weekly enterprise AI insights for technology leaders. No spam, no vendor pitches—unsubscribe anytime.
SubscribeRelated Products
Tines
Intelligent workflow platform where employees build AI apps, agents and automations in plain English while IT governs them
Metaview
Agentic recruiting platform with AI agents for sourcing, screening and interview notes
Jack & Jill
AI recruiting agents on both sides of the hire: Jill sources for employers, Jack represents candidates, and they introduce only on mutual fit
Sidetrade Aimie
Autonomous AI agents for order-to-cash: collections calls, cash application and credit risk