Writer
by Writer, Inc.
Full-stack enterprise AI built on models Writer trains and hosts itself
Writer is a full-stack generative AI platform for large enterprises that trains and hosts its own Palmyra language models rather than reselling OpenAI or Anthropic. It pairs those models with a Knowledge Graph over live business data and a no-code agent builder, so regulated organisations can deploy AI agents without customer data leaving the vendor's stack.
Writer is a full-stack enterprise generative AI platform founded in August 2020 by May Habib and Waseem AlShikh, the pair behind the earlier localisation company Qordoba, and headquartered in San Francisco. Its distinguishing decision is vertical integration: rather than reselling access to third-party frontier models, Writer trains and hosts its own Palmyra model family — launched February 2023 as Palmyra-Small (128M parameters), Palmyra-Base (5B) and Palmyra-Large (20B), and since extended with domain-specific models including Palmyra-Med for healthcare and Palmyra-Fin for financial services. Customer data therefore never routes through OpenAI or Anthropic APIs, which is the argument that wins Writer its regulated-industry deals. Three components sit on top of the models. Knowledge Graph, introduced May 2023, connects live business systems so agent output is grounded in the company's own data rather than generic web text. AI Studio is a no-code builder for custom agents that can search a CRM, reference brand guidelines, check compliance rules and execute multi-step workflows autonomously. A library of more than 100 prebuilt agents covers common marketing, sales, support and legal tasks. The company also maintains Writer Framework, an Apache-2.0 Python project on GitHub with roughly 1.4k stars for building AI applications with a visual UI editor and a Python backend. Writer reported $47M ARR in November 2024 on 194% year-over-year growth, was valued at about $1.98 billion in June 2025 on $326M raised, and serves 300-plus enterprise customers including Uber, Vanguard, Accenture, Qualcomm, L'Oréal and Prudential. It holds ISO/IEC 27001, 27701 and 42001 certifications alongside SOC 2 Type II including HIPAA.
The enterprise AI or content operations leader at a regulated Global 2000 company — financial services, healthcare, pharma — who needs agents grounded in internal data and cannot send that data through a third-party model API.
Custom AI agents grounded in live company data, running on models the vendor owns end to end, with the ISO 42001 and SOC 2 Type II paper trail procurement will ask for.
At a Glance
- Category
- Enterprise Platform
- Pricing
- Subscription, Usage-based, Contact for pricing
- Target Market
- CIOs, CTOs, Enterprise Content Teams, Marketing Leaders, Compliance and Risk Officers
- Deployment
- Cloud-first, Self-hosted, Hybrid
- Founded
- 2020
- Headquarters
- San Francisco, United States
- Team Size
- 500+
- Customers
- 300+ enterprise customers including Uber, Vanguard, Accenture, Qualcomm, L'Oréal, Spotify and Prudential (Sacra)
Key Features
- ✓Palmyra proprietary LLM family
Writer trains and hosts its own models, so customer data never routes through OpenAI or Anthropic APIs — the core procurement argument in regulated industries.
- ✓Knowledge Graph
Connects live business data sources to the models so agent answers reflect current company information rather than a stale document upload.
- ✓AI Studio no-code agent builder
Business users assemble custom agents that query systems, apply brand and compliance rules, and run multi-step workflows without writing code.
- ✓Domain-tuned models
Palmyra-Med and Palmyra-Fin are trained for healthcare and financial services, where general models routinely miss regulatory and terminology nuance.
- ✓Brand and compliance governance
Style, terminology and regulatory rules are enforced at generation time, which reviewers cite as the platform's most consistently praised capability.
- ✓Self-hosted and on-premise deployment
Because Writer owns the model weights it can deploy them inside a customer environment for full data isolation, which model resellers cannot offer.
- ✓Writer Framework (open source)
An Apache-2.0 Python project with a visual UI editor for building AI applications, roughly 1.4k GitHub stars, maintained by Writer.
Capabilities
Use Cases
- •Financial research and advisor enablement
Vanguard deployed custom Writer agents across investment research, financial advisor enablement, marketing and customer communications under existing compliance controls.
- •Support ticket triage
Uber's support organisation uses Writer agents to classify Jira and Slack requests and assess urgency, routing work before a human reads it.
- •Legal and trademark documentation
Qualcomm's legal team uses the platform to manage trademark documentation, applying consistent terminology across filings that would otherwise drift.
- •Governed marketing content production
Marketing teams generate campaign copy that automatically conforms to brand voice and approved claims, cutting review cycles with legal and compliance.
- •Knowledge-grounded internal answers
Employees query the Knowledge Graph for policy and product answers sourced from connected systems rather than searching several tools by hand.
Ideal For
Best For
- ✓Regulated enterprises that must keep data off third-party model APIs while still deploying generative AI
- ✓Enforcing brand voice, terminology and compliance rules consistently across marketing, support and sales content at scale
- ✓Building department-specific agents without engineering headcount, using the no-code AI Studio and 100+ prebuilt agents
- ✓Grounding agent output in live business systems through the Knowledge Graph rather than static prompt stuffing
- ✓Domain work where a specialised model helps — Palmyra-Med for clinical content, Palmyra-Fin for financial services
Not Ideal For
- ✗Individual creators and small teams wanting low-cost copywriting — the Starter tier caps at 20 users and reviewers consistently call the pricing prohibitive below mid-market
- ✗Teams that need value this quarter: reviewers describe a 2-4 week minimum for Knowledge Graph setup and agent configuration, with full rollouts commonly running months of discovery calls and technical planning
- ✗Organisations wanting best-of-breed model switching — Writer's advantage is that it owns the model stack, which is also a lock-in, and critics describe the end-to-end design as a walled garden that resists integration with existing tooling
- ✗High-volume content shops that expect uncapped generation, since Starter meters words per user per month
Integrations
Deployment
Market & Ratings
300+ enterprise customers including Uber, Vanguard, Accenture, Qualcomm, L'Oréal, Spotify and Prudential (Sacra)
Market Analysis
Pros
- ✓Owning the model stack is a genuine architectural differentiator, not marketing — it is what makes self-hosted deployment and the no-third-party-API guarantee possible at all
- ✓Reviewers rate brand and style governance as the strongest capability, with output described as more structured, accurate and business-ready than general-purpose assistants
- ✓Unusually complete compliance posture for a model developer: ISO/IEC 27001, 27701 and 42001 plus SOC 2 Type II including HIPAA
- ✓Real, named production deployments at Vanguard, Uber, Accenture and Qualcomm rather than logo-wall pilots
Cons
- ✗Reviewers report the advanced surface — custom agents, Knowledge Graph, granular governance — feels overwhelming, and describe a steep learning curve before value appears
- ✗Not plug-and-play: 2-4 weeks minimum for setup and configuration, and critics describe rollouts consuming months of discovery calls and technical planning meetings
- ✗Pricing is opaque at the Enterprise tier and widely called prohibitive for small and mid-sized businesses; independent estimates of the real per-seat cost disagree by more than 2x
- ✗Some users find generated suggestions strict or less creative — the flip side of the governance that regulated buyers are paying for
- ✗The end-to-end design is criticised as a walled garden that resists integrating with tooling a company already runs
- ✗Word-count metering per user per month is too restrictive for teams that want AI to draft the bulk of their content
Pricing
Starter
From $29/mo
- ✓$29/user/mo annual or $39/user/mo monthly
- ✓Up to 20 users per workspace
- ✓100+ prebuilt agents
- ✓Up to 5 custom agents in AI Studio
- ✓Basic Knowledge Graph with limited connectors
- ✓200K words per user per month
Enterprise
Contact for pricing
- ✓Unlimited users with org-wide roles and permissions
- ✓Full custom agent fleets
- ✓Complete Knowledge Graph with all data connectors
- ✓Higher API rate limits and advanced controls
- ✓Expert onboarding and priority support
Self-hosted / On-premise
Contact for pricing
- ✓Deploy Writer-built models inside your environment
- ✓Full data isolation
- ✓Custom model tuning
There is no public pricing page on writer.com — the site offers demo requests, and Enterprise quotes are seat-plus-usage and negotiated. Independent reviews put Starter at $29/user/month annual or $39 monthly, capped at 20 users with 200K words per user per month, though one review reported a $18/user Team tier and another cited a much lower 30K-word cap, so confirm the current metering in writing before signing. API access is metered separately at per-million-token rates starting around $7.50 for text models, with vision and video billed per image or per second, and Knowledge Graph storage and data extraction metered on top. The cost that surprises buyers is not the licence: reviewers consistently report 2-4 weeks minimum for Knowledge Graph setup, agent configuration and training, with full enterprise rollouts running months.
Security & Compliance
Connect
Sources
This page was written from 7 sources, 7 on domains other than writer.com.
Stay Ahead of the Curve
Weekly enterprise AI insights for technology leaders. No spam, no vendor pitches—unsubscribe anytime.
SubscribeRelated Products
Pigment
AI-native integrated business planning with Modeler and Analyst agents, AI-generated planning apps and an MCP server
Red Hat AI
Run any model on any accelerator across hybrid cloud, with safety evidence and cost attribution built in
Genesys Cloud
Agentic orchestration for customer experience — AI agents, employees and systems on one governed platform
VMware Tanzu Platform
Production AI agents inside your own private cloud, with a deny-by-default runtime