top of page
Business & Enterprise
AI Technology
Conversational AI / AI Assistant
Writer
Built an enterprise generative-AI business by turning brand-safe content generation into a governed AI platform, then expanding into enterprise agents and Palmyra models despite OpenAI’s general-purpose scale and hyperscaler distribution.
1
MODEL
BUSINESS MODEL
SaaS
- Founded in 2020 by May Habib and Waseem AlShikh after building Qordoba.
- Started with enterprise writing assistance, terminology management and brand-voice enforcement.
- Built the Palmyra family of enterprise LLMs to control quality, cost and customization.
- Added graph-based RAG, AI guardrails, no-code development tools and agent-building capabilities.
- Sold platform subscriptions and enterprise agreements to companies including Uber, Intuit, Mars, Vanguard, KPMG and L’Oréal.
- Raised $200 million at a $1.9 billion valuation in 2024 after earlier revenue growth of 10x.
1. Start with one high-frequency enterprise workflow where generic AI creates quality or compliance risk.
2. Build governance, brand rules and data grounding before expanding into broader AI applications.
3. Use a proprietary or controllable model layer only when it creates measurable quality, cost or customization advantage.
4. Expand from content generation into workflows where the same enterprise context and guardrails apply.
5. Price on platform value and enterprise deployment rather than only words generated.
Avoid: competing on raw model capability alone; enterprises need accuracy, governance and workflow integration more than another general chatbot.
DISTRIBUTION MODEL
Writer began with AI writing and terminology governance for enterprise teams, then expanded into Palmyra LLMs, graph-based RAG, guardrails, no-code development tools and AI agents. Revenue grew 10x in two years, net revenue retention exceeded 150% in 2023 and later reached over 200%, while customers expanded from writing workflows to enterprise-wide AI applications.
HOW THEY BUILT IT
HOW TO ARCHITECT IT
Direct Sales, Enterprise Sales, Product-Led Growth, Content Marketing
HOW THEY OPERATIONALIZED
HOW TO REPLICATE WHAT WORKED
dm
- Enterprise sales targeted marketing, customer support, HR and operations leaders with content and workflow pain.
- Self-serve and team plans let smaller groups test writing assistance before enterprise rollout.
- Enterprise pilots demonstrated brand governance, terminology accuracy and workflow ROI.
- Content on enterprise AI, LLMs, RAG and agents attracted AI-buying teams.
- Customer logos such as Uber, Intuit, Mars and Vanguard reinforced enterprise credibility.
What worked: Enter through a narrow, high-frequency workflow, prove governance and quality, then expand into broader enterprise AI applications.
Trap if copied blindly: Enterprise AI distribution fails when the product is a thin wrapper over a general model; the cost is low differentiation, price pressure and weak retention once customers can prompt a foundation model directly.
| PATTERNS OF THIS MODEL
PATTERNS IN ENTERPRISE GENERATIVE-AI PLATFORMS:
1. Governance creates enterprise willingness to pay. Companies pay for brand control, data grounding, permissions and auditability, not merely generation.
2. Workflow depth reduces commoditization. A writing tool becomes harder to replace when it connects to enterprise data, approvals and systems of record.
3. Model ownership increases customization control. Proprietary models allow tuning for industry language, cost and performance.
THE WEAKNESS: Building proprietary models increases capital intensity and creates pressure to match frontier labs on capability.
What companies with this model reveal
| OPPORTUNITY INTELLIGENCE
GOLDMINE 1 — ENTERPRISE AI GOVERNANCE LAYER. Companies need brand rules, source grounding, permissions and audit trails; governance can become the durable layer above interchangeable models.
GOLDMINE 2 — AGENT OPERATIONS PLATFORM. Enterprises need agents that execute workflows across systems with approvals, evaluation and observability; this is more defensible than standalone content generation.
GOLDMINE 3 — DOMAIN-TUNED SMALL MODELS. Smaller models tuned for legal, financial or industry language can reduce inference cost and improve accuracy for narrow enterprise tasks.
THE PIT — MODEL-LAYER CAPITAL INTENSITY. Proprietary LLMs require ongoing training, evaluation and compute investment, which can strain margins.
THE SECOND PIT — WRAPPER COMMODITIZATION. If the product only prompts a foundation model, customers may switch to the model vendor or build internally.
MOVE WITH CAUTION: Do not deploy enterprise agents without source grounding, permission controls and human approval for consequential actions; inaccurate automation can create compliance and operational risk.
Untapped Business Model / Gaps / Goldmines / Pits
Patterns & Insights
2
MARKET
mkt mt es
Writer competed against ChatGPT Enterprise, Jasper, Grammarly Business and hyperscaler AI platforms. Its differentiation was enterprise governance: brand voice, terminology, data grounding, guardrails and workflow deployment, rather than general-purpose chat or consumer writing assistance.
MARKET TYPE
Emerging Market
ENTRY STRATEGY
Greenfield Entry
WHY THEY WON
Enterprise generative AI was an emerging market with rapidly changing model capability and unsettled buying criteria. Writer exploited a structural gap: enterprises wanted AI productivity but also needed accuracy, brand consistency and compliance. Its positioning changed the buying criterion from “best general chatbot” to “most governable AI platform for enterprise work.”
EXECUTION
Writer created an enterprise generative-AI platform rather than acquiring an incumbent or building solely on another model provider. Its critical bet was that enterprises would pay for AI that respected brand, terminology and data governance. This was viable because Writer had enterprise content expertise from Qordoba and could package LLMs with workflow controls; the approach fails when governance alone does not improve output quality or business outcomes.
FOOTHOLD STRATEGY
Wedge Strategy
Writer first gained traction through a wedge of marketing, customer support and content teams that needed on-brand, accurate writing at scale. These users were reachable through enterprise content workflows, could pay from marketing or operations budgets, and experienced immediate quality and consistency gains. The path outward was enterprise-wide AI apps, Palmyra models, RAG and agents.
GROWTH CAMPAIGN
CAMPAIGNS THAT WORKED
SO WHAT: Palmyra gave Writer a strategic reason to be more than a wrapper: it could control model quality, cost and customization for enterprise use cases.
KEY LEARNING: Owning a component of the AI stack can create differentiation, but only if it improves customer outcomes rather than serving as a marketing claim.
HOW TO USE: If building on foundation models, identify the layer—data, evaluation, workflow or governance—where you can create durable customer value before investing in model development.
KEY LEARNING
THE STANDARD: In an emerging AI market, the winner defines the enterprise buying criteria before the category settles.
RULE 1 — SELL GOVERNED OUTCOMES, NOT RAW GENERATION: Enterprises need accuracy, permissions and auditability.
RULE 2 — START WITH A REPEATABLE WORKFLOW: Content, support and operations create frequent usage and measurable ROI.
RULE 3 — EXPAND INTO AGENTS CAREFULLY: Agents should operate within enterprise data, permissions and approval controls.
RULE 4 — KNOW WHEN ENTRY FAILS: If the foundation model alone solves the workflow, the application layer must add workflow, governance or integration value.
MARKET TYPE: Emerging Market (category).
gc
Market Context
| MARKET INTELLIGENCE
PATTERNS IN ENTERPRISE AI PLATFORM GROWTH:
1. Enterprise pilots create expansion. A successful departmental use case becomes the reference for broader deployment.
2. Governance content attracts AI buyers. Enterprises search for RAG, guardrails, model choice and compliance before selecting a platform.
3. Customer logos reduce adoption risk. Recognized enterprises signal that the platform can handle security, scale and procurement.
4. THE WEAKNESS: Enterprise AI adoption can stall when ROI is difficult to measure or when outputs require extensive human review.
| MARKET ENTRY PLAYBOOK
THE STANDARD: Greenfield entry is appropriate when existing AI tools do not address enterprise quality, governance or workflow requirements.
RULE 1 — TARGET A HIGH-FREQUENCY WORKFLOW: Choose a task where AI output errors create business cost.
RULE 2 — BUILD GOVERNANCE INTO THE PRODUCT: Brand rules, data grounding and permissions should be core, not add-ons.
RULE 3 — PROVE ROI BEFORE PLATFORM EXPANSION: Show productivity, quality or compliance gains in one department first.
RULE 4 — REJECT WHEN THE MODEL VENDOR OWNS THE WORKFLOW: If the foundation model can solve the task natively, build a vertical or data-advantaged layer instead.
How to enter
| FOOTHOLD STRATEGY PLAYBOOK
THE STANDARD: A wedge strategy succeeds when one workflow creates adoption, data and trust for broader enterprise AI.
RULE 1 — START WITH CONTENT AND COMMUNICATIONS: These workflows are frequent, visible and easy to evaluate.
RULE 2 — CAPTURE ENTERPRISE CONTEXT: Terminology, brand rules and approved sources make output more valuable over time.
RULE 3 — EXPAND INTO ADJACENT DEPARTMENTS: Support, HR, sales and operations can reuse the same governance layer.
RULE 4 — AVOID GENERAL-PURPOSE POSITIONING TOO EARLY: Broad AI claims invite direct comparison with frontier-model vendors.
How to get the first strong position
MARKET PATTERNS & PLAYBOOK
3
MONEY
money rev pri
Writer expanded from writing subscriptions into enterprise platform agreements, Palmyra model usage, RAG applications and agents. Its platform fee ranged from $30,000 to $1 million based on generated-word volume, while enterprise customers expanded across departments and use cases; net revenue retention exceeded 150% in 2023 and later surpassed 200%.
REVENUE MODEL
Subscription, Usage-Based, Contract Revenue
PRICING MODEL
Tiered Pricing, Usage-Based Pricing, Custom Enterprise Pricing
WHY THEY WON
Customers pay recurring platform fees for Writer’s enterprise AI platform, with usage influenced by generated words, model consumption and deployed applications. Enterprise agreements add committed spend, governance and support; revenue expands through seats, departments, use cases and agent workloads.
Writer uses team plans for smaller groups and custom enterprise pricing for platform deployments. Before April 2025, it offered a free Developer Plan, Team Plan and Enterprise Plan; enterprise platform fees ranged from $30,000 to $1 million annually based on generated-word volume.
TARGET AUDIENCE
Marketing and content teams (Team/Enterprise; want on-brand content at scale); Customer support teams (Enterprise; want accurate, governed responses); Enterprise AI and IT teams (Enterprise; want secure RAG, guardrails and agent deployment); Regulated industries (Enterprise; want compliance, auditability and data control).
CUSTOMER BUYING BEHAVIOUR
A marketing, support or operations leader identifies inconsistent content, compliance risk or manual work, then evaluates Writer through a pilot. IT and security review data handling and model controls before enterprise rollout; expansion follows as teams deploy additional AI applications and agents.
| PRICING INTELLIGENCE
What makes this model effective & make customers pay
Enterprises are not buying AI writing; they are buying governed productivity at scale.
RULE 1 — PRICE PLATFORM VALUE, NOT ONLY OUTPUT: Brand governance, data grounding and workflow deployment justify enterprise fees.
RULE 2 — EXPAND THROUGH USE CASES: Each new department or agent increases platform value.
RULE 3 — MAKE USAGE TRANSPARENT: Word or model consumption should map to customer value without creating bill anxiety.
RULE 4 — WATCH MODEL-COST PRESSURE: If inference costs rise faster than platform fees, enterprise gross margin suffers.
The buyer is purchasing governed enterprise productivity, proven by platform fees, active teams and expanding AI use cases.
PRICE & REVENUE
| Revenue Risk - The biggest threat to revenue stability
Foundation-model commoditization: OpenAI, Anthropic and Google can bundle writing, RAG and agent capabilities into enterprise offerings.
Model-cost escalation: Palmyra development and inference costs can pressure gross margin.
Enterprise ROI skepticism: If productivity gains are hard to measure, renewal and expansion slow.
AI-output risk: Inaccurate, biased or non-compliant content can create enterprise trust and legal exposure.
Agent-execution risk: Autonomous workflows that act across systems require strong permissions and auditability.
Competitive bundling: Microsoft Copilot, Google Workspace AI and Salesforce Einstein can bundle AI into software enterprises already use.
Writer raised $200 million at a $1.9 billion valuation in 2024 and has raised over $326 million total; it is private and does not publicly disclose current ARR.
Where the model can break
4
MOTION
motion ge cs
Writer built an enterprise sales motion around marketing, IT, security and business leaders. Enterprise deals required data-grounding reviews, security approval, workflow design and multi-team rollout; customers such as Uber, Intuit, Mars and Vanguard validated enterprise readiness.
GROWTH EXPANSION MODEL
Land & Expand, Platform Expansion
HOW THEY EXPAND
Writer landed with enterprise writing assistance, expanded within accounts through terminology, RAG and guardrails, then added Palmyra LLMs, no-code AI builders and agents. Each expansion reused the same enterprise context, governance layer and customer relationships.
COMPETITIVE STRATEGY
Differentiation
HOW THEY COMPETE
Against OpenAI and hyperscalers, Writer differentiated through enterprise governance, Palmyra models and workflow deployment. Against Jasper and Grammarly, it moved upmarket from writing assistance into enterprise AI applications and agents.
GROWTH ENGINE
GTM
Content Flywheel
A department adopts Writer for governed content → output quality and brand consistency improve → adjacent teams adopt the platform → enterprise data and workflows expand → AI agents increase platform usage. The loop compounds when governance, enterprise context and workflow integration make Writer harder to replace.
Writer runs an enterprise-sales-led GTM with product-led entry for teams. Self-serve and team plans create initial adoption; enterprise sales expands into multi-department deployments, security reviews and custom AI applications. The primary motion is enterprise sales, with product-led usage supporting expansion. The transferable implication is that enterprise AI needs a narrow proof point before broad platform adoption.
SUSTAINING MOATS
High Customer Lock-In
moat
The moat is enterprise context plus governed workflow dependence: terminology, brand rules, approved sources, permissions and agent workflows accumulate inside Writer, making replacement disruptive.
| MOAT INTELLIGENCE
THE STANDARD: Enterprise-AI moats last when proprietary context, governance and workflow integration reinforce one another.
RULE 1 — ENTERPRISE DATA CREATES OUTPUT ADVANTAGE: Grounded responses improve as companies add approved sources and terminology.
RULE 2 — GOVERNANCE CREATES ENTERPRISE TRUST: Permissions, auditability and brand controls reduce adoption risk.
RULE 3 — WORKFLOW DEPENDENCE CREATES RETENTION: Once agents operate across systems, replacement requires redesigning business processes.
THE SIGNAL: More than 250 enterprise customers, over 200% net revenue retention and a $1.9 billion valuation indicate enterprise expansion; the moat does not protect against frontier-model commoditization, weak ROI evidence or superior enterprise bundles.
Why this company remains defensible
ARR & TAKEAWAY
ARR Journey - what to do at each stage
PRE-$1M ARR — PROVE GOVERNED ENTERPRISE WRITING. After its 2020 founding, Writer needed to prove that enterprises would pay for AI writing with brand and terminology control.
$1–5M ARR — WIN CONTENT AND SUPPORT TEAMS. Enterprise writing workflows created the first repeatable adoption.
$5–10M ARR — EXPAND ACROSS DEPARTMENTS. Terminology, brand voice and workflow templates increased account value.
$10–50M ARR — ADD ENTERPRISE AI GOVERNANCE. RAG, guardrails and data grounding expanded Writer beyond content generation.
$50–100M ARR — BUILD PALMYRA AND AI APPLICATIONS. Proprietary models and no-code tools deepened platform control.
$100M+ ARR — SCALE AGENTIC ENTERPRISE AI. Writer raised $200 million at a $1.9 billion valuation in 2024 and positioned the platform around enterprise agents; revenue had grown 10x over the prior two years.
Rule: Prove governed value in one workflow, then expand into enterprise-wide AI applications.
COPY PLAYBOOK : What Worked → What Failed → What to Replicate → What to Avoid
THE STANDARD: Solve a governed enterprise workflow, build proprietary context and controls, then expand into broader AI applications.
SEQUENCE:
1. First, choose a workflow where generic AI creates quality or compliance risk; prerequisite: the customer can measure improvement.
2. Next, build governance and enterprise-data grounding; gate: output becomes materially more accurate, consistent or compliant.
3. Then, expand into agents and adjacent departments only after the first workflow proves ROI; gate: the expansion reuses enterprise context and increases retention.
WORKED: Writer used enterprise writing to create adoption, then expanded into Palmyra models, RAG and agents.
CAUTION: Do not build proprietary models without a clear quality, cost or customization advantage. Also avoid deploying agents without permissions, grounding and human oversight.
bottom of page