By Automated InsightsLearning Objectives
- Understand Automated Insights' role as the early NLG (natural language generation) pioneer
- Identify the Wordsmith workflow — data ingest, narrative templates, automated production
- Evaluate when Wordsmith fits versus modern LLM-based generation or general AI writing tools
What Is Automated Insights?
Automated Insights is the natural-language-generation (NLG) pioneer. Founded in 2007 in Durham, North Carolina, the company built Wordsmith — the platform that automatically writes data-driven narratives from structured datasets at scale. Wordsmith is most-known for the Associated Press partnership beginning in 2014, where it generated thousands of corporate-earnings recap stories per quarter, freeing AP human reporters for higher-value journalism. Other major customers include Yahoo Sports (fantasy football recaps), Allstate (claims summaries), and dozens of Fortune 500 enterprise customers.
The company was acquired by Vista Equity Partners in 2015 and subsequently absorbed into Stats Perform (formerly STATS LLC), where Wordsmith continues to operate as a sports + enterprise NLG platform. The pre-LLM-era NLG approach — template-driven, deterministic, structured-data-input — has been challenged by modern LLM-based generation (ChatGPT, Claude) but retains real advantages in deterministic accuracy, regulatory compliance, and high-throughput production.
💡Key Concept
Template-driven NLG vs. LLM-based generation: Template-driven NLG (Wordsmith) starts with structured data and a human-authored narrative template, deterministically filling in the template with data-derived language choices. Outputs are predictable, accurate, and identical for identical inputs — which matters for regulatory reporting, financial filings, and content where hallucination is unacceptable. LLM-based generation (ChatGPT, Claude) is much more flexible and creative but introduces hallucination risk and non-determinism. Each approach has its place; the boundary is shifting as LLMs improve.
✅Tip
Visit Automated Insights: automatedinsights.com — enterprise sales process via Stats Perform; pricing custom per-customer.
Pricing & Access
Wordsmith uses enterprise per-customer pricing tied to volume and use-case complexity.
- Template authoring + lower-volume generation
- Single-tenant workspace
- High-volume production
- API access
- Custom integrations
- SLA
- Sports-specific NLG (game recaps, fantasy)
- League data integration
- White-label content
- Custom NLG models
- Bespoke template development
- Dedicated implementation
For most enterprise customers, Wordsmith Enterprise is the primary tier with custom pricing tied to story volume, integration complexity, and SLA requirements.
Core Capabilities
Wordsmith Template Authoring
The flagship workflow. Authors build narrative templates that combine static text with data-driven variations — " reported for the quarter, the analyst consensus of ." The template includes branching logic for different data conditions ("if revenue beats by 5%+, use 'crushed'; if revenue misses by 10%+, use 'disappointed'"). Templates are author-controlled — no LLM hallucination risk.
Data-Driven Narrative Generation
Wordsmith ingests structured data (CSV, JSON, database) and applies the template to each data row, generating one narrative per row. For Associated Press, this means thousands of company-earnings recaps per reporting season; for Yahoo Sports, millions of personalized fantasy-football recaps per week; for enterprise customers, similarly high-volume routine reporting.
High-Throughput Production
The platform was built for scale — Yahoo's fantasy football integration generates millions of personalized recaps per Sunday afternoon, well beyond what any human-driven workflow could achieve. The deterministic template-driven approach scales linearly with compute.
Brand-Voice Consistency
Because templates are author-controlled, brand voice is consistent across all generated content. There's no LLM drift, no unexpected tone shifts, no creative liberties that don't match brand guidelines. For regulated industries (financial reporting, insurance claims), this consistency is a meaningful advantage.
Enterprise Integration
Wordsmith integrates with enterprise data warehouses, BI tools (Tableau, Power BI), CRM systems, and content management systems. Generated narratives can flow into emails, dashboards, reports, or web pages automatically.
Stats Perform Sports Integration
Following the Vista / Stats Perform acquisition, Wordsmith integrated tightly with Stats Perform's sports data feeds. Wordsmith now powers automated game recaps, player analyses, and fantasy commentary for major sports leagues and media partners — one of its largest current production deployments.
Strengths
- Deterministic accuracy — no hallucination risk, identical inputs produce identical outputs
- High-throughput production — millions of personalized stories per day in some deployments
- Brand-voice consistency across all generated content
- Regulatory compliance suitable for financial reporting, insurance, healthcare narrative requirements
- Pre-LLM-era pioneer with battle-tested production workflows
- Associated Press partnership validated the approach at journalism scale
- Stats Perform sports integration powers major league data narratives
- Mature enterprise integration with data warehouses, BI tools, CRM, CMS
Limitations & Considerations
- Less creative + flexible than modern LLM-based generation (ChatGPT, Claude)
- Template authoring requires upfront engineering investment per use case
- Lower variation in outputs versus LLMs (though that's also the strength)
- Acquired into Stats Perform — strategic priority lower than independent days
- Modern LLM competition — for many use cases, GPT-4-class LLMs now match or exceed Wordsmith quality with less engineering
- Smaller mindshare than ChatGPT or Claude in 2026 — though many large enterprises remain on Wordsmith for compliance
- Subscription model can be cost-prohibitive for smaller customers versus pay-per-API-call LLMs
Best Use Cases
| Use Case | Why Wordsmith Fits | Caveat |
|---|---|---|
| Financial earnings recaps (AP-style) | Deterministic accuracy + high throughput | Template engineering upfront |
| Fantasy sports + game recaps | Stats Perform integration + millions of personalizations | Sports-specific deployment |
| Regulatory financial reporting | No hallucination + audit-defensible templates | Less creative than LLMs |
| Insurance claims summaries | Brand-voice consistency + structured data input | Smaller volumes don't justify setup |
| High-volume routine enterprise reporting | Linear scale at predictable cost | Modern LLMs may match quality |
When to choose alternatives:
- Creative + flexible writing → ChatGPT or Claude
- Lower-volume / one-off generation → general LLMs (cheaper for ad-hoc work)
- Fiction writing → Sudowrite (purpose-built for fiction)
- Marketing-copy generation → Jasper or Copy.ai (marketing-specific tools)
- Conversational + interactive content → general LLMs (Wordsmith is one-shot template-driven)
Key Takeaways
- Automated Insights is the natural-language-generation pioneer — its Wordsmith platform automatically writes data-driven narratives from structured datasets at thousands of stories per minute
- The Associated Press partnership beginning in 2014 validated NLG at journalism scale; Wordsmith now generates corporate earnings recaps, sports analyses, fantasy football commentary, and enterprise reporting at production volume
- Acquired by Vista Equity Partners in 2015 and absorbed into Stats Perform; continues to operate as a sports + enterprise NLG platform with deep Stats Perform sports-data integration
- Template-driven approach is deterministic and accurate (no hallucination), scales linearly, and maintains brand-voice consistency — meaningful advantages versus modern LLM-based generation for regulated and high-throughput use cases
- Best fit for high-volume routine reporting where determinism and accuracy matter (financial recaps, sports analyses, regulatory filings); for creative + flexible writing use ChatGPT or Claude, for fiction-specific work use Sudowrite
