Narration 
FAQ 
Précis AI Wants Policy Wonks
BlogShare
Key Points- Précis AI has launched a beta for Willard, an advocacy AI aimed at improving accuracy in policy work by sourcing answers directly from verifiable public records.
- The AI landscape faces challenges, with general-purpose AIs like Claude and ChatGPT generating error-riddled policy content; the AI Index shows significant hallucination rates in leading models.
- Willard is designed to utilize the largest database of federal documents, ensuring all information provided is grounded in official records, enhancing reliability in legislative processes.
0:00
0:00
Précis AI Wants Policy Wonks
Opens Beta for Willard, the Policy-Native AI
WASHINGTON, D.C., (Sept. 17, 2026) — To help address AI accuracy in policy work, Précis AI today opened a beta for Willard, a federal advocacy AI that provides direct, sourced answers drawn from verifiable public records.
As Washington debates how to rein in powerful AIs, another risk is loose in the policy process: AIs like Claude and ChatGPT do policy work and produce fluent, confident, footnoted answers that are sometimes simply wrong.
But AI is already reshaping policy work in Congress. POLITICO reported last month that the House Office of Legislative Counsel is being swamped by error-riddled AI-generated legislation. In that reporting, the former House legislative counsel warned that using error-prone AIs for statutory language creates “the highest risk at the apex of failure.”
And the numbers back that up. Stanford’s 2026 AI Index found hallucination rates ranging from 22% to 94% across 26 leading models on a new accuracy benchmark — with many of the most capable models near the top of that range. Suprmind’s August 2026 hallucination rate tracking shows the newest flagships from OpenAI, Anthropic and Google hallucinate more than 60% of the time they don’t know the answer to a query, while OpenAI’s latest model has inaccuracies more than 90% of the time.
“Speculation that AI will destroy humanity rules the airwaves, but the pressing problem is a confident, cited AI generation that’s just false,” said Précis AI CEO David Fuscus. “Willard is built to dramatically increase accuracy by directly accessing warehouses of government data.”
General-purpose AIs generate mainly from training data and web sources; they aren’t built to ground every answer in official government documents. Willard is grounded in the world’s largest database of AI-accessible federal documents and engineered to retrieve, cite, and ground every answer in these public records.
Beta testers gain early access, with a direct line to the product team to further customize Willard for policy. “Getting in early isn’t about access. It’s about shaping how the AI works,” continued Fuscus.
Qualified professionals can sign up at precisai.com/Willard.
About Précis AI
Précis AI builds industry-specific AI tools. Précis Public Relations has users in 14 countries. Willard is a policy-focused AI designed to help, analyze and act on federal information for more accurate work. Details at PrecisAI.com.
Media Contact
Amelia McKee
(202) 289-4001
Frequently Asked Questions
01What is Willard?
Willard is a federal advocacy AI created by Précis AI, designed to provide accurate, sourced answers directly from verifiable public records.
02Why was Willard developed?
Willard was developed to address the high rates of inaccuracies produced by general-purpose AIs in policy work, which can lead to flawed legislation.
03How does Willard ensure accuracy?
Willard is built to access and cite the world’s largest database of AI-accessible federal documents, grounding every answer in official government sources.
04Who can sign up for the Willard beta?
Qualified professionals can sign up for early access to the Willard beta at precisai.com/Willard.
05What is the significance of AI in the policy-making process?
AI is increasingly shaping policy work in Congress, but reliance on error-prone AIs poses risks, highlighting the need for accurate tools like Willard.
