Chat(k) v4.3

AI Retirement Plan Assistant

Chat(k)

Practitioner-level retirement plan guidance for Senior TPAs, ERISA attorneys, and plan advisors. Powered by a curated knowledge base of 1,521 regulatory entries.

Contribution Limits

2025/2026 limits for 401(k), 403(b), IRA, catch-up, and annual additions

LTPT Employee Rules

Long-term part-time eligibility requirements for 401(k) and 403(b) plans

EPCRS Corrections

SCP, VCP, and Audit CAP correction programs for operational failures

All responses are for educational purposes only and are not legal or tax advice. Always consult qualified professionals for specific plan decisions.

Chat(k) may produce inaccurate information. Verify all guidance with authoritative sources.

Chat(k) is an AI-powered retirement plan assistant built for practitioners. It draws from a curated database of 1,521 regulatory knowledge entries covering 401(k), 403(b), SIMPLE IRA, SEP IRA, EPCRS corrections, SECURE 2.0 provisions, and more.

Current Limits & Thresholds

Up-to-date IRS contribution limits, HCE thresholds, and key dollar amounts for 2025-2026.

SECURE 2.0 Guidance

Comprehensive coverage of SECURE 2.0 provisions including LTPT rules, Roth catch-up, and emergency savings.

EPCRS Corrections

SCP, VCP, and Audit CAP guidance with correction methods and procedures.

Multi-Model AI

Choose between Claude, GPT-5.2, and Gemini. GPT-5.2 includes web search for real-time information.

Select Your AI Model
Choose from Claude (Opus 4.5), GPT-5.2 (with web search), or Gemini 2.0 Flash in the sidebar.

Ask Your Question
Type a retirement plan question in the chat box. Use specific terms like "safe harbor", "LTPT", or "402(g)" for best results.

Review the Response
Each answer begins with an educational disclaimer. Look for Key Points and Summary sections at the end.

Upload Documents (Optional)
Click the paperclip to upload plan documents (PDF, DOCX, XLSX) for analysis. Available with GPT-5.2.

Provision	2025	2026
401(k)/403(b) Deferral Limit	$23,500	$23,500
Catch-Up (Age 50+)	$7,500	$7,500
Super Catch-Up (Ages 60-63)	$11,250	$11,250
Annual Additions (415 Limit)	$70,000	$70,000
Annual Compensation Cap	$350,000	$350,000
HCE Threshold	$160,000	$160,000
Key Employee Officer	$230,000	$230,000
SIMPLE IRA Deferral	$16,500	$16,500
IRA Contribution	$7,000	$7,000
Roth Catch-Up Threshold	$145,000	$150,000

Source: IRS Notice 2024-80 and SECURE 2.0 Act provisions. Verify at irs.gov.

ADP/ACP Test

Actual Deferral Percentage / Actual Contribution Percentage tests to ensure nondiscrimination.

EPCRS

Employee Plans Compliance Resolution System - IRS correction programs for plan failures.

HCE

Highly Compensated Employee - earned $160,000+ in 2025/2026 or 5% owner.

LTPT

Long-Term Part-Time Employee - works 500+ hours in 2 consecutive years. Eligible for elective deferrals.

QNEC

Qualified Nonelective Contribution - fully vested employer contribution used in corrective calculations.

Safe Harbor

Plan design that satisfies ADP/ACP testing through required employer contributions and notices.

SECURE 2.0

Setting Every Community Up for Retirement Enhancement Act of 2022 - major retirement reform legislation.

TPA

Third Party Administrator - provides plan administration, compliance testing, and government filing services.

Top-Heavy

Plan where key employees hold more than 60% of total account balances. Triggers minimum contribution requirements.

415 Limit

IRC Section 415(c) annual additions limit - all employer + employee contributions to DC plans.

Chat(k) has a curated D1 database of 1,521 entries covering 401(k), 403(b), SIMPLE IRA, SEP IRA, defined benefit plans, EPCRS corrections, SECURE 2.0 provisions, IRS contribution limits, safe harbor rules, nondiscrimination testing, and more. The knowledge base is regularly updated with new IRS guidance.

GPT-5.2 is recommended for most queries as it combines the D1 knowledge base with web search for real-time information. Claude (Opus 4.5) excels at nuanced analysis and detailed explanations. Gemini 2.0 Flash is fastest for quick lookups.

Yes, when using GPT-5.2. Click the paperclip icon to upload PDF, DOCX, XLSX, CSV, or TXT files. Documents are processed through OpenAI's file search for analysis. Files are not permanently stored.

The knowledge base includes guidance through early 2026 including IRS Notice 2024-73, 2026 mandatory Roth catch-up rules ($150K threshold), and SECURE 2.0 provisions. GPT-5.2's web search can find the latest guidance published on irs.gov.

No. Chat(k) provides educational guidance for retirement plan professionals. All responses include a disclaimer. For specific plan decisions, always consult qualified ERISA counsel or compliance professionals.

v4.3 January 30, 2026

Critical search fix: newest entries now prioritized (ORDER BY id DESC)
Upgraded to Claude Opus 4.5 and GPT-5.2
Added diagnostic /test endpoint
Increased search coverage per term (10 to 25 results)

v4.2 January 2026

Search term limit increased from 5 to 10
Added phrase search for LTPT, SECURE 2.0, EPCRS, QNEC
403(b) LTPT knowledge base added (14 entries)
Roth catch-up threshold updated to $150,000 for 2026

Encrypted Transit

TLS 1.3 encryption for all API communications

Zero Data Retention

No prompts or responses stored by AI providers

SOC 2 Certified

All AI providers maintain SOC 2 Type II certification

Enterprise API with zero data retention. No training on API inputs. HIPAA eligible. trust.anthropic.com

Enterprise API with zero data retention by default. No training on API inputs. ISO 27001. openai.com/security

API data not used for model training. Google Cloud security infrastructure. ISO 27001 certified.

Workers serverless compute, D1 database. Enterprise-grade DDoS protection, global edge network. ISO 27001.

When you send a message, here is what happens:

Your Question
Sent over TLS 1.3 to the Cloudflare Worker endpoint.

Knowledge Search
Worker searches the D1 database for relevant entries. No data leaves Cloudflare at this step.

AI Processing
Your question + knowledge context are sent to the selected AI provider via enterprise API. Zero data retention policy applies.

Response Delivery
AI response returned through Cloudflare to your browser. No server-side conversation storage.

What is NOT sent:

Personal identifying information, Social Security numbers, employee census data, or plan participant details are never required and should not be included in queries.

No. All three AI providers (Anthropic, OpenAI, Google) are accessed via enterprise API endpoints with zero data retention. Inputs and outputs are not used for model training.

Conversation history is stored only in your browser's local storage for convenience. No server-side storage of conversations occurs. Clearing your browser data removes all history.

No. Do not include Social Security numbers, participant names, or other personally identifying information. Use generic terms or hypothetical scenarios when asking questions.

All providers maintain SOC 2 Type II certification. Anthropic and OpenAI are HIPAA eligible for enterprise use. Cloudflare infrastructure is SOC 2, ISO 27001, and PCI DSS certified.