Blog - Vellum
Thoughts on Personal Intelligence
Stories, best practices and ideas on how to use AI assistants for work and personal ops.
How to Turn Voice Memos into Client Invoices with an AI Assistant
Turn 20-second truck audio notes into itemized invoices in 30 seconds. How contractors calculate markups, eliminate paperwork backlogs, and stage bills.
GuidesSep 15, 2026 • 9 min
Read: How to Turn Voice Memos into Client Invoices with an AI Assistant
How to Audit Restaurant Vendor Invoices with an AI Assistant
Automate food vendor invoice auditing in 14 seconds. How restaurant owners catch silent price hikes, pack-size shrinkage, and stage approved accounting bills.
GuidesSep 11, 2026 • 10 min
Read: How to Audit Restaurant Vendor Invoices with an AI Assistant
How we use OpenSEO & Vellum Assistant to run our SEO engine
How we replaced my Ahrefs subscription with OpenSEO and Vellum to save time and money by running an agent-native organic search and GEO engine directly with an Assistant.
GuidesSep 9, 2026 • 4 min
Read: How we use OpenSEO & Vellum Assistant to run our SEO engine
GPT-6 Astra Benchmarks Explained
Every GPT-6 Astra benchmark explained, and compared head to head with GPT-5.6 Sol, Claude Fable 5.1, Opus 5, and Gemini 3.8 Flash.
Model ComparisonsSep 3, 2026 • 9 min
Read: GPT-6 Astra Benchmarks Explained
Agent Within Reach
Voice conversations run deeper than typed. The hard part is starting one.
AllSep 3, 2026 • 2 min
Gemini 3.8 Flash & 3.8 Flash Cyber Benchmarks Explained
Every Gemini 3.8 Flash benchmark explained, and compared head to head with Claude Opus 5 and GPT-5.6 Sol.
Model ComparisonsSep 3, 2026 • 14 min
Read: Gemini 3.8 Flash & 3.8 Flash Cyber Benchmarks Explained
Why Subscriptions > Credits for AI Products
We built our first monetization path around credits because AI costs are variable. Customers taught us that accurate metering and a good buying experience are not the same thing.
AllSep 2, 2026 • 7 min
Read: Why Subscriptions > Credits for AI Products
Claude Fable 5.1 & Claude Mythos 5.1 Benchmarks Explained
A breakdown of Anthropic's Claude Fable 5.1 and Mythos 5.1 benchmarks. Head-to-head scores against Fable 5, Opus 5, and GPT-5.6 Sol across scientific research, terminal coding, knowledge work, computer use, reasoning, business workflows, and coding, plus the safeguard zeros behind the table and the cache-read price cut that pulled Cognition's Devin traffic on launch day.
Model ComparisonsSep 2, 2026 • 10 min
Read: Claude Fable 5.1 & Claude Mythos 5.1 Benchmarks Explained
How Vellum Remembers
Learn how we built the Vellum Assistant's memory system to remember you, your context, and work.
AllAug 5, 2026 • 9 min
GPT-5.6 Sol vs Terra vs Luna: Which Tier Should You Actually Use?
A tier-by-tier breakdown of GPT-5.6's three models: benchmarks, pricing, reasoning modes, and which one to route your work to.
Model ComparisonsAug 3, 2026 • 9 min
Read: GPT-5.6 Sol vs Terra vs Luna: Which Tier Should You Actually Use?
Claude Opus 5 Benchmarks Explained
A side-by-side read of Claude Opus 5 against Fable 5, Opus 4.8, and GPT-5.6 Sol on every benchmark Anthropic published, including the ARC-AGI 3 result that triples the next-best model.
Model ComparisonsJul 24, 2026 • 8 min
Read: Claude Opus 5 Benchmarks Explained
How We Gave Your Assistant a Voice
How we made an AI speech pipeline feel like one continuous thing.
AllJul 24, 2026 • 4 min