· 5 min read
How Tamaton's AI Reads Your Inbox So You Don't Have To
A concrete walkthrough of Tamaton's inbox triage agent: how 84 unread messages become a ranked queue with searches run, drafts written, and calendar holds already placed.
Blog
From the Tamaton team
Practical writing on productivity, AI, and building software.
All#ai#ai-agents#ai-productivity#applied-ai#automation#calendar#context-switching#deep-work#documents#email#engineering#engineering-leadership#invisible-ai#knowledge-work#llm-evaluation#product-updates#productivity#prompt-engineering#remote-work#search#security#spreadsheets#startups#storage#tamaton#unified-workspace#workflow#workflows
· 4 min read
Aggregate Scores Lie: What EduFairBench Teaches AI Evaluation
Averaged benchmark scores hide the subgroup failures that break real workflows. Here's how to build disaggregated, reproducible evals for triage, summaries, and spreadsheets.
· 5 min read
One Model to Rule Them All Is a Myth: Routing for Real Work
Chasing the single 'best' LLM is a losing game. Here's a practical framework for routing each task — code, extraction, drafting, reasoning — to the model that actually fits.
· 5 min read
From Chatbots to Reasoning to Agents: What Changed in 2025
Reasoning models solved 'can it answer.' Agentic training moved the hard part to 'can it act reliably' — and that shift changes how everyday work gets done.
· 5 min read
When Your Agent Says "I Might Be Wrong": Honest AI Triage
Confident guessing is the most expensive failure mode in AI workflows. Here's how surfaced confidence and cited sources turn inbox triage and search into something you can actually trust.
· 5 min read
The Audit Trail Problem: Email as Agent Accountability
Enterprises can't deploy agents they can't review. Email threads already give every agent action a timestamped, human-readable, forwardable paper trail.
· 5 min read
Grounding Is the New Prompting: How to Stop Agents Guessing
Clever prompts don't make agents accurate — verified, fresh sources do. Here are the grounding architecture patterns that keep AI agents honest.
· 5 min read
Small Models, Big Inbox: When Not to Reach for a Frontier LLM
For classifying, routing, and summarizing email, a small fast model usually beats a frontier one on cost and latency. Here's how to choose per task instead of per vendor.
· 5 min read
Why Rigid APIs Break Agents — and Email Doesn't
Strict API contracts turn small agent mistakes into hard failures. Email tolerates ambiguity, supports clarification, and keeps the whole exchange on the record.
· 5 min read
Keeping LLMs Current: Knowledge Editing vs. RAG
Pretraining freezes your model's worldview on a date that has already passed. We compare knowledge editing research with retrieval as strategies for keeping an AI assistant accurate.
· 5 min read
The Agent-to-Agent Email Problem Nobody Has Solved
When agents email agents, threads, signatures and pleasantries become expensive noise. Here's the case for a structured A2A email layer with provenance, intent tags, and evaluation built in.
· 4 min read
Why Your AI Agent Fails at the Handoff, Not the Reasoning
Most agent failures aren't reasoning failures — they're broken context handoffs between tools, memory, and steps. Here's where the seams actually tear, and how to reinforce them.
Get started
Reserve your @tamaton.ai email
Claim your address before someone else does — free to start, with an AI-native inbox built in.