Trida Signals
Blog
AI engineering insights from the field — LLMs, agents, and production lessons.
Get updates
Field Notes
Designing Tool Errors for Agents
When a tool fails, the error is the only instruction the agent gets for what to do next. How to write errors that lead to a correction instead of a retry loop, and what to keep out of them.
Blueprints
Open-Model Coding Agent Blueprint
Blueprint: a governed coding agent on open models served by Baseten — scoped work orders, an approved plan, isolated work, independent re-grading, and the one figure that matters, cost per successful task.
Blueprints
Page Agent Admin Copilot Blueprint
Blueprint: a governed copilot for legacy admin consoles — a written, field-level plan a person confirms before any work runs, refusals at plan time for deletes and oversized credits, and an independent read-back that decides whether the job is done.
Blueprints
Deep Diligence Agent Blueprint
Blueprint: a diligence agent running on LangChain's deep agents harness inside Trida's governance shell — an approved plan before research starts, findings refused without evidence, an independent check on every claim, and a report a partner can defend sentence by sentence.
Blueprints
CX Resolution Agent Blueprint
Blueprint: a telecom support agent with one accountable supervisor and six specialist desks — identity proven before anything consequential, disputes decided on the evidence in the bill under written caps, promises kept in a ledger, and escalation by rule rather than by mood.
Blueprints
Month-End Close Agent Blueprint
Blueprint: a month-end close deployment built around aidnn in which governed, versioned definition packs shape the figures, an independent verifier recomputes every number from source, and a finance lead signs a pack that carries its own evidence.
Blueprints
Invoice Document Processing Agent Blueprint
Blueprint: an invoice-processing deployment built on Reducto that reads supplier documents with page-level citations, matches them three ways in code, holds bank-detail changes at a fraud review no approver can clear, and proposes payments instead of executing them.
Field Notes
Prompt Caching in Production Agents
An agent re-reads the same instructions, tools and history on every step, and pays full price each time unless the prompt is built for caching. Where the boundary belongs, which TTL fits which gap, and how to read a hit rate honestly.
Field Notes
Agent Memory That Earns Its Keep
An agent that remembers everything soon remembers it wrong — stale facts and one-off details included. Memory earns its keep only when what to store, when to write it, and when to delete it are designed decisions, not defaults.
Partners
Trida AI Joins Google Cloud Partner Network
Trida AI is a member of the Google Cloud Partner Network. What membership means for our customers, and the road from registered member to Select.