AI agents for small business: three jobs worth handing over
Finding tenders, drafting the blog and closing the month: what each agent does, what it hands back and what stays with you

Arclyx blog
Notes from building Arclyx.
Engineering research on running AI agents in production: why they fail, what they cost and how to keep them safe, plus product releases and company news, from the team building Arclyx.
Latest
The newest posts, across the blog
LLM inference explained: where the time and money go
What happens between sending a prompt and reading the answer, what it costs, and how to make it cheaper and faster
AI agent vs chatbot: what's the difference?
Who starts the work, where it happens, what comes back and what keeps it in bounds, with one job done both ways
AI agents for small business: three jobs worth handing over
Finding tenders, drafting the blog and closing the month: what each agent does, what it hands back and what stays with you
Ask Arclyx: an AI assistant that knows the page you're on
What it sees on each page, what to ask it there, what it can open and draft for you, and what it never does
Give your AI agents a knowledge base your team controls
What to put in it, which files agents can read, and how access tags decide which documents each person's agents can search
How to connect a remote MCP server to your AI agents
What a remote MCP server is, how it differs from a local one, and how to add one so every agent in your workspace can use its tools
How to debug an AI agent that went wrong
A method for finding the round where a run went off track, telling instructions from tools from data, and proving the fix
How to keep AI agent costs under control
The limits worth setting, where to see what each agent spends, and the settings that cut cost without hurting the result
LLM inference explained: where the time and money go
What happens between sending a prompt and reading the answer, what it costs, and how to make it cheaper and faster
The 1M Output Horizon
What Gemini 4 Argon is, what it costs, how it benchmarks, and how to engineer around its 1M-token runs
Solving LLM Serving Latency Interference
Cutting LLM inference latency with prefill-decode disaggregation and hybrid scheduling
The Evaluation Containment Crisis
What an AI agent sandbox is, how agents escape one, and which isolation to use
The Inference Cost Trap
What an AI agent really costs to run, and how frontier labs are cutting it
The Observability Gap
How AI agent monitoring works: traces, evals and the tools teams use in production
The Reliability Gap
AI agent reliability: why frontier agents still fail 1 in 3 benchmark tasks, and what is changing
The Idempotency Problem in Agentic Tool Calling
Why idempotency is the hardest reliability bug in AI agent tool calling
AI agent vs chatbot: what's the difference?
Who starts the work, where it happens, what comes back and what keeps it in bounds, with one job done both ways
AI agents for small business: three jobs worth handing over
Finding tenders, drafting the blog and closing the month: what each agent does, what it hands back and what stays with you
Ask Arclyx: an AI assistant that knows the page you're on
What it sees on each page, what to ask it there, what it can open and draft for you, and what it never does
Give your AI agents a knowledge base your team controls
What to put in it, which files agents can read, and how access tags decide which documents each person's agents can search
How to connect a remote MCP server to your AI agents
What a remote MCP server is, how it differs from a local one, and how to add one so every agent in your workspace can use its tools
How to debug an AI agent that went wrong
A method for finding the round where a run went off track, telling instructions from tools from data, and proving the fix
How to keep AI agent costs under control
The limits worth setting, where to see what each agent spends, and the settings that cut cost without hurting the result
How to use an AI agent in Slack
Ask an agent in a channel, get the answer in the thread, and let agents post their reports to the channel you pickBe first in when doors open.