Written by the people who build
Blog / Insights
Field notes from live deployments, including the ones that went badly.
Written by our engineers and architects, not by a content team. Published when we have something to say, which is roughly twice a month.
No newsletter pop-up. If you want it in your inbox, ask.
Latest
Your agent is fluent, confident, and wrong
Four out of five failed pilots we are asked to rescue have the same root cause, and it is not the model.
Prompt injection in a case description
A customer wrote instructions into a support ticket. The agent followed them. Here is what we changed.
The second agent is harder than the first
One agent is a project. Two agents is an architecture. What changes, and what breaks.
We told a client not to buy Agentforce
Their volume was too low and their knowledge base was a decade of undocumented tribal memory. The maths did not work.
Latency is the whole game in voice
Where the milliseconds go, and the four places we found to get them back.
How to write an instruction an agent will follow
Instructions are source code. Most of them are written like a memo.
Also in resources
Guides & Playbooks
The documents we use internally, lightly edited so they make sense outside the building.
Read Customer storiesCase Studies
Named where we are allowed to name them, and specific where we are not.
Read Model it yourselfROI & Pricing Calculators
Two calculators. Neither of them asks for your email.
ReadStart with the assessment.
Three to four weeks, a fixed fee, and an answer we are willing to say out loud — including if the answer is no.