I build Giant Context.
An autonomous marketing platform, built and run by one person
Latest writing
New field notes from the build, as they land
Cutting the homepage from 5.7MB to 217KB
Ten days of frontend performance work on the public site, including the four hypotheses that turned out to be wrong. Images were served raw because a ...
|
Sep 1, 2026Compute gets cheaper per request at scale
I price Giant Context at $25 per million tokens and had no defensible answer to what it costs to serve a thousand customers. Four of the five cost cla...
|
Aug 30, 2026Removing the test cache halved CI time
The test cache made the suite fast by replaying old results instead of running the tests. Deleting it was correct and it made everything slower. Then ...
|
Aug 29, 2026The build in milestones
Nine months, one platform, one person. Every turn that mattered.
The business case for Giant Context
Where the series begins, and the market one autonomous platform opens.
Building analytics in week two
One event contract and one table, built before there was a product to measure.
Building the page editor first
Foundation before features. A typed tree of sections, columns and blocks.
API keys, the SDK, and MCP from one spec
One specification becomes the console's client, a developer SDK and an agent's toolbox.
A single prompt cannot build a website
The agentic pipeline begins. Stages in sequence, not one prompt.
My errors come with fixes attached
Production errors arrive already diagnosed, with a fix attached.
Email without campaigns or segments
Marketing email decided one recipient at a time, no campaigns.
A website chat that only answers from published content
A public chat grounded in the site's own knowledge, safe to ship.
Turning uploaded files into retrievable context
The system learns a business from the files it hands over.
Deleting 20,000 lines of rules from the builder
The pivot. Twenty thousand lines deleted, the model left to reason.
Automating multichannel marketing
The product. One mind runs web, mail, and social from one grounding.
Reading an SDK to write its docs
A model reads a real SDK and writes the documentation it never had.
What one person actually built
The census. 505 operations, one integrated platform, one person.
Compute gets cheaper per request at scale
Little's Law, a closed form, and thirty days of real usage against it.
About me
Jesse Richard
I'm Jesse James Richard. Twenty years building software from the first line of code taught me the engineering only counts when it opens a market. Now I'm building Giant Context on my own, one architect directing a crew of AI agents, writing down how as I go. I live in Canada with my family, play more hockey than I should, and read more than anyone thinks is reasonable.