Brendon Garner

[Brendon Garner @ittsbrendon]

we hit this with Podium. the enrichment layer lives behind an API and can surface anywhere. but the review surface, where a human approves what the agents ship, stays in Canopy. the data being portable is fine. the judgment loop has to be yours

[Brendon Garner @ittsbrendon]

built Canopy around this. every draft routes through a human before it ships. the non-obvious part: that review step is where the agent learns. reviewer notes become memory the model reads on the next run

[Brendon Garner @ittsbrendon]

the restructuring buys output capacity, not revenue. we shipped 10x more campaigns with Canopy in the first quarter. review queue grew. distribution didn't follow. scaling the production layer is the easy part

[Brendon Garner @ittsbrendon]

@danshipper @lennysan built https://t.co/Yve2qZ6qNt around this. every draft routes through a human before it ships. the non-obvious part: that review step is where the agent learns. reviewer notes become memory the model reads on the next run

[Brendon Garner @ittsbrendon]

two weeks iterating on Canopy's voice gate trying to coax the model into catching the right patterns. the day we wrote the rule directly it started working

[Brendon Garner @ittsbrendon]

@cursor_ai bringing branch management, stashing, overwrites, regressions and all the fun back into workflows - the pains of progress - 'pls rebase and build on fresh branch pls' ... 'you deployed to dead branch' so many tokens used on that phrase 😇 preventative prompting helps!

[Brendon Garner @ittsbrendon]

Building agent pipelines made this concrete: the bottleneck is not the initial the code generation; it's how to get the most out of the planning stages, invoking specialists (and making sure they're not ignored!), and iterating - code's always going to need reviewing, but it makes building agent teams at @mycanopyai lot more fun when you start off right

[Brendon Garner @ittsbrendon]

maintenance is harder than the build. rebuilt the context-feeding layer for Sage's AI companion 3 times this year. not because the model changed, because what actually mattered as inputs kept shifting as we learned

[Brendon Garner @ittsbrendon]

@karrisaarinen Tryin me hardest

[Brendon Garner @ittsbrendon]

@jameygannon So good!

[Brendon Garner @ittsbrendon]

@stephenhaney @QuiverAI 👏👏👏

[Brendon Garner @ittsbrendon]

in consumer AI the repetition is often invisible to the user. beauty routines, outfit picks, shopping. they don't experience it as a job to automate so you can't sell them automation. you sell them the outcome. we learned this the hard way with Familiar -> try it yourself https://t.co/3LPiUVjEEd

[Brendon Garner @ittsbrendon]

@jameygannon Brillliaaant! Combo’ing @mobbin Savee @fuserstudio @lummipics and the rest into a design/prod monster I can run from a terminal/editor is the twisted dream…one step closer!

[Brendon Garner @ittsbrendon]

@ericosiu a lot of symmetry with what we’re doing at https://t.co/Yve2qZ6YD1 - would love to share more

[Aaron Levie @levie]

Forward deployed engineers, or equivalent, are about to become one of the most in-demand jobs in tech. And one of the most important functions for AI rollouts.

Deploying agents is far more technical of a task than most people realize, often far more involved than deploying software. Software generally works the same way every time, and generally for the past few decades has been updated versions of an existing technology or concept (which basically means easier for the enterprise to update their workflows on a newer system).

With agents, you’re actually deploying the equivalent of work output within the enterprise. The customer is effectively using you as a professional services provider for a task, which they expect to get solved nearly end-to-end now. This means you need to actually deeply understand the business process as a vendor, and get the customer from the current to the end state seamlessly.

Companies need help figuring out which models will work best for their workflows, they need extensive evals setup often, they need change management support for workflows, they need to get their data setup for the agents, and constant tuning of the agentic system for their process.

Massive role in tech now. And another example of the kind of highly technical work that AI is creating.

[Brendon Garner @ittsbrendon]

the decisions that felt arbitrary in year one are the ones that show up everywhere now. not the big bets. the small calls made before there was enough data to make them carefully

[Brendon Garner @ittsbrendon]

the overnight run is the whole thing. we do this for Podium's pipeline work now. stuff that blocked a dev for a morning just runs while nobody's watching. the hard part is writing the spec tight enough that you actually trust what comes back

[Brendon Garner @ittsbrendon]

@ericosiu Working on this w more on the growing brand/modernization PE arb space with https://t.co/Yve2qZ6YD1

[Brendon Garner @ittsbrendon]

@levie It’s strangely satisfying to go service in some ways after so much time on the ‘easy scale’ - going bespoke is what we’re doing at https://t.co/Yve2qZ6YD1

[Brendon Garner @ittsbrendon]

@scottbelsky Pretty much it