Bernardo García

Building data-intensive agents quickly teaches you one thing: large tool outputs don’t belong in the context window. A simple request can explode into thousands of IDs. Once those flow through tool calls and multi-agent plans, tokens blow up, latency spikes, and agents start failing. We fixed this with memory pointers instead of raw payloads. This week I read an IBM Research paper that independently lands on the same solution. Different domain, same failure mode, same conclusion. Wrote a technical breakdown 👇

THIS GUY BUILT AN AUTOMATED PIGEON DEFENSE SYSTEM FOR HIS BALCONY pigeons kept nesting on his balcony so he engineered a full detection and deterrent system here's how it works: 1\ camera captures video in real time 2\ an AI model identifies the pigeon in real time 3\ a water gun mounted on servo motors turns toward it 4\ sprays the pigeon automatically the hardware: > an orange pi 5 running the detection model > a disassembled electric battery-driven water gun > USB camera > 2 servo motors for aiming > resistors and a transistor to trigger the water gun the detection runs on an AI vision model (yolo world v2) using the rockchip 3588's built in neural processing unit. the best part is that it's not limited to pigeons. because it uses open vocabulary detection, you can reprogram the target to any object. squirrels, cats, raccoons, whatever is messing with your balcony fully automated, runs 24/7, no manual intervention needed

We’re reimagining a 50-year-old interface - the mouse pointer - with AI. 🖱️ These experimental demos show how people can intuitively direct Gemini on their screens using motion, speech, and natural shorthand to get things done 🧵

On a long trip back home after a week in SF.. want to learn about more rust during the flight so downloaded qwen3.5:9b to have some backup.. positively impressed by how much the small models have improved!!

Link to recording: youtube.com/watch?v=VFAovr…

Spoke at Context is King #4 in SF yesterday about why we ended up building our own specialized agent harness instead of reusing an existing one. I walked through the default behaviors we encoded into the harness, and the implementation choices they led to: how to make schema, organizational knowledge, and business rules available to the agent all at once; how to let our semantic data layer learn at the same pace as knowledge evolves; and how to efficiently manage the context window when working with indivisible data. Thanks @aiven_io for co-organizing, and to everyone who came out.

@cyntro_py I totally agree. Do you think people will be still around in stage 6?

Six stages every company goes through to become AI-native: 0. CEO says AI is important, but nobody does anything except pretty presentations, meetings, and breathy sighs of "yes, important" and "we're being replaaaaaced" 1. People manually feed context to chatbots: dropped an excel into claude, pasted a paper into GPT, opened a miro board through claude code, asked it to post in slack and pretend you're working 2. People assemble their personal OS with connectors and skills: local data structure, automatic message and document ingestion once a day, regular skills that sort through email and prep reports in obsidian 3. Teams have shared workspaces, scripts, and skills. Engineering has shared AGENTS[.]md for every repo, sets of prompts and skills, common practices. Marketing has its own templates and skills. 4. The company has shared AI infrastructure, every function is visible to agents. An agent can find a message in support, connect it to a ticket in engineering, and flag it for marketing. There's access control, sync, execution and cost monitoring across the whole company. 5. People build self-optimizing processes: e.g., a CMO launches a self-developing marketing campaign run by a swarm of agents. The system can improve itself, but humans build, orchestrate, and supervise it. 6. Cybernetic company: the company has sensors and feels everything happening inside and in the market, turns it into tokens and makes it visible to AI systems, proactively improves itself. Agents form hypotheses, test them, and ship changes. You can't skip stages. Most companies today are at level 1-2.

Sharing my experiences from building specialized harnesses for analytical SaaS companies. It's likely that your harness requires your own defaults around data, context, multi-tenancy, and evolving business rules. After all knowledge work is different from software development. Which default behaviors do you encode in your harness today? Are you encoding them in the best way?

The most common mistake we see in analytical agents is dumping data into the context. Letting the agent work from references to the data instead keeps the system fast, the numbers right, and each customer's data separated. @ksariola does the best job I have heard of explaining why that matters and what it changes about how you ship reliable agents on top of real customer data.

@flowaicom @ksariola Super high-quality content

Claude Code is a great agent harness, for coding. For analytical SaaS, it is the wrong default. Our CTO @ksariola took that case to AgentCon Silicon Valley this week, drawing on our experience of building specialized harnesses for analytical SaaS. youtube.com/watch?v=pikC5I…

Now the whole setup is a template the rest of the team can clone for their own talks. What are people using to build slides these days?

Then I pushed everything to a repo so my co-founder could polish it. He cloned it, took the design from 80 to 100 with Claude, and opened a PR -> Git-based, on-brand, collaborative slide design.

Building slides with Claude made PowerPoint feel unnecessary. 🧵

We are back in San Francisco for Context is King no. 4 on Tuesday, with over 100 builders already in and a few last spots left. My co-founder @bergr7 is going deep on the harness we built at @flowaicom for data-heavy analytical agents. Search structure, memory pointers, multi-tenancy, the parts that make our agents actually work with real data. Joined on stage by Itai Smith (@trychroma), @RomainSestier (@StackOneHQ), @OtsoVeistera (@thetokenco), and @NathanBurg (@GitHits_com). Let's go 🌉