Proto Typing Too (Page 2)

Latest

A coffee-stained napkin reading “Important Project Rule: Do Not Change” sits atop an organized filing cabinet labeled for permanent project information.

The AI Forgot My Instructions.

AI may not have forgotten your instructions. They may be stored in the wrong place. Why persistent project rules need more than a conversation to survive.

AI retrieval pipeline discards a valid building match before it reaches the model.

Your AI Isn’t Stupid.

When an AI agent gives the wrong answer, the model may not be the problem. A real debugging story shows why developers should check what the model actually received before rewriting the prompt.

Illustration of a red Save button, a green "Saved Successfully" message, and a "Missed Schedule" stamp, showing how software can appear successful while quietly failing.

Trust Signals in Software

The blog about "The Save Button That Doesn't" missed its own scheduled publish date. You can't make this stuff up.

A multigenerational software team combines an experienced architect’s blueprint with fresh ideas to build a glowing digital bridge together.

Great Software Teams Need Both Experience and Fresh Energy

by Brian Pollack If you needed to build a wall, would you hire a young go-getter with lots of energy and muscles or a wise mason with 30 years of experience? Honestly, you need both. Tech teams especially need a mix of senior developers and leads who have seen all

AI model variants pass through evaluation gates, with the strongest version selected to continue an automated cycle of recursive improvement.

AI Models Are Improving Themselves—and the Pace Is Accelerating

by Brian Pollack There is a new model every other day, and there is a reason the pace keeps picking up. The models are now improving themselves. The industry term is RSI, or recursive self-improvement. Google has been the loudest about it, but every major lab is running some version

Computer monitor stuck at 99% with two minutes remaining, surrounded by cobwebs while a loading spinner continues to turn.

The Progress Bar Lies

Progress bars look precise, but the percentages aren't always what they seem. A look at how software uses progress indicators to communicate status, manage expectations, and quietly shape what we trust.

Blue, a small robot, skips across stepping stones while pieces of conversation context disappear beneath the stream.

Your AI Doesn't Experience the Time Between Your Questions

Illustration of how ChatGPT and generative AI process conversations as successive interactions rather than experiencing continuous time, using stepping stones and a river of lost context to explain AI context loss and conversational memory.

An AI team watches a glowing prototype pause at a gap before production, with five readiness checkpoints above the stalled launch pipeline.

Your AI project worked in the demo. So why didn't it launch?

by Prateek Sharma, AI Engineer Almost every AI trial we see does what it promised in the meeting room. Most still never reach a real customer. The gap is rarely the technology — it's five things nobody scoped at the start. Download a pdf of this white paperProtovate_Why_

A massive data center built inside a cloud sends glowing data streams to people using streaming video, gaming, video calls, and online services across a city.

WiFi Doesn’t Come From the Clouds: The Internet’s Data Center Problem

by Brian Pollack Do they think WiFi comes from the actual clouds? The internet needs a new PR firm. Let me explain. If you're posting about data centers while binge-watching Netflix with Reels scrolling by on your phone, you might want to learn where WiFi goes to sleep

Cluttered engineering workbench covered with tools, coffee cups, schematics and wires, with a partially buried wrench highlighted by a subtle blue glow.

The AI Didn't Forget.

AI doesn’t always forget instructions. Sometimes the context window gets overloaded. Two very different AI workflows reveal the same problem - and why managing context matters.

Prompt engineer working at a computer as structured prompt blocks flow through an AI system toward a verified result.

How to become a prompt engineer in 2026

by Prateek Sharma, AI Engineer From writing your first effective prompt to building real AI systems: a practical roadmap for learning prompt engineering, and an honest look at what the career has actually become. Download a pdf of this white paperProtovate_How_to_Become_a_Prompt_Engineer.pdf3 MBdownload-circle What

Worker cuts through a server room wall with a power saw, exposing digital code and illustrating the security risks of rogue IT and unauthorized AI tools.

The Hidden Risk of AI-Powered Rogue IT

by Brian Pollack Nobody hands out cordless saws at orientation. Think about the building you work in. Somebody designed the way people get into it. There is one main entrance where visitors sign in, badge readers on the doors that matter, cameras on the parking lot, a fire marshal who

Jana sits on a giant mushroom with a laptop and coffee as smoke curls into several small robots.

Replacements?

After being laid off, I’m helping build an AI version of myself to preserve decades of institutional knowledge. But how much can AI capture before knowledge gives way to judgment?

Rough wall repair blending into an IT workspace with technical diagrams, illustrating quick fixes versus specialized expertise.

Brian Pollack

Fully Staffed Is Not Fully Covered

Nerds aren’t interchangeable. Why even strong IT teams sometimes need specialized expertise in accessibility, security, performance, and other high-risk areas.

Software Ownership

The Cost of Owning Software

by Brian Pollack Clients rarely misunderstand the cost of building software. That part is simple: the client asks for software with a goal, we build it, the transaction is complete. But that’s only the first step. Once software exists, someone has to own it. Below are those ownership costs,

AI chip connected to customer service, data processing, workflow automation, and follow-up processes, promoting a free 30-minute AI consultation.

AI Lab

You Know AI Matters. But Where Does It Actually Pay Off?

by Brian Pollack After 30 years of building software for healthcare, telecom, and many enterprise companies, I've pointed Protovate at one thing: AI software development. Custom software with AI doing the actual work inside it. Not chatbots. The unglamorous stuff. Recently we automated customer service requests and followup

Carpenter given a pocketknife while his professional tools remain locked away.

It’s 2026, Not 2016

Technical interviews still test experienced developers on memory and syntax, even as real-world development increasingly depends on judgment, problem-solving, and intelligent use of modern tools. Prateek Sharma asks whether it’s time for the interview process to catch up.

AI engineer viewing a roadmap with five stages: Learn, Build, Deploy, Iterate, and Lead.

White Paper

How to Become an AI Engineer

The roadmap I actually walked. Sixteen years of shipping software, two years of shipping AI, and everything that broke in front of real users along the way. 7 stages 90-day plan No PhD required AI engineering is not ML Stop treating it like a chatbot Where the intelligence belongs Your

Abstract neural network made of interconnected nodes and lines, representing a locally running AI model.

White Paper

We stopped paying per token and put the models on our own machine

One workstation, a stack of open models from Hugging Face, a few hundred lines of Python, and an ngrok tunnel. Here is how Protovate builds and tests AI apps before a single API bill arrives. GPU memory32 GBenough for 7B–13B System RAM128 GBoffload headroom Models pulledText + imageopen weights Shared

A computer monitor displays a huge Excel spreadsheet named "JANAs-spreadsheet-COPY-203.xlsx," representing the business-critical spreadsheet everyone is afraid to change.

The Spreadsheet Everyone Was Afraid to Touch

A temporary spreadsheet built to save time became a business-critical system that outlived its creator. A funny story about institutional knowledge hiding in plain sight.

Computer screen showing multiple spreadsheet files named Budget_FINAL, Budget_FINAL_v2, Budget_FINAL_v2_REAL, and Budget_FINAL_v2_REAL_THISONE, illustrating version-control workarounds.

Version Control for Normal Humans

Why do so many shared drives contain files named FINAL_v2_REAL_THISONE.xlsx? It isn't bad file naming. It's how people adapt when software doesn't match the way work actually gets done.

Illustration of a project manager on vacation while an AI assistant handles routine project questions, highlighting the value of captured knowledge and documentation.

The Vacation Test

A 35-day vacation, an AI clone, and the real lesson about knowledge capture. AI doesn't replace experience. It depends on it.

2 office workers chat by a server rack used like a water cooler. A monitor above joins the conversation, showing how an AI assistant can feel like a coworker after repeated interactions.

The New Work Relationship

We spend years teaching computers to communicate like people. Then we accidentally start treating them like coworkers. Why the relationship change has more to do with us than AI.

A traveler uses a smartphone while airport workers prepare an aircraft outside, illustrating that people still perform most of the work behind modern travel.

Typewriters. ATMs. AI?

What do the printing press, the typewriter, the ATM, and AI have in common? Every one sparked predictions of widespread job losses. The reality was far more interesting.

See all