"To Read or Not to Read": Digital Haunting is Over

In the modern workplace, Shakespeare's timeless "To be or not to be?" has transformed into a question relevant for our digital age: "To read or not to read?". Every day, we face a barrage of push notifications, emails, and messages, each demanding an immediate response. This toxic culture of 'agility' has put instant response ahead of real productivity, creating a digital spectral onslaught that sucks our time into a black hole, leaving nothing for productivity and creativity.

At kor.gy, we've been actively building a solution against this digital haunting since 2022. We understand that today's companies must be agile to survive and be competitive. However, should humans be the ones responsible for that agility? While everyone was talking about autonomous agents, we were actively building them and integrating them into real business processes, gaining valuable real-life feedback.

Our final approach involves autonomous agents technology taking over everyday operations, transforming human employees' roles towards those akin to executives focused on analysis, decision-making, and strategy. We're introducing a new concept of "institutional intelligence", a kind of AI that is not a companion of a single employee but a complex unified technical entity designed to take over everyday business operations of the whole company.

As far as we are a software development company, our customer service largely involves handling bug reports from customers. Some of our corporate customers have thousands of end users, and for us as a small company, this often felt like a tsunami. Before we built kor.gy AI, our customer service was a bottleneck, struggling to process the volume of cases and provide immediate, competent insights.

Now, not only do we automate the process, but we also collect around 18x more relevant information. The bot is neither tired nor nervous, it’s asking the user exact questions needed to get into the case and have it resolved ASAP. The bot is capable of providing basic technical assistance and even proposing a workaround for the issue, thanks to all software documentation being transferred to a formal logic backed format. It can access the codebase, logs, and a wealth of data from the customer device, providing immediate, highly competent insights for both customers and our dev team. This has reduced the time needed for bug fixing by around ten times.

Let's take a real bug description as an example. A customer reported that their application crashes when they try to upload a file larger than 2GB. Previously, this would have required a developer to replicate the issue, dig through logs, and manually identify the problem. With kor.gy AI institutional intelligence platform, the bot immediately accesses the codebase and logs, identifies the issue in the file upload module, crafts a bug fix, and even deploys it on a test environment on its own. The only task left for human employees is code review. To add that the AI also builds an automated test case to check all the future code versions against this specific issue, would be obvious.

Unplanned task handling, such as when a bug report comes in, has been streamlined by kor.gy AI. Previously chaotic and stressful, it's now seamlessly inserted into the regular sprint-based workflow with zero human involvement. Developer has a personal task queue and unplanned bug report is unplanned only for the customer support agent bot.

Mission chats are another highlight. These are per-task chats automatically created by bots who form a mission team of dedicated bots and humans. As a fully aware institutional intelligence platform it picks only the right people (for example, developers that built a particular service) and would not disturb someone who is out of office. This replaces chaotic discussions with structured, focused communication, making bug handling fast and seamless, delighting customers, and removing any stress from the team.

Advanced technology behind the platform is combination of mainstream LLMs, custom LLMs, custom domain specific ML solutions (we call them LBMPs what stands for large business process models) and symbolic processing. Symbolic processing, particularly formal logic and term rewriting (steered by an ML model that defines strategy for applying term rewriting rules) are forming a backbone which makes that platform real business ready.

While competing autonomous agent solutions are like road vehicles with reasonable safety concerns, the thing we are building shares a lot with the world of high speed railways, ensuring a safety envelope for AI to operate within, guarded by the formal logic and symbolic processing in the core of the platform.

At this moment we're ready to start discussing cooperation with potential early customers who will participate in building up the technology, improving LBPMs, and crafting domain-specific LBPMs. If you're interested in joining us on this journey, we'd love to hear from you.

Our priority with kor.gy AI is not just developing technology; we're crafting a new paradigm for work in the AI era. The project is covering a vast span of R&D including user interfaces and building pioneering E-Ink based user interfaces that look like nothing before and follow the idea of leaving the agile, flashy and bright digital world for its natives (bots) and bringing back the warm analogue charm of les trente glorieuses to humans.

There is nothing to worry about, there is nothing to hurry for.

Let the machines work, let the humans dream and invent.