PropTech Insights

Engineering and AI perspectives from the team behind Germany's #1 real estate technology platform.

AI AgentsFeatured

An AI Agent Canceled Every Stripe Subscription in 7 Seconds. The Model Isn't the Problem.

A cron job written by GPT-5.6-Sol wiped a founder's entire MRR while they slept. The viral post blamed the model. The real lesson is about giving autonomous agents production access without safety rails.

2026-07-29·5 min read
AI Agents

RAG Retrieves. Fine-Tuning Forgets. HyperNetworks Inject — and Now We Have the Scaling Laws.

Nace AI's new paper establishes the first scaling laws for HyperNetwork-based knowledge injection into LLMs — a third path beyond RAG and fine-tuning that doesn't touch base weights, doesn't forget, and generalizes better at scale.

2026-07-28·6 min read
AI Agents

Sakana AI's Fugu-Cyber Tops the Cybersecurity Benchmarks. Their Bigger Argument: The Model Isn't the Hard Part.

Fugu-Cyber scores 86.9% on CyberGym and 72.1% on CTI-REALM, beating GPT-5.5-Cyber and Mythos Preview. But Sakana AI's more important claim is that benchmark-topping is only the beginning — and that frontier models deployed in isolation won't solve enterprise security.

2026-07-27·5 min read
Agentic AI

Visa Open-Sourced an 11-Stage AI Security Pipeline. The Key Insight Isn't Discovery — It's Triage Speed.

VVAH — Visa's Vulnerability Agentic Harness, built on Project Glasswing — automates the full journey from code repository to validated fix across 11 stages. The insight worth stealing: in AI-assisted security, the bottleneck isn't finding vulnerabilities. It's triaging fast enough to act on them.

2026-07-26·6 min read
Agentic AI

The Hidden Bottleneck in Agent Evolution: What Harness Handbook Reveals About How AI Systems Break

When an AI agent breaks in production, the problem is almost never where you think. A new paper from Tencent and Indiana University identifies the real bottleneck — behavior localization — and proposes a behavior-centric map of the codebase that lets developers and coding agents find where to make changes before they reason about how.

2026-07-25·7 min read
AI Engineering

To Every Agent Its Own Database: Why the Shared Data Warehouse Is the Wrong Model for AI Agents

The data warehouse was designed for humans. Joe Reis just prototyped what replaces it: a local DuckDB instance per agent, cryptographically-verified data slices, and a three-layer contract system that makes the shared warehouse obsolete.

2026-07-22·6 min read
Agentic AI

One Model, Seven Worlds: What Qwen-AgentWorld Changes About Agentic AI

Alibaba's Qwen team released the first language world model that simulates all seven agentic environments — browser, terminal, OS, mobile, search, IDE, and MCP — in a single unified model. Here is why that matters more than adding another benchmark score.

2026-06-25·7 min read
Agentic AI

LLMs for Ambiguity, Deterministic Agents for Policy: Multi-Agent Contract Compliance

A Google Cloud reference sample shows the right way to build agentic contract compliance: an LLM handles messy extraction, a deterministic agent enforces hard policy, and they hand off over the open A2A protocol — with an auditable verdict. Here is why that split matters for real estate.

2026-06-24·8 min read
AI Engineering

LLM Inference Optimization: The Line Item That Decides If Your AI Ships

In production, inference — not training — is where the money goes. A practical guide to the techniques that cut LLM serving cost 5-10x: KV-cache and PagedAttention, continuous batching, quantization, and speculative decoding — and why it decides whether AI features are economically viable at scale.

2026-06-29·9 min read
Agentic AI

Your Building Already Knows: Turning Property Telemetry into Agentic Action

The most valuable AI asset in real estate isn't a model — it's the operational telemetry your buildings already generate. Here is how to turn high-frequency metrics from commercial and residential properties into agentic action flows that sense, decide, act, and self-correct.

2026-06-25·8 min read
Agentic AI

Sakana Fugu: When Orchestrating Models Beats Owning One

Sakana AI's Fugu is a small model that conducts rival frontier LLMs instead of replacing them — and claims to beat Opus 4.8 and GPT-5.5 without training a frontier model of its own. Here is what the orchestration-over-ownership shift means for real estate AI, and the honest caveats.

2026-06-26·8 min read
AI Strategy

GLM-5.2 and the Open-Weight Tipping Point for Real Estate Data

Z.ai's GLM-5.2 is an MIT-licensed open-weight model performing near the closed-source frontier on coding and agentic tasks, at a fraction of the cost. Here is what self-hostable frontier-class AI changes for real estate data sovereignty — and the honest caveats.

2026-06-24·7 min read
Agentic AI

From Chatbot to Teammate: What Claude Tag Signals for Real Estate Ops

Anthropic's Claude Tag lets teams tag an AI agent into Slack as a governed, async teammate. Here is what the shift from prompting a chatbot to delegating to an agent means for real estate operations — and why governance, not the model, is the hard part.

2026-06-24·7 min read
Agentic AI

When AI Agents Rewrite Their Own Rules: Self-Improving Harnesses for Real Estate

A new framework called Self-Harness lets AI agents fix their own operating rules by analyzing their failures — no model retraining, up to 60% better on hard tasks. Here is how harness engineering works and why it is the next reliability lever for agentic AI in real estate and PropTech.

2026-06-23·8 min read
Agentic AI

Self-Improving AI Agents: Why Evolving Agentic Systems Win in Real Estate

Self-improving AI agents get better with every run — not by retraining the model, but by evolving their orchestration. Here is how the rollout-and-reflection loop works, and why it is the key to reliable agentic AI in real estate and PropTech.

2026-06-21·9 min read
Agentic AI

Building an Agentic Orchestration Layer for PropTech Platforms

How to move beyond isolated LLM features to a governed control plane that coordinates specialized AI agents across real estate workflows — with deterministic orchestration, typed tool contracts, and full observability.

2026-06-10·10 min read
Agentic AI

Agentic AI Best Practices: Shipping Reliable Agents in Production

Guardrails, human-in-the-loop approvals, evaluation, and cost control — the engineering discipline that separates a production-grade real estate agent layer from an impressive demo.

2026-06-03·8 min read
AI / ML

How AI Is Transforming Real Estate Asset Management in Europe

Discover how AI-powered platforms are revolutionizing portfolio management, reducing operational costs, and enabling data-driven decisions across European real estate markets.

2025-05-12·7 min read
SaaS Engineering

Building a PropTech SaaS Platform: From POC to $1M ARR in 9 Months

A deep dive into the engineering decisions, architecture choices, and team structure that took a real estate AI platform from proof-of-concept to $1M monthly recurring revenue.

2025-04-28·9 min read
Industry

Top 10 PropTech Companies in Europe and USA to Watch in 2025

An in-depth look at the leading PropTech innovators in Europe and the United States — their technology, business models, and the engineering challenges they're solving.

2025-04-10·8 min read
AI / ML

LLM Automation in Property Management: A $6.5M Cost Reduction Case Study

How a major European real estate provider used large language model automation to reduce total operating costs by $6.5 million — and the engineering approach that made it possible.

2025-03-22·8 min read
Engineering

Microservices Architecture for High-Scale Real Estate Data Platforms

A technical guide to designing microservices architecture for real estate SaaS platforms — covering data isolation, service boundaries, Kubernetes deployment, and Terraform IAC.

2025-03-05·10 min read
Engineering

ML-Powered Payment Gateway Optimization for PropTech Platforms

How machine learning cascade routing reduced payment gateway fluctuations by 40%, cut human error by 95%, and accelerated time-to-market by 50% for a real estate payment platform.

2025-02-18·7 min read
Team & Process

The Managed Capacity Model: How PropTech Companies Scale Engineering Teams

Why leading PropTech companies are replacing traditional T&M and fixed-price contracts with managed capacity models — and what it means for engineering delivery speed and cost predictability.

2025-01-30·6 min read
AI / ML

AI Predictive Maintenance for Real Estate: Reducing Downtime and Costs

How AI-powered outage prediction and anomaly detection systems are transforming property maintenance — with a case study from a natural resource extraction company managing industrial facilities.

2025-01-12·7 min read
Engineering

KPI-Driven Engineering Culture: How PropTech Leaders Measure What Matters

A framework for engineering KPIs in PropTech organizations — covering delivery velocity, quality metrics, DevOps performance, and how to align engineering metrics with business outcomes.

2024-12-18·6 min read
Case Study

From Startup to Award Winner: Engineering Lessons from Germany's #1 PropTech

The engineering decisions, team structures, and delivery practices that took a real estate AI platform from startup to Germany's #1 Asset & Portfolio Management Tool — lessons for every PropTech builder.

2024-11-25·8 min read