Introduction
Technology radars help software teams spot emerging trends and make informed decisions. The 34th edition of Thoughtworks' Technology Radar — published just last week — packs 118 blips covering tools, techniques, platforms, and languages. This guide shows you how to turn that radar into a concrete action plan for your organization. You'll learn to interpret AI-centric blips, revisit foundational practices, address agent security, and apply harness engineering concepts. By the end, you'll have a repeatable process for extracting value from each biannual release.
What You Need
- Access to the Radar – The full 34th edition PDF or interactive web version (available on Thoughtworks' site).
- A team or stakeholder group – At least 3–5 people familiar with your current tech stack and goals.
- Note-taking tool – A shared document, wiki, or Miro board to capture insights.
- Prioritization framework – A simple matrix (e.g., effort vs. impact) to sort blips.
- Time commitment – 2–4 hours for initial review and discussion.
Step-by-Step Guide
Step 1: Understand the Radar's Purpose and Scope
Before diving into blips, orient yourself. The Technology Radar is a biannual survey capturing Thoughtworks' hands-on experience across client engagements. Each blip offers a brief, opinionated take — it is not an exhaustive review. In this edition, AI topics dominate, but the radar also revisits established techniques like pair programming, zero trust architecture, mutation testing, and DORA metrics. This balance is intentional: AI is accelerating complexity, so the radar calls for a counterweight of software craftsmanship.
- Skim the radar's introduction and theme sections.
- Note the four quadrants: Techniques, Tools, Platforms, and Languages & Frameworks.
- Recognize that blips are classified as Adopt, Trial, Assess, or Hold.
Step 2: Identify AI-Oriented Blips and Their Implications
The radar's leading theme is AI. Many blips examine familiar tools and techniques through an LLM-assisted lens. For example, agents like OpenClaw and Claude Cowork are highlighted for supervising real work tasks, while Gas Town coordinates agent swarms across entire codebases. These blips indicate where the industry is heading — but also raise questions about control, security, and cost.
- Filter the radar to show only blips in the “AI / ML” category (or manually tag them).
- For each AI blip, ask: What problem does this solve? What risks does it introduce?
- Pay special attention to blips marked Assess or Trial — these are areas where you might experiment.
Step 3: Revisit Foundational Techniques as a Counterbalance
Assembling this edition, Thoughtworks found themselves returning to core practices. AI tools can generate code quickly, but they also generate complexity. The radar highlights the need for clean code, deliberate design, testability, and accessibility as a first-class concern. Interestingly, the command line is experiencing a resurgence: agentic tools are pushing developers back to the terminal.
- Cross‑reference the foundational blips (e.g., pair programming, zero trust architecture, mutation testing) with your current practices.
- Identify any gaps — for instance, if you’ve abandoned test-driven development, consider doubling down.
- Schedule a discussion around how to reinforce these fundamentals alongside AI adoption.
Step 4: Address Security Concerns with Permission-Hungry Agents
One of the most important themes is the security risk of “permission hungry” agents. As Jim Gumbley — now on the radar writing team — emphasises, agents that manage real work require broad access to private data, external communication, and live systems. This appetite collides with unsolved problems like prompt injection, where models cannot reliably distinguish trusted instructions from untrusted input.
- List any AI agents or LLM integrations your team is using or planning to use.
- Apply a security review to each: what permissions does it need? Can you limit them?
- Investigate the radar’s blips on threat modeling and harness engineering — they often suggest sensors and guardrails.
Step 5: Apply Harness Engineering Concepts to Your Workflow
Harness engineering is a major source of ideas in this edition. The radar includes several blips about guides and sensors needed for a well-fitting harness — a framework that lets you safely introduce tools (especially AI) with controlled risk. As Birgitta’s excellent article on the subject explains, harness engineering provides the structure to experiment without breaking production.
- Define the “harness” for your environment: feature flags, canary deployments, monitoring dashboards, and fallback plans.
- Map each blip classified as “harness” to a concrete action item (e.g., “implement a guardrails API”, “set up anomaly detection”).
- Create a prioritized backlog of harness improvements to run alongside your AI experiments.
Tips for Maximizing the Radar
- Don’t adopt everything. Blips are suggestions, not mandates. Pick 3–5 that align with your strategic goals.
- Group similar blips. Many blips (e.g., “Permission Hungry Agents”, “Prompt Injection Prevention”) belong together — treat them as a package.
- Involve security early. Use the radar’s security blips to build a risk-aware culture before agents go live.
- Revisit periodically. The radar changes every six months. Archive your notes and update your priority list.
- Share with the team. A shared understanding of trends helps everyone make better technical decisions.