As of 13 August 2026, AI can only partly choose the best programming language for your business app.

What to actually do

1. Use a tool built for this

   The route this page recommends

2. Do it yourself

   Second choice

   A chat interface, chat-fluent skill, and roughly 1 hour until you can act on the result.

   How to actually do it

   1. Open a document for the decision and record the app's purpose, users, platforms, essential features, integrations, data types and expected usage.
   2. Ask your technical lead or developer to add the existing languages, frameworks, hosting arrangements, deployment process, team skills and likely maintenance constraints.
   3. Paste the completed information into the prompt and run it in a chatbot, asking it to identify missing facts rather than filling gaps with assumptions.
   4. Compare the shortlist against your actual requirements, existing systems, UK GDPR responsibilities, security needs, hiring plans and operational support capacity.
   5. Ask the model to turn the recommended language and strongest alternative into small technical-spike plans, then have your developer implement the same representative feature in each option.
   6. Measure the spikes against agreed criteria such as integration difficulty, testability, deployment, security controls and maintainability, and record the evidence rather than accepting the model's predictions.
   7. Send the recommendation, assumptions, spike results and unresolved risks to the person accountable for delivery, and record the final language choice and reasons in the decision document.

   Prompt

   Act as a software architect helping me choose a programming language for a business app. Make a recommendation, but do not present it as certain and do not invent facts about tools, frameworks, costs or performance.
   
   Use only the information below. If important information is missing, list the missing decision and explain how it could change the recommendation.
   
   Business purpose: [describe what the app does]
   Users and roles: [describe who will use it and their permissions]
   Platforms: [web, mobile, desktop, or other]
   Core features: [list the essential features]
   Data and integrations: [databases, payment services, APIs, identity systems and other integrations]
   Expected usage: [estimated users, traffic patterns and data volume, or say unknown]
   Security and compliance needs: [UK GDPR, sensitive data, audit requirements or none known]
   Existing systems: [languages, frameworks, hosting and deployment systems already used]
   Team skills: [languages and technologies the team can maintain]
   Delivery constraints: [deadline, budget, hiring constraints and need for rapid iteration]
   Operational needs: [availability, offline use, background jobs, reporting and support requirements]
   
   Compare no more than four suitable language options. For each option, explain its fit, important libraries or ecosystem considerations, hiring and maintenance implications, operational risks, and what would make it a poor choice. Separate facts from assumptions. Then provide:
   1. a recommended language and the conditions behind that recommendation;
   2. the strongest alternative and when to choose it instead;
   3. the assumptions that need confirming;
   4. a short discovery plan using documentation checks, a technical spike and a security review;
   5. questions I should take to my technical lead before making the decision.
   
   Do not write production code. Do not claim that one language is universally best. Keep the recommendation tied to the requirements above.

   Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.

3. Hand it to a person

   The distant third

   A person who owns the outcome does this end to end, worth it when the failure is dear.