Large language model access and AI-assisted workflows.
Key Functions
| Function | Description | AI Opportunity |
|---|
| Chat Interface | Conversational AI interaction | Core function |
| Context Management | Conversation history, memory | Long-term memory |
| File Analysis | Document, image, code understanding | Multimodal |
| Code Generation | Write, explain, debug code | Agentic coding |
| Writing Assistance | Draft, edit, summarize content | Style adaptation |
| Research | Search, synthesize information | Real-time data |
| API Access | Programmatic model access | Custom integration |
| Team Collaboration | Shared conversations, workspaces | Knowledge capture |
| Custom Instructions | Personalized behavior | Learning preferences |
| Tool Use | Web browsing, code execution | Autonomous agents |
Core Entities
| Entity | Fields | Volume | Sensitivity |
|---|
| Conversations | messages, timestamps, model, tokens | Very High | High |
| Files | uploaded documents, images, code | High | Variable |
| Custom Instructions | system prompts, preferences | Low | Medium |
| API Keys | tokens, permissions, usage limits | Low | High |
| Usage Metrics | tokens, costs, model usage | High | Low |
| Shared Workspaces | team conversations, permissions | Medium | High |
| Plugins/Tools | enabled integrations, configs | Low | Medium |
| Memory | learned facts, preferences | Medium | High |
Integration Points
| System | Data Flow | Direction |
|---|
| IDE/Code Editors | Code context, completions | Bi-directional |
| Documents | Content analysis | Inbound |
| Web Browsing | Search, page content | Inbound |
| APIs | Custom tool execution | Bi-directional |
| Slack/Teams | Bot integrations | Bi-directional |
| Automation | Workflow triggers | Bi-directional |
Data Retention
| Data Type | Typical Retention | Compliance Driver |
|---|
| Conversation history | 30 days - indefinite | User preference |
| Uploaded files | Session - 30 days | Privacy policy |
| Usage logs | 30-90 days | Billing/audit |
| API logs | 30 days | Debugging |
Evaluation Criteria
| Criteria | Weight | Notes |
|---|
| Model quality | High | Reasoning, accuracy |
| Context length | High | Complex tasks |
| Speed | Medium | Latency matters |
| Privacy/data handling | High | Enterprise concern |
| API availability | Medium | Integration needs |
| Pricing | Medium | Token economics |
| Tool/plugin ecosystem | Medium | Extensibility |
Market Leaders
| Product | Strength | Best For |
|---|
| Claude | Reasoning, safety, long context | Complex analysis |
| ChatGPT | Ecosystem, plugins, familiarity | General use |
| Perplexity | Search integration, citations | Research |
| Gemini | Google integration, multimodal | Google ecosystem |
| Copilot | Microsoft integration, enterprise | Microsoft shops |
AI Disruption Potential
This IS the disruption layer. The question is which interfaces and models win.
| Function | Current State | 2027 Projection |
|---|
| Context length | 100K-200K tokens | 1M+ tokens |
| Tool use | Basic function calling | Autonomous agents |
| Memory | Session-based | Persistent learning |
| Multimodal | Text + images | Full sensory |
| Reasoning | Chain of thought | Multi-agent systems |
Build vs Buy: Buy the models, build the interface if you need custom workflows. Most should use existing interfaces unless AI is core to product.
Questions
Which engineering decision related to this topic has the highest switching cost once made — and how do you make it well with incomplete information?
- At what scale or complexity level does the right answer to this topic change significantly?
- How does the introduction of AI-native workflows change the conventional wisdom about this technology?
- Which anti-pattern in this area is most commonly introduced by developers who know enough to be dangerous but not enough to know what they don't know?