docs: add project roadmap

Author: nirbhays

ROADMAP.md

@@ -0,0 +1,38 @@
+# TokenMeter Roadmap
+
+## Vision
+Be the de-facto observability layer for LLM applications — as essential as logging is for traditional services.
+
+## ✅ Shipped
+- One-line drop-in integration (replace a single import)
+- Real-time cost + token + latency tracking
+- Smart model routing (up to 60% cost savings)
+- OpenAI, Anthropic, Google Gemini support
+- `<5ms` overhead
+
+## 🔨 In Progress
+- [ ] Mistral AI provider support
+- [ ] LangChain callback integration
+- [ ] CSV / JSON cost export
+
+## 📋 Planned — Q2 2025
+- [ ] Web dashboard for cost visualisation
+- [ ] Per-feature / per-user cost attribution tags
+- [ ] Budget alerts (Slack / email when threshold hit)
+- [ ] OpenAI Assistants API support
+- [ ] Streaming response cost tracking
+
+## 📋 Planned — Q3 2025
+- [ ] Ollama (local model) cost estimation
+- [ ] Team / org cost sharing and allocation
+- [ ] CI budget gates — fail PRs that exceed cost thresholds
+- [ ] AWS Bedrock provider support
+
+## 💡 Under Consideration
+- Grafana dashboard plugin
+- OpenTelemetry metrics export
+- Cohere provider support
+- LlamaIndex integration
+
+## Contributing
+See [CONTRIBUTING.md](CONTRIBUTING.md). Provider integrations are a great first contribution.