From stateless chatbots to memory-driven sales agents: how DealMind was built
Kanvitha · reddit · 2026-09-29
The author's team built DealMind, a sales agent prototype using Hindsight as a persistent memory layer to retain and recall customer objections, stakeholder concerns, competitor info, preferences, and past commitments. The workflow: recall relevant account history → feed context to the agent → generate a response → retain new information. Built with Python, Streamlit, Hindsight, and Groq; the core challenge was designing the memory workflow so each interaction compounds into future ones. A Medium write-up details the architecture.
More from coding & agent
- Open-Source P(doom) Music Video Fully Built With Claude Code — StefanoGogioso · 2026-09-30
- Twin launches unlimited agent plans from $29/mo, drops credit system entirely — socialwithaayan · 2026-09-30
- Celesto Launches GitHub Actions Runners, Claims 12x Cheaper Than GitHub — aniketmaurya · 2026-09-30
- Wasmer's Pi runs AI agents unmodified in the browser and on iPhone via WebAssembly — JosephJacks_ · 2026-09-30
- RecursiveMAS: Agent swarms share latent thoughts like looped Transformers, cutting tokens by 75% — james_y_zou · 2026-09-30
- When code review needs memory: adding persistent context with Hindsight — Ambitious_Menu1385 · 2026-09-30