Merge's Gil Feig on context graphs: semantic search synced, live APIs for lookups
shensi · x · 2026-10-15
At AI Engineer World's Fair, Merge API co-founder Gil Feig walked through a practical architecture for agent data access. Asking an agent "which customers were upset last week?" fails with a few MCP servers — live API lookups can't search across that much historical data.
His solution: build a context graph that syncs data for semantic search while keeping live API calls for simple lookups, letting each path handle what it's good at. Directly useful for developers building enterprise agents over customer data.
Related event: Merge's Gil Feig: Why Companies Need a Context Graph, Not Just MCP Servers(2 posts)→
More from coding & agent
- AI agents hammered Amap with 2,600 scrapes, forged tokens, and even read coverage of themselves — dotey · 2026-10-15
- trq212: being technical really means going deep on the right details at the right level — trq212 · 2026-10-15
- Magicare AI cuts nursing admissions to under a minute, slices inference costs 33% on Baseten — baseten · 2026-10-15
- Codex tip: you can ask it to find a specific past conversation in a new conversation — burkov · 2026-10-15
- Watchdog blocks all 609 prompt injections by checking every agent tool call against the mission — jagga_tech · 2026-10-15
- dotey's multi-agent dev workflow: git worktree isolation, agents resolve merge conflicts via rebase rule — dotey · 2026-10-15