Salesforce Agentforce / Einstein
https://www.salesforce.com/artificial-intelligence/
Agentforce agents grounded in Data Cloud — unified customer record across silos. Memory is the CRM record made available to AI; not an independent memory system. AI reads/writes through the existing data model.
At a glance
- Type
- Customer 360 + Data Cloud as context layer
- Tier
- T2
- Created
- 1999 (Salesforce founded 1999; Einstein AI launched 2016; Agentforce launched 2024)
- Latest release
- not applicable — not OSS
- License
- not applicable — not OSS
- GitHub
- not applicable — no GitHub repo
- Pricing
- searched not found
- Funding
- Public company (NYSE:CRM)
Taxonomy
- storage
- relational
- retrieval
- similarity
- persistence
- long-term
- update
- overwrite
- unit
- fact
- governance
- auditable
- conflict
- pii
When to use
Optimised for: cross-channel customer graph + agent handoff + CRM integration
Anti-fit: not for non-customer-facing use cases
Pros & cons
Pros
Memory grounded in Salesforce CRM data — customer / opportunity / case context is structurally first-class.
Cons
Salesforce-only; pricing tied to Einstein / Agentforce tier; ecosystem dependency.
Claims & capabilities
Dominant CRM market position. Agentforce GA'd 2024.
Technical surface
- API surface
- REST, SOAP, SDK: many
- Backend storage
- custom (Salesforce platform)
- Deployment
- searched not found
- Embedding model
- locked
- Multi-tenancy
- Hyperforce: dedicated functional domain per tenant; multi-tenant with isolation + zero-trust controls; no LLM training on customer data
- MCP
- via official adapter — Salesforce MCP
- A2A
- no A2A protocol support advertised — vertical product, no A2A integration documented
- OpenTelemetry
- first-class — Hyperforce + Datadog + OTel
Similar systems
Other vertical / domain-specific ai memory in the catalog, ranked by inbound references.
- NVIDIA ReMEmbR T3
Builds long-horizon memory by captioning video segments with VILA, storing captions with timestamps + 3D position coordinates in MilvusDB. At query time, LLM iterates retrieval across text, time, and position modalities. Deployed on Nova Carter robot (Jetson Orin).
- Abridge T1
Clinician-assist ambient documentation. Source mapping: every AI-generated summary element traced back to the source utterance. Audit-and-trust layer over episodic memory. Built on proprietary 1.5M+ medical-encounter dataset.
- ASAPP GenerativeAgent T1
Treats memory as first-class architecture. Captures the digital footprint of every interaction; retrieves preference and history at engagement time. Public example: airline knowing a frequent flyer wants aisle seats with her son — preference-aware, not just history-lookup.
- BenevolentAI T1
Target identification / drug repurposing / mechanism tracing. 85+ data sources, petabyte-scale, rebuilt every few weeks. Wet-lab results re-enter the graph and shift downstream predictions — institutional experimental memory closing a feedback loop.
- Causaly T1
Drug discovery / target identification / causal mechanism tracing. The graph is the memory: 7 years of curated biomedical cause-effect relationships compounding with each new ingestion. Scientific RAG retrieves from a versioned causal substrate.
- Character.ai T1
Chat Memories (user-defined facts), auto-memories for c.ai+ subscribers, pinned memories, in-context retention. PipSqueak 2 model (April 2026) reduces in-conversation drift. Memory Visualization meter forthcoming.