Relational Memory for AI

Your AI finally thinks
the way you do

You have a mental model of your business—contacts, projects, decisions, relationships. Describe it once, and your AI remembers it forever.

See the difference
5 min setupWorks with Claude, ChatGPT, CursorNo credit card required

Facts, not guesses

Typical memory solutions give you summaries and hedging. Relational memory gives you exact values, real names, and actual dates.

With typical memory solutions

"What's our pipeline with Acme Corp?"

"Based on the meeting documents, it appears to be a significant opportunity... The team seems to be working on next steps..."

"Appears to be" — uncertain
"Seems to be" — hedging
No specific numbers or dates
With Xtended

"Show me deals closing this month"

Acme Corp ($450K) is in final negotiations. Contact: Sarah Chen. Due: April 1st.

Exact values, not summaries
Real names, real dates
Facts, not guesses
See the full comparison
5 min
to describe your structure
Forever
of queryable memory

Tell your AI how you think about your business once. It creates the structure, remembers every detail, and answers questions instantly—forever.

Why relational memory?

Vector search can't represent your business structure. Graph databases make you learn their language. Relational memory lets you describe what you're thinking.

Vector DBs

Semantic search, fuzzy recall

Can't represent your mental model
No precise field queries
Results are probabilistic
Can't aggregate or count

Best for: Finding similar content

Graph DBs

Entities and connections

Makes you learn their language
Overkill for simple structures
Steep learning curve
Complex to query

Best for: Complex relationship traversal

Relational Memory

Structure that matches how you think

Describe it naturally, AI structures it
Exact queries, real answers
Aggregate, filter, join
Human-readable, always

Best for: Persistent agent memory

How it works

Describe → Structure → Query. Your AI handles the database, you get the answers.

Describe your world

Tell your AI what you want to track, in plain English. It creates the structure.

"I need to track my sales pipeline—contacts, companies, deal stages, and follow-up dates"

AI structures it

Tables, fields, relationships—all created automatically to match your mental model.

Contacts → Companies, Deals → Contacts, automatic date tracking

Query for insights

Ask questions, run aggregations, get real business intelligence. Not just retrieval.

"Total pipeline value by stage" or "Contacts I haven't talked to in 30 days"

Query for insights, not just retrieval

This isn't a fancy filing cabinet. Run real queries, get real business intelligence.

Aggregations

Sum, count, average, group by. Real business intelligence, not just lookups.

"Total revenue by quarter" or "Average deal size by industry"

Complex Queries

Filter by any field, join across tables, combine conditions.

"Deals over $50K closing this month where we haven't sent a proposal"

Export & Integrate

Pull data into spreadsheets, reports, or other tools.

"Export my Q4 pipeline to CSV" or send to Notion, Slack

Works everywhere via MCP

Connect once, use in any AI tool. Same memory, same structure, everywhere.

MCP Operations
list_tablesSee all your structures
describe_tableGet schema and relationships
create_tableDefine new structures
create_recordAdd data
list_recordsQuery with filters
searchFull-text search across everything

Works with

Claude
ChatGPT
Cursor

+ any MCP-compatible client

What will you stop re-explaining?

Any structured knowledge your AI should remember. Describe it once, query it forever.

CRM

Your AI remembers every contact, company, and interaction. Query your relationships.

"Who at Acme Corp should I follow up with this week?"

Project Tracking

Tasks, milestones, blockers. Ask what's overdue, what's at risk, what's next.

"What's blocking the Q1 release?"

Decision Logs

Record decisions with context. Never lose the 'why' behind choices.

"Why did we choose Postgres over MongoDB?"

Research & Notes

Structured notes linked to sources, topics, projects. Actually find things later.

"Summarize my notes on competitor pricing from last quarter"

See your data, not just query it

A clean interface to browse tables, view records, and understand what your AI remembers.

Deals

4 records · $183K total

companyvaluestageclose_date
Acme Corp$65KNegotiation
Jan 28Closing soon
TechStart$45KProposal
Feb 3
DataFlow$38KNegotiation
Jan 31Closing soon
CloudSync$35KDiscovery
Feb 15
Total pipeline$183,000

Stop re-explaining your business

5 minutes to describe your world. Forever of instant answers.

Free tier includedNo credit cardMCP config provided