Memory
Memory is how Ordify learns about you and your organization over time. As you interact with agents, Ordify automatically extracts key facts — your role, preferences, communication style, and organizational context — and stores them as a Core Profile. Every agent you chat with can then use this profile to give more relevant, personalized responses without you repeating yourself.
Why use memory
Without memory, every conversation starts from scratch. You end up re-explaining your role, your team's tools, or how you like responses formatted. Memory solves this:
- Automatic learning: Ordify picks up facts from your conversations — no manual setup needed.
- Consistent personalization: Every agent knows your preferences, role, and context from the first message.
- Organization-wide context: Share company-level facts (tech stack, processes, goals) so all team members' agents understand the business.
- Always editable: You control what's remembered. View, edit, or delete any fact at any time.
How it works
Memory operates at two levels: user and organization.
User memory
You chat with agents
↓
Ordify analyzes the conversation
↓
Key facts are extracted with confidence scoring
↓
High-confidence facts are added to your Core Profile
↓
Your Core Profile is injected into every future agent conversation
After each conversation is summarized, Ordify scans the transcript and extracts facts that fall into six categories:
| Category | What it captures | Examples |
|---|---|---|
| Profile | Identity, role, job title, location | "Product manager in Berlin, GMT+1" |
| Preference | How you want the AI to behave | "Prefers concise answers under 200 words" |
| Behavior | Work habits and routines | "Reviews PRs every morning before standup" |
| Style | Tone, formatting, vocabulary | "Uses bullet points, avoids emojis" |
| Knowledge | Expertise and domains | "Expert in Python and data pipelines" |
| Workflow | Repeated procedures and checklists | "Always runs staging tests before deploying" |
Organization memory
Organization memory works the same way, but stores facts about your company rather than about you individually. It captures:
| Category | What it captures | Examples |
|---|---|---|
| Domain | Industry, products, customer segments | "B2B SaaS in the fintech space" |
| Tech stack | Platforms, languages, tools | "Uses React, Python, and PostgreSQL" |
| SOP | Processes, approval chains, schedules | "Releases ship every two weeks on Thursday" |
| Goal | OKRs, targets, roadmap themes | "Q2 goal: reduce churn by 15%" |
| Culture | Communication norms, values | "All-hands every Monday at 10am" |
| Environment | Offices, remote policy, compliance | "Fully remote, data hosted in EU" |
Organization facts require admin approval before they're added to the profile. When Ordify extracts a new organizational fact, it's queued as a pending suggestion that an admin can approve or reject.
Viewing and editing your memory
User memory
- Click on your profile and the gear icon to open Settings
- Select the Memory tab
- Your Core Profile appears as editable text
- Add, edit, or remove any fact
- Click Save Changes
Your Core Profile is a simple text organized by category. For example:
## Profile
- Product manager at Acme Corp
## Preference
- Prefers concise responses under 200 words
- Wants code examples in Python
## Workflow
- Reviews pull requests every morning
- Uses Linear for task tracking
Organization memory
- Navigate to Organization Settings
- Select the Memory tab (admin-only)
- The Organization Core Profile appears as editable text
- Review any pending facts suggested by the system:
- Facts to add appear with a green border
- Facts to remove appear with a red border
- Click Approve to accept a fact or Reject to discard it
- Click Save Changes
Non-admin members can view the organization profile but cannot edit it.
How agents use memory
When you start a conversation, Ordify injects your user profile and organization profile into the agent's context. The agent sees these facts as background knowledge and adapts its responses accordingly — adjusting tone, referencing your tools, and skipping information you already know.
This happens automatically. You don't need to configure anything per agent. Every agent in your workspace benefits from the same memory.
Use cases
- Onboarding: A new team member chats with agents for a week. By the end, every agent knows their role, tools, and preferences — no configuration needed.
- Consistent formatting: You prefer bullet points and short answers. Once memory captures this, every agent responds in your style.
- Organization context: Your company uses Jira, ships on Thursdays, and targets enterprise customers. Agents reference these facts when drafting plans, emails, or reports.
- Role-aware responses: A developer and a sales rep on the same team get different responses from the same agent — the developer sees code examples, the sales rep sees customer-facing language.
What's next
- What is an agent — understand how agents use memory as part of their context
- Create an agent — build an agent that benefits from memory
- Knowledge — add document-level knowledge on top of memory