Platform
MCP Server Build
A production MCP server for one of your systems, with up to 8 tools, secure sign-in, action logging and tests in AI…
Platform
AI Agent Memory System
Your agents remember prior sessions and conversations, scoped and retained by rules you set, not lost between calls.
An agent that forgets everything between conversations makes customers and staff repeat themselves. AI Agent Memory System gives your agents durable memory across sessions, scoped per user, thread or workflow, so they recall what matters. You set how long memory lasts and what must never be stored. Access controls keep one user's history away from another.
- Agents recall prior sessions instead of starting blank
- Retention rules decide how long memory lasts
- Memory scoped so one user's history stays private
$4,500
20–30 working days
Optional maintenance $249/mo
What it is
AI Agent Memory System is a build that gives your AI agents durable memory across sessions and conversations, storing what happened before under retention rules you define, so an agent recalls prior context instead of starting blank every time. Memory is scoped per user, thread or workflow, with access controls.
Why businesses buy it
Agents without memory feel clumsy. Customers re-explain their situation, staff paste in the same background, and useful context from last week is lost. Building memory badly is worse, because it can keep data too long or show it to the wrong person. After this build, your agents remember what they should, for as long as your rules allow, with scope and access tested before launch. The Scope Card sets the agents, retention rules and controls in writing before you buy, so you know exactly how memory will behave.
Problems this solves
Agents start every conversation from a blank slate
The Agent Memory System gives agents durable memory across sessions and conversations. They recall prior context instead of asking the same questions again or losing track of earlier steps.
Customers repeat details they already gave last time
What happened before is stored and made available to the agent in the next session. Memory can be scoped per user, thread or workflow, so each conversation sees only what belongs to it.
Stored memory raises questions about retention and access
Retention rules you define decide what is kept and for how long. Access controls limit which agents can read each memory, set as up to five controls.
A good fit if
Your agents currently lose all context between sessions, forcing users to repeat information they already gave.
You want memory scoped per user, per conversation or per workflow, not one shared pool everything writes into.
You have retention rules in mind, such as how long memory persists or what should never be stored.
You have a named technical owner who can supply cloud access and review the memory configuration before launch.
Not the right fit if
You need agents to reference your organization's documents and knowledge base, not their own session history. See AI Agent Knowledge Layer.
Your priority is deciding which model handles a request, not agent memory. See AI Model Router.
You need the full integration layer for routing requests across providers, not memory storage. See AI Gateway.
Your priority is versioning the prompts agents use, not what they remember. See AI Prompt Management.
How it works
- We map the sessions and conversations your agents currently handle, and where context is being lost between them today.
- Memory scope is defined: per user, per conversation thread, or per workflow, matching how your agents are actually used.
- Retention rules are configured, covering how long memory persists and what categories of information should never be stored.
- Agents are connected to the memory store, so a returning session recalls relevant prior context automatically, not through manual prompting.
- Access controls are set so memory scoped to one user or workflow isn't visible to another without explicit permission.
- A control test log documents memory behavior against real sessions, reviewed with your technical owner before launch.
What you provide within 72 hours
Your portal lists these the moment you order. Delivery starts the working day after everything is in and your access is checked.
- Written authorization confirming which agents and applications are in scope for persistent memory.
- A list of the AI applications and agents that should retain memory across sessions.
- Your retention rules, including how long memory should persist and what should never be stored.
- Cloud or hosting access needed to deploy the memory store alongside your existing infrastructure.
- A named technical owner who can review the memory configuration and sign off before launch.
- Admin access and API keys for every connected system, shared by adding our team as a user or creating a key you control, never by email.
- Your business information (company basics, hours and brand voice), shared during intake, and one decision-maker who can approve the final configuration.
The 72-hour rule. Send everything on your checklist, including access, within 72 hours of ordering so your build starts on time.
Timeline
Start
Intake verified, agents in scope and retention rules confirmed.
Weeks 1–2
Memory store built with scoping and retention rules configured.
Week 3
Agents connected to the memory store and access controls set.
Weeks 3–4
Control test log produced and shared with your technical owner.
Week 4
Your review and two rounds of changes applied to the configuration.
Go-live
Live, recorded handoff session, 30 days of support begins.
| Option | Delivery time | Price |
|---|---|---|
| Standard | 20–30 working days | $4,500 |
| Express | 15–25 working days | $5,300 |
| Rush | 10–20 working days | $6,100 |
What you receive at handover
- AI Agent Memory System configured and deployed for up to three agents or AI applications
- Memory scoped per user, thread or workflow, with retention rules you define
- Access controls on memory (up to five controls configured)
- A test report with a control test log for the agreed scenarios
- An admin guide and a recorded 60-minute handoff session
- 30 days of post-launch support
Expected benefits and how to measure them
- Can make agent conversations feel continuous rather than repetitive.
- Helps agents finish multi-step work across sessions.
- Keeps stored context under clear retention and access rules.
- Repeat questions per conversation
- Review a sample of conversations before and after launch and count questions asked twice.
- Task completion across sessions
- Track multi-session tasks finished without a restart, using agent logs.
- Memory past its retention date
- Check monthly that stored items older than your rules allow are removed.
Before you start: Review a sample of recent agent conversations and note where context was lost. Write down the retention rules you want applied.
These are things to track, not promised results. Outcomes depend on your data, your team and your market.
Running costs
The build price is one-time. Running the Memory System afterward costs roughly $20 to $300 a month in your own accounts, covering hosting, logging and monitoring tools, and it scales with how many sessions and how much retained context are stored.
AI model usage during the build and testing is billed to your own accounts too, so every provider relationship stays yours, not ours.
Ask about running costsCustom requirements
Need more than three systems in scope, more than five retention rules configured, or a second admin console? Add a note at checkout and you get a written Change Quote, only starting once you accept it. Larger or bespoke projects go through the Custom AI request form instead.
Request a custom versionKeep it running
For this build we recommend monthly maintenance, covering monitoring, fixes, and a monthly allowance of change hours, enough to adjust a retention rule or extend memory scope to a new agent. It starts once the 30 days of support end and can be cancelled any month.
$249/mo (the minimum fee, includes 2 change hours)
Optional. Starts after the post-launch support period. Cancel any month.
How maintenance worksHow this compares
| Do it yourself with a SaaS tool | Hire a custom agency | Upmark Labs | |
|---|---|---|---|
| Price | An agent memory platform charges a monthly fee per stored session volume | A systems integrator quotes per project after a discovery phase | Ours is one fixed price, shown on this page before you buy |
| Scope | A memory platform is infrastructure your engineers configure and maintain | An integrator agrees scope during the project | Ours is written on the Scope Card before purchase |
| Retention rules | A memory platform ships default retention settings you adjust yourself | An integrator build depends on who is staffed on it | Ours is configured against the retention rules you define |
| Delivery | A memory platform is live once your team finishes integrating it | An integrator usually takes several weeks to staff and start | Ours has a fixed delivery time shown before you buy, with faster options |
Questions about this product
How much does AI Agent Memory System cost?
The fixed price for each delivery speed is shown in the price and delivery panel on this page before you order. It covers everything on the Scope Card: memory scope for up to three agents or applications, retention rules, up to five controls and the control test log you review against real sessions before launch.
How is memory scoped between different users or workflows?
You choose the scope during intake: per user, per conversation thread, or per workflow, matching how your agents are actually used. Access controls are configured so memory scoped to one user or workflow isn't visible to another without explicit permission, and this is tested against real sessions before the system goes live.
What does AI Agent Memory System cost to run each month?
Roughly $20 to $300 a month in your own accounts, covering hosting, logging and monitoring tools such as Langfuse or Grafana. You pay providers directly, so there's no markup from us. Light session volume sits toward the bottom of that range, high-traffic agents with long retention toward the top.
How is AI Agent Memory System different from AI Agent Knowledge Layer?
Memory System stores what happened in an agent's own prior sessions and conversations, its experience over time. AI Agent Knowledge Layer gives agents structured access to your organization's documents and knowledge base, which is reference material rather than session history. The two fit together, giving agents both a memory and a library.
Can we set how long memory is retained?
Yes. Retention rules are defined during intake, covering how long memory persists before it expires and what categories of information should never be stored at all. These rules are configured into the memory store as part of the build and documented in the control test log you review before launch.
Can we add more agents to the memory system later?
Yes. The standard build covers up to three systems or AI applications, and adding a fourth or beyond is quoted as a Change Quote once you are ready to expand. We suggest starting with your highest-traffic agent, confirming scoping and retention behave correctly, then extending memory to further agents.
What do we need to provide before AI Agent Memory System can launch?
Written authorization for the agents and applications in scope, a list of the agents that should retain memory, and your retention rules on duration and excluded data categories. We also need cloud access, a named technical owner, and admin access shared by adding our team as a user or creating a key you control, never sent by email.
Can I cancel maintenance for AI Agent Memory System?
Yes, any month. Maintenance starts after the 30 days of post-launch support end and runs month to month with no cancellation fee or lock-in. It includes a monthly allowance of change hours that scales with the build, and unused hours in a given month don't carry over to the next one.
Stop making users repeat themselves every session.
Order AI Agent Memory System and your agents retain scoped, rule-bound context on a fixed delivery date. The scope above is what you get, in writing, before you pay.
Works well with
Platform
AI Agent Orchestration System
Coordinates multiple agents, tools and handoffs in one runtime.
