// WHAT_AGENTS_KNOW

AGENT MEMORY

Your organisation knows more than your agents can find. The reasoning behind decisions, the knowledge your analysts have built, the context your teams carry in their heads — scattered across emails, buried in documents, or walking out the door when people leave.

Documents are what survived. Reasoning is what mattered.

// THE_IDEA

A context graph that connects facts, reasoning, and decisions — and remembers how they relate. Ask “why did we choose Vendor X?” and get the reasoning. Ask “what do we know about Y?” and get everything we've learned. Not a search engine. Not a document store. A knowledge system that your AI agents can reason with. We call it Alexandria.

[ TALK_TO_US → ]

// THE_ACCELERATOR

Alexandria

Alexandria is our agent memory accelerator — not a product you buy off the shelf, but a proven architecture we customise and integrate into your stack as part of a delivery engagement. We adapt it to your data sources, your access controls, and your agent workflows, so your agents reason with your organisation's full context from day one.

Provenance

Every claim traces back to a source document, email, or conversation. Nothing is asserted without evidence.

Permissions

Graph nodes inherit access controls from source systems. If you couldn't see the original, you can't see the reasoning.

Time-aware

Decisions are versioned. The graph knows you chose Vendor X in 2023 for reason Y, and switched in 2025 when Y changed.

Compounding

The graph grows with every interaction. Your agents get smarter every week. When people leave, their knowledge stays.

// THE_DIFFERENCE

Same Question. Different Intelligence.

// QUERY

“What's our policy on customer data retention?”

Typical RAG

Customer data is retained for 7 years in line with regulatory requirements. Data is stored in the EU-West region and encrypted at rest using AES-256.

Source: Data_Retention_Policy_v4.2.pdf

Technically correct. But it was 5 years until last quarter. Why did it change? Who approved it? Was it a regulatory requirement or a legal recommendation? RAG doesn't know — it only sees the current document.

Alexandria

The current policy is 7 years, effective from January 2026.

This changed from 5 years following advice from external counsel (Baker McKenzie, Nov 2025) that the original period was insufficient under the updated FCA guidelines for financial services firms.

The change was approved by the DPO and CFO in December 2025. The CFO initially pushed back on storage costs — resolved when Infrastructure confirmed the incremental cost was under £12k/year.

Sources: Baker McKenzie memo (8 Nov 2025), DPO ↔ CFO email thread (3 Dec), Policy change request #4471, Infrastructure cost estimate

Previous versions: 5 years (2021–2025), 3 years (2018–2021)

Your agents are only as good as what they remember.

Build Your
Agent Memory

We customise and integrate Alexandria into your agent stack — adapted to your data sources, your permissions model, and your workflows.

01

Seed

Ingest your existing knowledge — documents, emails, meeting notes, interviews. Build the initial context graph. 4-8 weeks.

02

Connect

Wire Alexandria into your agents and tools. Your team starts querying immediately. Your agents start reasoning with context. 8-12 weeks.

03

Grow

The graph grows with every decision, every interaction. Your agents get smarter every week. When people leave, their knowledge doesn't. Ongoing.

[ TALK_TO_AN_ENGINEER → ]

What you get

  • A context graph seeded from your existing knowledge
  • Agents that reason with your organisation's full context
  • Provenance on every claim — traceable back to source
  • Permissions inherited from your existing systems
  • A system that compounds — smarter every week it runs
Barnacle Labs
Barnacle_Labs

AI for breakthroughs, not buzzwords.

34 Tavistock Street, Covent Garden, London WC2E 7PB

Google Cloud Partner
  • Barnacle Labs Ltd. England & Wales.
  • Company No. 14427097
  • © 2026 Barnacle Labs Ltd.