Storage

Files and memory that persist between agent runs

Keep the documents an agent produces and the facts and decisions it saves, so each run starts from what the last one found.

Storage

Files

reports/q3.mdJust now
exports/accounts.csv2 days ago
contracts/northwind.pdfLast month

Memories

Northwind renews in MarchJust now
Fiscal year starts in FebruaryLast week
Fernhill prefers email3 weeks ago

Renewals

Run 1

Write reports/q3.md
Save what it learned

Month-End Close

Run 2

Read reports/q3.md
Recall what Renewals saved

Northwind renews in March, so its invoice moves to Q1.

Use cases

What you can build with storage.

  • Persist working files

    Keep documents, datasets and reports after the sandbox that made them stops.

  • Remember across runs

    Save facts and decisions as memories that the next run can retrieve.

  • Resume where you left off

    Let a later task read an earlier result instead of rebuilding it from scratch.

How it works

Start each account review where the last ended.

Keep the files and decisions future reviews depend on, not just the conversation.

  1. Save the source material

    Store the approved account exports and documents the agent will analyze.

  2. Preserve the result

    Write the account summary and report to storage when the task completes.

  3. Build on it next time

    The next run compares new data with the last report and saves an update.

storage.ts
import { Adapt } from '@adapt/sdk'

const store = new Adapt().storage

await store.files.write('reports/q3.md', report)
await store.memories.add('Fiscal year starts in February')

const last = await store.files.read('reports/q3.md')
Illustrative API · final SDK may differ

Pricing

What it costs to run.

Pay for what you keep: files by the GB-month, memories by the thousand.

  • $0.09per GB-month of files
  • $0.50per 1,000 memories a month

Works with

Use it with the rest of Adapt.

The company agent combines them with your tools, knowledge and team’s apps.

Start building with Storage