1
Your sources
Files, pages, recordings, notes — whatever your agent should remember.
Born working
Your business knowledge and AI, together.
Thyme RAG Capsules organize your documents, recordings and notes into reusable capsules. Sage AI answers from them. Run both on Cloudflare, or privately on hardware you own, and connect Claude, ChatGPT or Gemini through authenticated MCP and APIs.
Maple Harbor Guide
2 sources · 14 passages
Connect
…/c/maple-harbor/mcp
ClaudeChatGPTGemini
How it works
One pipeline, from a folder of files to a grounded answer.
A capsule is memory your agents read from and write to. The stack that builds it — the embeddings and the index — is disposable and rebuildable. The corpus, manifest, and eval set are the permanent asset.
2
Build
We distill the facts, embed them, and index them into a namespace that is yours alone.
Disposable — rebuildable at any time.
3
The capsule
corpusmanifesteval set
The permanent asset. Agents read and write — every write carries provenance and can be reverted.
4
Any agent
Ask, remember, recall — grounded answers with citations back to your sources.
Score each version against your own eval set, and export the whole capsule — rebuild the stack on any vector database, embedding model, or cloud.Read the open capsule spec →
Pair it with any model
Your capsules, with the AI you choose.
A Thyme RAG Capsule does the retrieval. The model you pick writes the answer from the passages it finds. Choose per capsule and switch at any time. Your knowledge stays the same.
Sage Cloud · available
Powered by Cloudflare
Assistants on Cloudflare Workers AI models, on the same network as your capsule. Questions and passages stay on Cloudflare. Pick from 25 models.
Compatible AI
Claude · ChatGPT
Claude from Anthropic or ChatGPT from OpenAI writes the answer. Each question sends that lab the passages retrieved for it, never your whole library.
Sage Private · scoped pilot
Custom cloud or on-prem
Models configured for your organization in a custom cloud deployment or on hardware you own, with optional fine-tuning scoped separately. Each deployment is assessed and validated before it answers.
Two ways to run it
Managed on Cloudflare, or private on your hardware.
Both use the same capsule format: corpus, manifest and eval set. Moving between them rebuilds only the index.
Cloudflare-hosted · available
Managed by Retenta on Cloudflare
Self-serve and running in minutes. Retrieval is Retenta's own pipeline on Cloudflare: documents in R2, Workers AI embeddings, and a Vectorize index for each capsule.
- Sign up and build today
- Free Developer tier
- Pairs with Sage Cloud, Claude, ChatGPT or any MCP client
Private · scoped per customer
On hardware you own
A custom retrieval deployment on servers you own, scoped to your workload. Retenta can help you buy and install the hardware. Each deployment is assessed, quoted and validated before it goes live.
- Requirements intake, assessment, then a quote
- Hardware purchase and installation facilitated
- Validated on your workload before activation
Request a private deployment
Tell us about your corpus, tasks and network. A private customer is never enrolled in the cloud by default; a cloud trial is a separate choice.
Beyond RAG-as-a-Service
A capsule is memory your agents write to — not a read-only search box.
RAG-as-a-Service answers “search my docs” — a read-only pipeline welded to one provider. A Retenta capsule is memory an agent both reads from and writes to: every write is provenanced and revertible, every version is scored, and the corpus is portable across any vector database, embedding model, or cloud.
Write API
Agents read and write. Every write carries provenance — agent, session, source — and a revert leaves an append-only tombstone in the audit log.
Versions you can score
Write an eval set once and grade any capsule version against it with an LLM judge, so you see the score before it ships.
Quality gates
Set a threshold, score every source snapshot, and keep below-gate snapshots out of the promoted release history.
Versioning
Diff recorded source snapshots and pin the source set retrieval may use. Immutable historical file revisions are next on the roadmap.
Portable export
Export the whole capsule — corpus, manifest, evals, write log — and rebuild it on any vector database, embedding model, or cloud.
MCP + REST per capsule
A first-class MCP server and REST API for every capsule, single-sourced from one tool spec so they never drift.
Connect in seconds
One call loads the whole memory spine.
The same tools over MCP or REST, single-sourced so they can’t drift. Recall the full state, append ordered progress, and remember findings — from Claude, ChatGPT, Gemini, or any headless harness.
recallrememberasksearchlog_progress
# Load the whole memory spine in one call
curl -H "Authorization: Bearer rk_live_…" \
…/c/maple-harbor/recall \
-d '{"query":"nightly triage"}'
# Write a fact back — provenance stamped, revertible
curl …/c/maple-harbor/remember \
-d '{"text":"CI is green tonight","topic":"ci-status"}'Pricing
Start with a capped trial. Scale with clear limits.
Developer includes one capsule, 10k stored vectors, a one-time 10k indexing grant and 1k monthly queries. Paid plans renew indexing and query capacity monthly. Core allowances stop at a hard limit unless you upgrade. Hosted AI uses prepaid credits, so unfunded model work is refused instead of becoming a surprise bill. Your stored data stays yours.
Every plan includes MCP + REST, writable memory with rollback, eval scoring, version pinning, and portable export. Plans scale by capacity and support; Team and Scale also include multi-user workspaces, role controls, and the CI quality-gate endpoint.
Developer
$0
Capped trial · no card
1 active capsule included · 10k stored vectors
100 sources · 250 MiB raw
10k vectors indexed once
1,000 queries /mo
200 analysis retrievals /mo
- Core memory + governance included
- Community support
Pro
$29/mo
$319/yr · 1 month free
5 active capsules included · 100k stored vectors
2,000 sources · 5 GiB raw
100k vectors indexed /mo
25,000 queries /mo
5,000 analysis retrievals /mo
- Core memory + governance included
- Email support
Team
$99/mo
$1,089/yr · 1 month free
15 active capsules included · 400k stored vectors
10,000 sources · 25 GiB raw
400k vectors indexed /mo
100,000 queries /mo
20,000 analysis retrievals /mo
- Core memory + governance included
- Multi-user workspace and role controls
- CI quality-gate endpoint
- Priority support
Scale
$399/mo
$4,389/yr · 1 month free
30 active capsules included · 1.5M stored vectors
50,000 sources · 100 GiB raw
1.5M vectors indexed /mo
500,000 queries /mo
100,000 analysis retrievals /mo
- Core memory + governance included
- Multi-user workspace and role controls
- CI quality-gate endpoint
- Priority support
FAQ
Questions, answered.
What can I put in a capsule?
PDF, Word, Excel, CSV, HTML, text and markdown, source code and notebooks, plus audio (.mp3, .m4a, .wav, .ogg, .webm, .flac) — audio is transcribed on build, then chunked and embedded like any other source. You can also point a capsule at a URL, or crawl a whole site from its sitemap. Images need OCR and PowerPoint isn’t supported yet, and a single very large file may not finish indexing.
Is my data isolated from other customers?
Yes. Every capsule is a private namespace, and tenant isolation is enforced in code — not just configuration — and covered by an automated isolation suite that runs on every change and must be green to merge. One customer can’t read, write, or list another’s capsules, vectors, storage, or usage.
Can I take my data with me, or am I locked in?
You can export the whole capsule — corpus, manifest, eval set, and write log — as a single archive with a re-materialize target, then rebuild it on any vector database, embedding model, or cloud. The embeddings and index are disposable build artifacts; the corpus and eval set are the permanent asset, and they’re yours.
Do agents only read, or can they write to memory?
Both. A capsule is memory an agent reads from and writes to. Every write carries provenance — agent, session, and source — and is revertible. A revert removes the content from retrieval while preserving an append-only tombstone and the write history.
How do I know a new version of my capsule is any good?
Write an eval set — the questions your capsule has to get right — and grade any source snapshot against it with an LLM judge. A quality gate keeps below-threshold snapshots out of promoted release history, and an owner can pin the source set retrieval may use. The eval set travels with the capsule in every export.
Which AI writes the answers?
You choose, per capsule: Sage Cloud (Cloudflare Workers AI, 25 models), a compatible external model (Claude from Anthropic or ChatGPT from OpenAI), or Sage Private in a custom cloud or on-prem deployment once it is validated. Retrieval always runs inside your capsule. The model only sees the passages found for each question, and you can switch at any time without rebuilding.
Can a capsule run on our own hardware?
Yes, as a scoped private deployment. Thyme RAG Capsules run managed on Cloudflare (self-serve), or as a custom deployment on hardware you own, which Retenta can help you buy and install. Choose On-premises on the setup page and send us your requirements; each private deployment is assessed, quoted and validated first. The capsule format is the same either way.
Which agents and tools can connect?
Each capsule exposes a first-class MCP server and a REST API, single-sourced from one tool spec so they never drift. Connect from Claude, ChatGPT, Gemini, or any headless harness or cron loop.
What happens when I hit my plan’s limit?
We warn before the hard limit. Upgrade to raise storage, indexing, query, or capsule capacity; Retenta does not silently run core overages. Retenta-funded hosted AI operations use a separate prepaid wallet and stop before work when it is empty. Your data is never deleted over billing: downgrading or cancelling keeps every capsule, source and passage, and you can export at any time.
Encrypted, isolated, and yours to take with you.
AES-256 at rest and TLS 1.3 in transit, per-capsule isolation covered by an automated test, and a full export whenever you want it. We’re also honest about the one limit: to answer over your content we can read it, so we scan a narrow set for illegal material (CSAM and credible threats) and cooperate with law enforcement — never general monitoring, never selling your data.
How we protect your dataBuilt on Cloudflare.
Workers, D1, R2, Vectorize and Workers AI — one runtime, no cold-start theatre. On built-in AI, no second cloud touches a request. A frontier lab or Sage is called only for capsules you point at one. Because the corpus, manifest and eval set are the permanent asset, the runtime underneath is an implementation detail we can change without touching what you own.
CloudflareRe-materialize targets — coming soon
AWS Bedrock
Google Vertex AIpgvectorYour capsule exports as corpus + manifest + eval set, so it can be rebuilt on infrastructure you choose. These targets are on the roadmap, not shipped.
More to come.
The stack is disposable. Make the memory permanent.
Embeddings and indexes are build artifacts you can rebuild anytime. The corpus, manifest, and eval set are the asset — and they’re yours to take anywhere.