ἀλήθεια /a.lɛ́ː.tʰeː.a/

The memory
you can actually reason with.

Aletheia turns scattered notes — Joplin, Obsidian, Notion, meeting transcripts — into a synthesized memory. Every answer is retrieval-first, every claim links back to the exact note. No hallucination. No hedging.

Hosted on Azure · your private workspace | Bring your own OpenAI key
aletheia · session · 14:02 forest · 1,284 notes indexed
A

what's going on with the avalara migration? blockers, owners, what's next.

synthesize_topic · matched 47 chunks across 18 notes
find_relationships · resolved 6 entities, 4 owners
building synthesis…

Status

  • Migration is ~70% complete; pilot tenant cutover succeeded May 8 avalara-pilot-recap.md
  • Production rollout blocked on tax-jurisdiction mapping for 4 EU regions eng-sync-2026-05-12

Owners

  • Sarah Brobston — tax mapping, EU jurisdictions 1:1 sarah · may 09
  • Devraj P. — invoice template diff, integration tests devraj-handoff.md

Open questions

  • Are we deprecating the legacy TaxJar fallback at cutover, or running both in parallel for Q3? avalara-decisions
why aletheia

Four things that change
everything else.

Most "chat with your notes" tools are a vector search with a chatbot bolted on. Aletheia is structured retrieval first, generation second.

01 / synthesis

Synthesize, don't summarize.

Purpose-built workflows hunt for decisions, blockers, ownership chains, and open questions — not just keyword matches. Ask a question, get a synthesis.

workflow → synthesize_topic
02 / citations

Citations or it didn't happen.

Every claim links back to the exact note. Inline pills you click to jump straight to source. No "according to my training data" — if we don't have a source, we say so.

retrieval-first · zero hedging
03 / portability

Your notes, your platform.

Bring it in from anything — Joplin, Obsidian, Notion, Logseq, plain markdown. Take it out the same way. No lock-in, no proprietary format you can't escape.

markdown + yaml frontmatter
04 / entities

Built to know things.

Auto-resolves people, companies, and projects across notes. Sarah, Sarah B, and Sarah Brobston are one person. "What do I know about X" is a first-class query.

typed-edge graph · entity resolution
seven workflows

The LLM picks
the right one for you.

Each workflow is a separate retrieval + reasoning chain tuned for a specific shape of question. You don't choose — Aletheia routes based on intent.

01
Synthesize topic

Status, decisions, open questions, next actions for any topic.

"what's going on with the avalara migration?"
02
Find relationships

Entities, tags, and supporting passages connected to any subject.

"what's connected to sarah?"
03
Build timeline

Chronological narrative of any topic, every beat linked to source.

"timeline: q1 pricing debate"
04
Find notes

Filtered, scoped search with rich previews and provenance.

"notes tagged #1on1 in the last 30 days"
05
Find quotes

Verbatim passages with context windows, ready to drop into a draft.

"quotes from the october customer interviews"
"
06
What's on my plate

Open action items extracted across notes, grouped by owner.

"my open items, by project"
07
Find contradictions

Conflicting statements surfaced across notes with both sources side-by-side. The one workflow nothing else does.

"contradictions in my notes about quantization tradeoffs"
who it's for

Built for people
who actually take notes.

persona / 01

The program manager

Five live projects, dozens of meetings a week. Notes scattered across Joplin, Slack threads, email summaries.

"what's the status of the avalara integration? what blockers? who owns what?"
30-second synthesis with named owners, the actual decisions, and what's still pending Monday.
persona / 02

The independent consultant

Multiple clients, each with their own thread of context. Needs to recall what was agreed and who owns it.

"what do I know about acme corp's billing setup?"
Entity dossier: the people involved, key meetings, the open questions, dates.
persona / 03

The researcher / analyst

Reading and note-taking across hundreds of sources. Threading arguments, finding contradictions, building timelines.

"find contradictions in my notes about quantization tradeoffs."
Conflicting claims surfaced with the source citations side-by-side.
persona / 04

The writer / journalist

Source organization, fact-checking, draft assembly with rigorous citation discipline.

"build a timeline of how the policy debate evolved."
Chronological narrative with each beat linked to the underlying source.
how it works

Corpus → workflows →
answer with citations.

Hybrid retrieval (BM25 + embeddings + tag-match), section-boosted rerank, then the right synthesis workflow for the question.

stage / 01
Corpus
1,284 notes
joplin/raw-export412
obsidian-vault318
notion-export204
meetings/plaud187
research/papers163
stage / 02
Retrieval
rrf · k=24
lexical · bm25 0.78
semantic 0.91
tag-match 0.64
## decisions +rerank
## open q's +rerank

structured chunks (## decisions, ## open questions) get amplified.

stage / 03
Synthesis
cited
Pilot cutover succeeded May 8 avalara-pilot. Blocked on EU jurisdiction mapping eng-sync. Owners: Sarah Brobston (tax) 1:1 sarah, Devraj P. (invoices) handoff.

every claim → click → source note.

interop

Bring it in. Take it out.

No lock-in. Universal markdown is the substrate — every supported tool round-trips.

Import from anything

One-click import. Drag the export, watch the count tick up, ask a question.

Joplin RAW Obsidian vault Notion export Logseq graph Markdown ZIP PDF · 100MB audio images

Export to anything

Markdown plus YAML frontmatter, packaged as a ZIP. Opens in any tool, forever.

Markdown + YAML JSON dump SQL export git-friendly

$ aletheia export --format=md --out=./vault.zip

privacy & deployment

Your workspace.
Honest about scope.

Aletheia runs on Azure with managed Postgres and a hard monthly cap with auto-halt. You bring the OpenAI key — OpenAI sees only the chunks your query retrieves, nothing else.

infrastructure
Azure-managed Postgres, Container Apps, Static Web Apps.Your data lives in your own isolated workspace.
cost guardrail
Hard $10/mo cap with auto-halt before spend overruns.No surprise bills. Ever.
inference
Your OpenAI key. Stored encrypted, used only at query time.OpenAI sees only the retrieved chunks for the question you asked.
data ownership
Notes are yours. Export at any moment to portable markdown.No proprietary format. No retention games.
pricing

Simple. Honest.
Not a SaaS racket.

Pay for managed infra. Pay for your own model inference. No margin on tokens, no SaaS racket.

hosted
$10 /mo · hard cap
Managed Azure stack. Notes stored in your private Postgres. We run the infra, you bring the key.
  • All seven synthesis workflows
  • Universal import & export
  • Managed Postgres + Container Apps
  • Auto-halt when budget hits cap
  • Daily backups, restore on demand
Start free trial
roadmap
enterprise
Talk · custom
Multi-tenant, SSO, role-based scoping. Not shipped yet — we're being honest about it.
  • SSO & SCIM provisioning
  • Team workspaces, shared corpora
  • Audit log, retention policies
  • On-prem deployment options
Join waitlist
ἀλήθεια · unconcealment

Reveal what your notes
already know.

Create a workspace, drop in your vault, and ask the first real question of your own corpus. It takes about four minutes.