Question 1

What is ThreadMemory?

Accepted Answer

ThreadMemory is an AI knowledge router that connects team chat platforms — Slack, Discord, and Telegram — to productivity tools like Notion, Jira, GitHub Issues, and Confluence. ThreadMemory's 7-stage AI pipeline automatically classifies important discussions into 5 types (bugs, ideas, decisions, knowledge, process) and routes them to the right destination. Zero manual effort — your team's knowledge is captured as it happens.

Question 2

What problem does ThreadMemory solve?

Accepted Answer

ThreadMemory solves the knowledge loss problem that affects 90% of remote and hybrid teams using Slack, Discord, or Telegram. Important decisions, bug reports, and process changes get buried in chat history within 24 hours. ThreadMemory's AI monitors selected channels, classifies 5 content types with 88% accuracy across 92 test scenarios, and routes important discussions to Notion, Jira, GitHub Issues, or Confluence — automatically, in under 3 minutes after the conversation matures.

Question 3

Who is ThreadMemory built for?

Accepted Answer

ThreadMemory is built for small-to-medium teams of 5 to 20 people — startups, developer teams, and remote-first organizations that use Slack for communication and Notion, Jira, or GitHub for project management. ThreadMemory's per-workspace pricing starts at $0/month (Free plan) with 4 paid tiers up to $149/month (Team), making it accessible for early-stage teams. The setup takes under 3 minutes — no code, no workflow builders, no IT department required.

Question 4

How do I set up ThreadMemory?

Accepted Answer

ThreadMemory connects to your Slack workspace in under 3 minutes with a 4-step setup: (1) sign in with Google or Slack OAuth, (2) authorize the ThreadMemory Slack bot with channels:history and channels:read permissions, (3) select which channels to monitor, and (4) create your first routing rule in plain English — for example, "Send bug reports from #engineering to Jira." No code, regex, or workflow builders required. The Free plan includes 1 source, 1 destination, and 20 documents per month with no credit card.

Question 5

What chat platforms does ThreadMemory support?

Accepted Answer

ThreadMemory currently supports Slack as a fully operational source connector with OAuth bot installation, channel selection, thread detection, and real-time event webhooks. Discord and Telegram source connectors are on the roadmap. For destinations, ThreadMemory routes to 4 platforms: Notion (formatted pages with blocks and properties), GitHub Issues (issues with labels and descriptions), Jira (tickets with project mapping), and Confluence (structured documentation pages). Each of the 4 destination adapters formats content natively for its platform.

Question 6

What destination tools does ThreadMemory route to?

Accepted Answer

ThreadMemory routes classified content to 4 destination tools, each with a dedicated format adapter that restructures chat discussions into native document formats: Notion creates formatted pages with proper blocks, properties, and database entries. GitHub Issues creates issues with labels, descriptions, and markdown formatting. Jira creates tickets with correct fields, project mapping, and severity levels. Confluence creates structured documentation pages with headers and context.

Question 7

Can I control which channels ThreadMemory monitors?

Accepted Answer

ThreadMemory monitors only the Slack channels you explicitly select — never your entire workspace. Each source connection displays all available channels, and you toggle monitoring on or off per channel. The Starter plan supports up to 5 monitored channels, Pro supports 15, and Team offers unlimited channels. You can also create channel-specific routing rules: "#engineering bugs → Jira" and "#product decisions → Notion." Unmonitored channels are completely ignored — ThreadMemory's bot receives no events from them.

Question 8

How does ThreadMemory's AI pipeline work?

Accepted Answer

ThreadMemory processes every message through a 7-stage AI pipeline powered by 9 River job workers running on a Go backend. The stages are: (1) Ingest & Classify — messages are classified into 5 types (bug, idea, decision, knowledge, process) with importance scoring, (2) Embed — text is converted to 1,536-dimensional vectors via OpenAI embeddings and stored in pgvector, (3) Cluster — semantically similar messages are grouped using 0.85 cosine similarity threshold, (4) Summarize — each cluster gets a human-readable summary, (5) Triage — type-specific routing and urgency are determined, (6) Maturity Check — the system waits for discussions to conclude before routing, (7) Materialize — content is formatted for the destination tool and delivered.

Question 9

How does ThreadMemory decide what is important?

Accepted Answer

ThreadMemory uses a 2-layer AI classification system to determine importance. Layer 1 (Ingest & Classify) scores each message from 0 to 100 across 5 content types — bugs, ideas, decisions, knowledge, and process items — using GPT-4o-mini with structured output schemas. Layer 2 (Triage) applies type-specific thresholds: bug reports can create draft artifacts at a low 0.35 confidence score, while decisions require 0.55 and at least 2 participants. This dual-layer approach achieves 88% accuracy across 92 golden test scenarios, with type-specific tuning that prevents premature routing of half-finished discussions.

Question 10

What does "smart maturity" mean in ThreadMemory?

Accepted Answer

ThreadMemory's smart maturity system uses type-specific thresholds to determine when a discussion is ready to route — preventing half-finished conversations from becoming incomplete Jira tickets or Notion pages. Bug reports use a low draft threshold (0.35) so they route quickly as drafts, then auto-update as discussion continues up to the full 0.65 threshold. Decisions require a higher 0.55 score and at least 2 participants to ensure a conclusion was reached. Knowledge items need 0.55 with substantive content validation. Each of the 5 content types (bug, idea, decision, knowledge, process) has its own independently tuned maturity prompt template.

Question 11

Does ThreadMemory hallucinate or make errors?

Accepted Answer

ThreadMemory includes a self-review gate in the materialization stage that validates AI-generated content before it reaches your Notion, Jira, or GitHub. The self-review checks factual consistency against the original messages, verifies that no information was fabricated, and confirms the content matches the classified type. Additionally, ThreadMemory uses a 0.85 cosine similarity deduplication check to prevent the same discussion from creating duplicate artifacts. The pipeline achieves 88% accuracy across 92 test scenarios — and every routed document links back to the original chat messages for human verification.

Question 12

How does ThreadMemory handle long discussions?

Accepted Answer

ThreadMemory's token budget system handles discussions of any length through a 4-layer adaptive strategy. Layer 1: individual messages are capped at 1,500 tokens at ingest, ensuring consistent DB storage. Layer 2: cluster prompts use an adaptive cascade — if a prompt exceeds the 100,000 token budget, the system progressively reduces recent messages, then reduces clusters, then splits the batch. Layer 3: for clusters with more than 80 messages, a head+tail strategy selects the earliest 10 and latest 70 messages to preserve context. Layer 4: embedding text is truncated to 8,000 tokens before the OpenAI API call (8,191 token limit).

Question 13

How does ThreadMemory connect to Slack?

Accepted Answer

ThreadMemory uses 2 separate Slack OAuth flows for security isolation. Flow 1 (Sign-in): standard OAuth with openid, profile, and email scopes — authenticates the user and returns basic team info. Flow 2 (App Installation): bot OAuth with channels:history, channels:read, chat:write, and users:read scopes — gives ThreadMemory's bot read access to selected channels and the ability to respond. The bot token is encrypted with AES-256-GCM envelope encryption (per-tenant data encryption key) before storage. ThreadMemory receives messages via Slack's Events API webhooks in real-time — no polling, no delays.

Question 14

Can ThreadMemory route different topics to different tools?

Accepted Answer

ThreadMemory's routing rule engine lets you create unlimited rules that map specific content types and channels to different destination tools. For example: "Bug reports from #engineering → Jira," "Product decisions from #product → Notion," "Architecture knowledge from #backend → Confluence," and "Feature ideas from any channel → GitHub Issues." Each rule uses plain English — ThreadMemory's AI interprets the intent and matches incoming classified content. The Pro plan ($79/month) and Team plan ($149/month) support unlimited routing rules; Starter ($29/month) supports up to 5 rules.

Question 15

Does ThreadMemory work with tools besides Slack and Notion?

Accepted Answer

ThreadMemory currently supports 1 source (Slack) and 4 destinations (Notion, GitHub Issues, Jira, and Confluence). Discord and Telegram source connectors are on the product roadmap. Each destination has a dedicated format adapter — Jira tickets include fields, severity, and project mapping; GitHub issues use markdown with labels; Confluence pages have structured headers and sections. All 4 adapters include rate limiting, retry logic, and circuit breakers for reliability. The Atlassian connectors (Jira and Confluence) use OAuth 2.0 3LO (3-legged OAuth) for secure workspace access.

Question 16

How is ThreadMemory different from Zapier or Make?

Accepted Answer

ThreadMemory replaces manual workflow building with AI-powered classification and routing. Zapier and Make require you to design a separate "zap" or "scenario" for each automation — choosing triggers, filters, and actions. ThreadMemory's 7-stage pipeline does this automatically: it classifies messages into 5 types, waits for discussions to mature (smart maturity with type-specific thresholds from 0.35 to 0.65), and formats content natively for each destination. Zapier charges per task ($0.01-0.05 per execution); ThreadMemory uses flat per-workspace pricing starting at $29/month for unlimited message processing.

Question 17

How is ThreadMemory different from Slack AI?

Accepted Answer

ThreadMemory and Slack AI solve different problems. Slack AI ($10/user/month) provides in-Slack search and channel summaries — the knowledge stays inside Slack. ThreadMemory ($29-149/month per workspace, not per user) extracts important content from Slack and routes it to external tools: Notion pages, Jira tickets, GitHub issues, and Confluence docs. For a 10-person team, Slack AI costs $100/month for search-only; ThreadMemory's Starter plan at $29/month captures and routes up to 100 documents per month to your existing project management tools.

Question 18

Why not just copy-paste important messages manually?

Accepted Answer

ThreadMemory automates what teams currently do manually — and does it better. Manual copy-paste from Slack to Jira or Notion takes 2-5 minutes per item, requires someone to notice the message, and loses context (no thread history, no participant list, no related discussions). ThreadMemory's pipeline processes every message in monitored channels, classifies 5 content types with 88% accuracy, groups related messages into semantic clusters using pgvector embeddings (0.85 cosine threshold), and delivers formatted artifacts with full context. For a team generating 10-20 important discussions per week, ThreadMemory saves 2-4 hours of manual documentation work.

Question 19

How much does ThreadMemory cost?

Accepted Answer

ThreadMemory uses flat per-workspace pricing (not per user), making it predictable for teams of any size. 4 plans are available: Free ($0/mo — 1 source, 1 destination, 20 docs/mo), Starter ($29/mo — 2 sources, 2 destinations, 100 docs/mo), Pro ($79/mo — unlimited sources and destinations, 300 docs/mo), and Team ($149/mo — unlimited everything, 1,000 docs/mo). All plans include the full 7-stage AI pipeline, AES-256-GCM encryption, and format adapters for all 4 destinations.

Question 20

Is there a free trial or free plan?

Accepted Answer

ThreadMemory offers a permanent Free plan — not a time-limited trial. The Free plan includes 1 source connection, 1 destination, 1 monitored channel, and 20 documents per month with no credit card required and no time limit. When you're ready for more capacity, upgrade instantly from the dashboard: Starter ($29/mo) adds 2 sources, 2 destinations, 5 channels, and 100 documents. Downgrade anytime — takes effect at the end of your billing period, no cancellation fees.

Question 21

What counts as a "document" in ThreadMemory?

Accepted Answer

ThreadMemory counts 1 document each time an artifact is delivered to a destination tool — 1 Notion page, 1 Jira ticket, 1 GitHub issue, or 1 Confluence page equals 1 document. Messages that don't match any routing rule, messages below importance thresholds, and skipped messages do not count. Draft artifacts (created for bugs at the 0.35 threshold before the discussion matures) count as 1 document even if they're later updated at the 0.65 full-maturity threshold. The Free plan includes 20 documents/month, Starter includes 100, Pro includes 300, and Team includes 1,000.

Question 22

How does ThreadMemory protect my data?

Accepted Answer

ThreadMemory implements 5 layers of data protection: (1) Encryption — AES-256-GCM envelope encryption with per-tenant data encryption keys (DEK) for OAuth tokens, bot tokens, and API credentials. (2) PII detection — 7 regex patterns automatically scrub AWS keys, JWT tokens, credit card numbers, email addresses, IP addresses, API credentials, and phone numbers from AI-generated content. (3) Transport — TLS 1.3 for all connections between services. (4) Database — PostgreSQL on Neon with encrypted connections. (5) Audit — all access and pipeline decisions are logged in audit_logs.

Question 23

Does ThreadMemory store my messages?

Accepted Answer

ThreadMemory stores message content encrypted in the database for pipeline processing — classification, clustering, and materialization. Each message is truncated to a maximum of 1,500 tokens at ingest to bound storage and prompt size. Messages are encrypted with AES-256-GCM envelope encryption using per-tenant keys. Raw messages are subject to automatic purge policies — they are not retained indefinitely. Once an artifact is materialized and delivered to your destination tool, the message content is used only for deduplication checks (0.85 cosine similarity threshold) and cluster updates.

Question 24

Is ThreadMemory GDPR compliant? Can I delete my data?

Accepted Answer

ThreadMemory supports GDPR compliance with full data portability and deletion capabilities. You can: (1) disconnect sources — immediately stops all message processing, (2) delete individual sync logs and artifacts from the Activity Feed, (3) delete your entire workspace — triggers cascading deletion across all tables including messages, clusters, artifacts, rate_limits, audit_logs, and processing_logs. Data deletion is permanent and processed within 30 days. ThreadMemory stores data on Neon (PostgreSQL) servers and can provide a Data Processing Agreement (DPA) upon request.

Frequently Asked Questions

What is ThreadMemory?

Getting Started

How It Works

Integrations

Comparison

Pricing & Plans

Security & Trust

Still have questions?