PKM & AI Knowledge Graph Tools Landscape 2026

MOKA-338 | Deep Research | 2026-03-20 Purpose: Competitive positioning for Neuron — AI-Powered Team Knowledge Graph

Executive Summary

The PKM and knowledge management market is in a consolidation phase: Notion approaches $1B ARR, Obsidian runs a $25M ARR business with 18 people, and dozens of AI-native startups compete for the “second brain” niche. The knowledge graph market is projected to grow from $1.07B (2024) to $6.94B (2030) at 36.6% CAGR (MarketsAndMarkets, 2025). The broader note-taking app market is valued at ~$7.9B in 2024, growing to $26.7B by 2032 at 16% CAGR (Credence Research, 2025).

Despite this activity, a clear gap persists: no product combines local-first architecture + AI-native knowledge graph + team-first collaboration. Obsidian dominates individual PKM but lacks native team features and AI. Notion dominates team collaboration but has no knowledge graph. Tana has the strongest schema-based graph but is cloud-only. Neuron can own the intersection — but the window is narrowing.

Go/No-Go Recommendation: Conditional GO. Build individual PKM first (free tier), prove graph differentiation, then layer team features. Do not attempt team + local-first + AI simultaneously in MVP.

1. Should Moklabs Build This?

Verdict: Conditional GO.

Reasons to build:

The intersection of local-first + AI knowledge graph + team collaboration is genuinely unoccupied
Obsidian proves a tiny team (18 people) can build a $25M ARR business in PKM (GetLatka, 2025)
Knowledge graph market growing at 36.6% CAGR provides tailwinds
Mem AI’s struggles ($40M raised, stalled product, poor reviews) demonstrate that cloud-only AI notes without solid fundamentals fail (Medium, 2025)

Reasons for caution:

PKM fatigue is real: users cycle between tools without committing
Notion ($600M ARR, $11B valuation) could add graph features with massive resources
Obsidian’s community plugin ecosystem is a moat — any feature Neuron ships, a plugin probably exists
Local-first team sync is technically hard and unsolved at scale

2. Market Overview

2.1 Market Sizing

Metric	Value	Source
Knowledge Graph Market (2024)	$1.07B	MarketsAndMarkets, 2025
Knowledge Graph Market (2030 proj.)	$6.94B	MarketsAndMarkets, 2025
Knowledge Graph CAGR	36.6%	GlobeNewsWire, 2025
Note-Taking App Market (2024)	$7.9B	Credence Research, 2025
Note-Taking App Market (2032 proj.)	$26.7B	Credence Research, 2025
PKM Software Market (2024)	$2.45B	DataIntelo, 2025
PKM Software CAGR (2025-2033)	15.8%	DataIntelo, 2025
KM Software Market (2025)	$23.2B	Fortune Business Insights, 2025

2.2 TAM/SAM/SOM Methodology

TAM (Total Addressable Market): $2.45B — PKM software market in 2024. This includes all tools in the personal knowledge management category (Obsidian, Notion personal, Logseq, Mem, etc.).

SAM (Serviceable Addressable Market): ~$490M — 20% of PKM TAM that represents users who value knowledge graph features and local-first architecture. Derived from: Obsidian’s 1.5M+ users represent the local-first segment; graph-interested users estimated at 30-40% of power users based on Obsidian forum activity around graph features.

SOM (Serviceable Obtainable Market, Year 3): ~$4.9M — 1% of SAM. Capturing 5,000-10,000 paid team users at $8/user/mo within 3 years. This assumes strong product-market fit and viral growth within developer/research communities.

Methodology note: These estimates use conservative multipliers. The broader KM software market ($23.2B) and knowledge graph market ($6.94B by 2030) provide additional upside if Neuron expands into enterprise use cases.

3. Competitive Landscape (Deep Dive)

3.1 Tier 1: Dominant Players

Obsidian

Metric	Value	Source
ARR	~$25M	GetLatka, 2025; Fueler, 2026
Active Users	1.5M+	Fueler, 2026
Team Size	18	GetLatka, 2025
Revenue/Employee	~$1.4M	Derived from ARR / team size
Revenue Mix	80%+ from Sync subscriptions	Fueler, 2026
Pricing	Free core; Sync $4/mo; Publish $8/mo	CostBench, 2026
Architecture	Local-first, Markdown files	—
AI	Community plugins only (no native AI)	Eesel, 2025
Team Collab	Basic shared vaults via Sync; Relay plugin for real-time CRDT editing	Obsidian Help; Relay.md
Knowledge Graph	Visual graph view (display only, not semantic)	—

Threat Assessment: Obsidian’s DNA is individual-first and extensibility. Adding native AI or team features would alienate their core “bring your own tools” philosophy. The Relay plugin adds CRDT collaboration but it’s third-party and niche. Obsidian is unlikely to build a semantic knowledge graph natively — their graph is purely visual. Risk: Medium.

Notion

Metric	Value	Source
ARR	~$600M (2025), trajectory to $900M-$1B by end 2026	GetLatka, 2025; SaaStr, 2025
Valuation	$11B (Dec 2025 tender)	SaaStr, 2025
Total Funding	~$344M (no primary raise since Oct 2021)	Sacra; Tracxn, 2026
Customers	4M+	GetLatka, 2025
Revenue Growth	$3M (2019) -> $31M (2021) -> $250M (2023) -> $400M (2024) -> $600M (2025)	SaaStr, 2025
Pricing	Free; Plus $10/seat/mo; Business $15/seat/mo; Enterprise $20+/seat/mo	Notion Pricing
Architecture	Cloud-first	—
AI	Notion AI (unlimited on Business+); AI agents launching	Max-Productive, 2026
Knowledge Graph	None native; third-party IVGraph plugin exists	IVGraph, 2026

Threat Assessment: Notion has the resources and data model (relations + databases) to build knowledge graphs. However, their architecture is cloud-only, and their most-requested feature for years was offline mode — suggesting a fundamental tension with local-first approaches. Notion is focused on becoming the “AI workspace” (agents, AI chat), not a knowledge graph tool. Risk: Medium-High (could add graph visualization but unlikely to go local-first).

3.2 Tier 2: Direct Competitors

Anytype

Metric	Value	Source
Total Funding	$13.4M (Series A, Aug 2023, led by Balderton Capital)	Anytype Blog; Crunchbase
Team Size	~175	Tracxn, 2026
Pricing	Free - $20/mo	FahimAI, 2026
Architecture	Local-first, IPFS, E2E encrypted, P2P sync	—
AI	Basic	—
Knowledge Graph	Object graph view	—

Why Anytype is the closest threat: Same vision (local-first + team + graph) but with 175 employees and $13.4M in funding. Their IPFS-based architecture is novel but adds complexity. AI capabilities are basic. Neuron differentiates on: (1) AI-native knowledge graph (auto-linking, entity extraction), (2) simpler architecture (SQLite vs IPFS), (3) Apple ecosystem integration.

Tana

Metric	Value	Source
Total Funding	$25M ($11M seed + $14M Series A, Feb 2025)	TechCrunch, 2025
Waitlist	160K+	TechCrunch, 2025
Active Testers	30K+	SiliconANGLE, 2025
Community	24K+ Slack members	PR Newswire, 2025
Pricing	Free core; Plus $10/mo; Pro $18/mo	Tana PKM
Architecture	Cloud-only	—
Revenue	Pre-revenue / early-revenue (just launched Feb 2025)	—

Why Tana matters: Best supertag-based schema graph in the market. $25M in funding and strong enterprise interest. However: cloud-only, no local-first, and just launched publicly. Neuron’s window to establish local-first graph positioning before Tana scales is ~12-18 months.

Heptabase

Metric	Value	Source
Total Funding	$2.2M (YC-backed)	Crunchbase
ARR	~$7M (2026)	Fueler, 2026
Monthly Revenue	$1.2M (Dec 2024)	GetLatka, 2024
Valuation	$80-100M	Fueler, 2026
Team Size	8	GetLatka, 2024
User Growth	28% YoY	Fueler, 2026
Pricing	$8.99-$17.99/mo	—
Architecture	Cloud + local cache	—

Signal: Heptabase proves spatial/visual PKM can reach $7M ARR with 8 people and only $2.2M in funding. Capital-efficient model Neuron should emulate.

3.3 Tier 3: Adjacent Players

Mem AI (Cautionary Tale)

Metric	Value	Source
Total Funding	$29.1M (OpenAI led $23.5M at $110M valuation)	TechCrunch, 2022
Status	Product stagnation; poor reviews; lack of basic features	Medium (Theo James); G2 Reviews

Lesson for Neuron: AI features without solid note-taking fundamentals fail. Ship a great editor first, then layer AI. Mem’s mistake was prioritizing “gimmicky AI marketing features” over note-taking essentials.

Reflect

Metric	Value	Source
Funding	$3.75M (Sep 2025)	Tracxn, 2025
Revenue	$326.4K (2024); ~$218K MRR run-rate (2025)	GetLatka, 2024
Team Size	3	GetLatka, 2024
Status	Profitable, semi-bootstrapped	Reflect Blog

Capacities

Metric	Value	Source
Funding	$0 (100% user-funded)	Capacities Pricing
Architecture	Cloud	—
Differentiator	Object-based note-taking, not document-based	—

Logseq

Metric	Value	Source
Funding	$4.1M seed (2022, led by Patrick Collison, Nat Friedman)	Logseq Blog
Architecture	Local-first, Markdown/Org, open source	—
Status	Active but niche; outliner-first approach limits mainstream adoption	—

4. Differentiation Axes

Axis 1: Individual vs Team

Individual --------------------------------- Team
Obsidian  Reflect  Logseq  Heptabase  |  Notion  Confluence  Slite
Mem  Capacities  Anytype  Tana        |
                                       |
                          NEURON TARGET |
                    (Individual -> Team) |

Axis 2: Local-First vs Cloud

Local-First -------------------------------- Cloud
Obsidian  Logseq  Anytype  |  Mem  Notion  Confluence  Tana
                            |  Heptabase  Capacities  Reflect
                            |
               NEURON TARGET |
         (Local + AI + Team) |

Axis 3: Knowledge Graph Sophistication

None/Basic --------------------------------- Full Graph
Notion  Confluence  |  Obsidian  Heptabase  |  Tana  Anytype  |  Atlas
Reflect  Mem        |  Logseq  Capacities   |                 |
                    |                       |  NEURON TARGET  |
                    |                (AI-native team graph)    |

5. What Specifically Would We Build? (ICP + MVP)

5.1 Ideal Customer Profile

Primary ICP: Individual developers and researchers (25-45) who currently use Obsidian but want:

Smarter linking (auto-suggested connections, not just manual [[wikilinks]])
A semantic knowledge graph (not just a visual display)
Better mobile experience (Obsidian mobile is functional but clunky)

Secondary ICP (post-MVP): Small engineering teams (3-15 people) who use Obsidian individually but need team knowledge sharing without moving to Notion.

Willingness to Pay:

Individual: $0-8/mo (must compete with Obsidian’s free tier)
Team: $8-12/user/mo (undercut Notion at $10-15/user/mo)
Evidence: Obsidian Sync at $4/mo captures 80%+ of their revenue, showing individual PKM users will pay for sync. Heptabase at $8.99-$17.99/mo proves premium PKM pricing works.

5.2 MVP Feature Set

Feature	Why	Priority
Rich text editor (TipTap)	Table stakes — must be as good as Obsidian editor	P0
Wiki-links (`[[` trigger)	Core differentiator — entry point to graph	P0
Knowledge graph visualization	Visual “aha moment” — the feature people screenshot	P0
Full-text search (FTS5)	Basic utility — users won’t adopt without fast search	P0
Daily notes	Habit formation — daily entry point to the app	P1
Backlinks panel	Shows graph value without opening graph view	P1
AI auto-linking (v1)	Neuron’s core differentiator — suggest connections users missed	P1
Tags	Organization primitive	P2
Entity extraction	Semantic graph enhancement — people, projects, concepts	P2

5.3 Pricing Strategy

Tier	Price	Target
Personal	Free	Individual PKM, local-only, basic graph
Pro	$8/mo	AI auto-linking, entity extraction, semantic search, cloud backup
Team	$12/user/mo	Shared graphs, real-time co-editing, team insights
Enterprise	Custom	Self-hosted, compliance, SSO, dedicated support

6. What’s the Unfair Advantage?

Local-first + AI knowledge graph: No one combines these. Obsidian is local but no AI. Mem is AI but cloud-only and failing. Tana has graph + AI but is cloud-only.
Apple ecosystem integration: On-device AI via Apple Foundation Models means AI features that are private, fast, and free. No API costs for inference. No competitor has this.
Capital efficiency model: Obsidian ($25M ARR, 18 people) and Heptabase ($7M ARR, 8 people) prove PKM can be built lean. Neuron doesn’t need VC to reach profitability.
Why now: (a) Knowledge graph market at 36.6% CAGR, (b) Apple Foundation Models enable on-device AI in 2025-2026, (c) Tana just launched publicly — the “AI knowledge graph” category is being defined right now, (d) cr-sqlite and SQLiteSync make local-first CRDT sync feasible.

7. What Kills This Idea? (Top 3 Risks)

Risk 1: Obsidian Adds Native AI + Team Features

Probability: Medium (30%) Impact: High Analysis: Obsidian’s 2026 roadmap includes syncing vaults without the full app (Obsidian Roadmap), suggesting incremental improvements to collaboration. However, their philosophy of extensibility-via-plugins makes native AI unlikely. The Relay plugin already adds CRDT collaboration, which could be “good enough” for many teams. Mitigation: Ship faster. Neuron’s AI graph must be demonstrably better than any Obsidian plugin within 12 months.

Risk 2: PKM Fatigue / “Another Note App”

Probability: High (50%) Impact: High Analysis: Users cycle between PKM tools endlessly. “Perfectionism and configuration burden” are systemic failures of PKM adoption (Medium). The PKM Paradox: “most PKM systems fail because they’re missing destinations — a purpose for the knowledge being acquired.” Mitigation: Position Neuron as “the team brain” not “another note app.” Focus marketing on team knowledge sharing, not individual productivity. Make the graph the hero feature that differentiates from the 50+ note-taking apps.

Risk 3: Local-First Team Sync is Technically Unsolved at Scale

Probability: Medium-High (40%) Impact: High Analysis: CRDT-based sync (cr-sqlite, SQLiteSync) works for small teams but faces challenges: storage limits on user devices, debugging distributed states, compliance/audit trail requirements (DEV Community; Tech-Champion, 2026). If an application needs strong consistency or centralized audit trails, cloud-first may be more suitable. Mitigation: Start individual-first (no sync needed). Add cloud backup as Pro feature. Tackle team sync only after validating individual product-market fit. Use hybrid architecture: local-first for writing, cloud for team graph aggregation.

Bonus Risk: Notion Adds Knowledge Graphs

Probability: Low-Medium (20%) Impact: Very High Analysis: Notion’s database relations already enable graph-like structures. Third-party IVGraph is building graph visualization for Notion (IVGraph, 2026). Notion’s focus on AI agents and offline mode suggests knowledge graphs aren’t on the immediate roadmap, but they have the data model and resources to pivot quickly. Mitigation: Neuron’s local-first architecture is structurally incompatible with Notion’s cloud approach. Even if Notion adds graph visualization, it won’t be local-first or privacy-preserving. Lean into privacy and ownership as core values.

8. Key Technical Differentiators to Build

AI Auto-Linking — Entity extraction + semantic similarity to automatically suggest connections between notes. No competitor does this natively with local-first architecture.
Semantic Knowledge Graph — Not just visual links (Obsidian), but typed entities and relations (person -> works_at -> company) extracted automatically.
Meeting-to-Graph Pipeline — Auto-extract entities, decisions, and action items from meeting transcripts.
Graph Analytics — “Knowledge gaps” detection: areas with thin coverage, knowledge silos where only 1 person holds context.
Daily Notes with Auto-Context — Surface relevant past notes when writing daily notes, using on-device Apple Foundation Models.

9. Competitive Comparison Matrix

Dimension	Obsidian	Notion	Tana	Anytype	Heptabase	Neuron
Architecture	Local (files)	Cloud	Cloud	Local (IPFS)	Cloud+cache	Local (SQLite)
Knowledge Graph	Visual only	None	Supertag schema	Object graph	Spatial	Semantic (entity-relation)
AI	Plugins	Notion AI	AI generation	Basic	Credit-limited	On-device (Apple)
Team Support	Basic (Sync)	Excellent	Workspace	Shared Spaces	Basic	CRDT sync (planned)
Search	In-memory	Server	Server	Local	Cloud	FTS5 + semantic
Pricing (ind.)	Free	Free	Free	Free	$8.99/mo	Free
Pricing (team)	$4/mo sync	$10-15/seat	$10-18/mo	$20/mo	N/A	$12/user/mo
Funding	$0 (bootstrapped)	$344M	$25M	$13.4M	$2.2M	—
ARR	$25M	$600M	Pre-rev	Unknown	$7M	—