All case studies
    WORKED EXAMPLE · WRITING TOOLSMay 2026· 10 min

    Jasper, Grammarly, Copilot in Word: Same Category, Three Structural Fates

    Jasper logoJasper
    Grammarly logoGrammarly
    Copilot in Word logoCopilot in Word
    L4L7
    Verdict: L7c surface vs L4a railroad

    Jasper Valuation

    Peak

    $1.5B (Oct 2022)

    Now

    ~$300M

    -80%

    Layer Scoring

    L-1
    Resources
    L0
    Infra
    L1
    Data
    L2
    Models
    L3
    Gates
    L4
    Access
    L5
    Execution
    L6
    Orchestration
    L7
    Surface
    L8
    Memory
    L2 Models
    The L2 commoditization (ChatGPT free) is what triggered the cascade, Jasper's surface lost its model scarcity overnight.
    L4 Access
    Grammarly built L4 (browser/editor extensions), defensible until Microsoft brought a bigger L4 (Office, Teams, 365 install base) to the same fight.
    L6 Orchestration
    Tone, style, voice orchestration, Grammarly's deeper layer, but increasingly replicable inside any L4 owner's stack.
    L7 Surface
    Jasper lived here alone, prompt templates and brand voice presets. Absorbed when the model itself became conversational.
    L8 Memory
    The unclaimed layer. Whoever owns the memory of *your* writing voice across years and teams wins the next decade. No one owns it yet.

    Sublayer Impact Map

    Which of the 50 sublayers this case actually touches, and at what magnitude.

    L2 Models
    Models
    Foundation models
    plays here: OpenAI, Anthropic, Google
    Owns
    L4 Access
    Access
    Browser extension
    plays here: Grammarly
    Owns
    Editor integration
    plays here: Microsoft 365
    Owns
    L6 Orchestration
    Orchestration
    Tone & style
    plays here: Grammarly, Copilot
    Share
    L7 Surface
    Surface
    Prompt templates
    plays here: Jasper
    Owns
    Writing surface
    plays here: All three
    Owns
    L8 Memory
    Memory
    Voice memory
    plays here: Up for grabs
    Touch
    Impact: Touch = enters · Share = meaningful · Owns = dominates· bars = magnitude

    Intelligence Cube · 2D

    Footprint across Functions × Verticals × Layers, the three axes that determine structural fate.

    Layers × Verticals

    8 cells · 4×2

    L-1
    L0
    L1
    L2
    L3
    L4
    L5
    L6
    L7
    L8
    FinTech
    EdTech
    Legal
    Health
    Travel
    eCom
    Media
    Gov
    SaaS
    Horizontal

    Layers × Functions

    16 cells · 4×4

    L-1
    L0
    L1
    L2
    L3
    L4
    L5
    L6
    L7
    L8
    Dev/Eng
    Design
    Product
    PM/Proj
    Ops
    Mktg
    Sales
    CustCare
    Strategy
    Finance

    Two 2D projections of the Intelligence Cube (Functions × Verticals × Layers). Filled cells = this move occupies that intersection.

    Timeline

    Nov 2021

    Grammarly hits $13B valuation at peak, owns browser + editor extensions.

    Oct 2022

    Jasper raises $125M at $1.5B, pure L7 prompt-template moat.

    Nov 2022

    ChatGPT launches. Free, conversational, GPT-3.5. Jasper's L7 moat begins evaporating in weeks.

    Mar 2023

    Microsoft announces Copilot in Word, Outlook, Teams, bundled into 365.

    2024

    Jasper reportedly trading at ~$300M, 80% mark-down.

    2025–26

    Grammarly squeezed: same moat layer, but Microsoft owns more of it. Pivots toward team-voice memory (L8).

    - Who Wins

    • Microsoft. Already owned the bigger L4. Added L2 inside it. Bundled at no marginal price.
    • OpenAI / Anthropic / Google. L2 is now the price-setting layer, every L7 wrapper pays them rent.
    • Any future L8 owner. The memory of your voice, across every doc you've ever written, is the only layer still unclaimed in writing.

    - Who Loses

    • Jasper. L7-only is the new GPT-wrapper exposure pattern. 80% mark-down reflects the structural exposure.
    • Pure-play AI writing startups. Same fate as Jasper unless they own L1 (voice data) or L8 (cross-doc memory).
    • Grammarly (partially). Still defensible in the browser, but squeezed inside Office. Needs an L8 sprint.

    - Steelman: The Counter-Thesis

    The counter is that Grammarly has built genuine L8 over a decade, billions of corrections per user, tone profiles, team style enforcement. Microsoft has the bigger L4 but no equivalent depth on *individual voice memory* across non-Microsoft surfaces (Slack, Gmail, browser, mobile keyboards). If Grammarly converts its corrections corpus into a true cross-surface voice memory product, and if enterprises actually pay for "the AI that knows how your VP of Marketing writes", then L8 + multi-L4 distribution beats single-L4 dominance. The honest read: that's a 30% probability bet, not a base case.

    Three writing products. Three structural positions. Three different fates, and none of them were predicted by ARR or product quality alone.

    Jasper, surface only (L7). At its Oct 2022 peak Jasper raised $125M at a $1.5B valuation as the canonical "GPT wrapper for marketers." Defensibility lived entirely at the surface: prompt templates, brand voice presets, a polished UI. When ChatGPT shipped six weeks later, the model layer absorbed the value. By 2024 Jasper was reportedly trading hands at a fraction of that mark. Law I in motion: intelligence commoditized downward.

    Grammarly, distribution + orchestration (L4 + L6). Already a $13B writing leader at its Nov 2021 peak, Grammarly was never a thin wrapper. It owned L4 (railroad), extensions in Chrome, Word, Gmail, Slack, the OS keyboard, and L6 (orchestration), tone detection, style enforcement, team voice. That depth carried it through the first wave: the surface might be a copilot, but the distribution was the moat.

    Copilot in Word, the L4 owner that ate L4. Then Microsoft did the thing that breaks the framework's most important asymmetry. Microsoft already owned a bigger L4, Word, Outlook, Teams, the entire 365 install base. They integrated GPT-class models directly into the document surface and bundled it with E3/E5. The same kind of moat (distribution into the editor) was now owned by an entity with vastly more of it. Grammarly's L4 didn't disappear, it just got out-scaled.

    Why this is the right triad to study. Most decks compare Jasper to ChatGPT. That's a category mismatch, ChatGPT is a general assistant. The honest comparison is Jasper, Grammarly, and Microsoft Copilot in Word: all three are literally writing-assistance products embedded in the prose-creation moment. The three fates are then driven entirely by structural depth.

    The structural read:
    • Jasper (L7 only), exposed; displaced as L2 commoditized
    • Grammarly (L4 + L6), defended for years; now squeezed by a bigger L4 owner above it
    • Copilot in Word (L0 + L2 + L4 + L7), vertically integrated from compute to surface, bundled into existing distribution

    Law III, the surface captures attention; the chain captures power. Same category does not equal same future. The market is repricing layer ownership, not feature quality. And when a bigger owner of your moat layer arrives, owning that layer is no longer enough, you need a layer they don't own. For Grammarly, that question now becomes: where is the L8 (memory of your writing voice across years and teams) that Microsoft can't easily replicate?

    Illustrative strategic patterns only, not investment recommendations. Public reporting; numbers approximate.

    What This Means for You

    Product Leader

    Map your product to the layers it actually owns vs. rents. The rented ones are where the counter-move work belongs.

    Investor

    Underwrite layer ownership, not feature count. The Cube footprint is the moat.

    Operator

    Audit your stack against Supply Chain of Intelligence. Anything sitting only at L7 is the layer to watch.

    AA

    Anand Arivukkarasu

    Ex-Meta product leader. Creator of Supply Chain of Intelligence™. Writes about where AI value accrues, and who can fire your product. LinkedIn

    Get the next teardown in your inbox.

    One issue when something structurally important happens, usually weekly. No spam, no filler, unsubscribe anytime.

    Worth sharing? Pull-quote: "L7c surface vs L4a railroad"