History of Spreadsheets: From VisiCalc to Taskade Genesis (2026)

In October 1979, Dan Bricklin and Bob Frankston released VisiCalc on the Apple II. The product cost $100. Within three years, Apple had sold a million Apple IIs — most of them to accountants, financial analysts, and small business owners who bought the computer specifically to run VisiCalc.

The spreadsheet built the personal computer.

Forty-seven years later, the same architectural pattern — a programmable document where users define computations and relationships without writing code — runs from VisiCalc through Lotus 1-2-3, Excel, Google Sheets, Airtable, Notion, and Taskade Genesis. This is the story.

TL;DR: The spreadsheet was the original "no-code." VisiCalc (1979) by Bricklin and Frankston built the Apple II. Lotus 1-2-3 (1983) built the IBM PC. Excel (1985) built Microsoft. Google Sheets (2006) made spreadsheets collaborative. Airtable (2014) made them relational. Notion (2016) made them mix with documents. Taskade Genesis (2025) extends the programmable-document lineage to AI-native app generation: describe the app in plain English, get a working system with database + agents + automations in seven minutes. 150K+ Genesis apps built. Try Genesis →

🗺️ 47 Years of Programmable Documents

Forty-seven years. Eight inflection products. One throughline: the document where users define computation.

📜 The Big Idea That Started It All

The spreadsheet is one of computing's least appreciated inventions because it's so ubiquitous it disappeared into furniture. But VisiCalc in 1979 was as transformative for personal computing as the iPhone was for mobile.

Before VisiCalc: Mattessich (1961) and LANPAR (1969)

VisiCalc wasn't the first electronic spreadsheet. Richard Mattessich at UC Berkeley described the conceptual spreadsheet in a 1961 paper. René Pardo and Remy Landau built LANPAR (Language for Programming Arrays at Random) in 1969 — the first true electronic spreadsheet, running on Bell Canada, AT&T, and General Motors mainframes. Pardo and Landau later won a 1995 patent lawsuit confirming priority. LANPAR was the missing link between the academic concept and the personal-computer mass-market product.

But mainframe-only spreadsheets stayed inside large enterprises. The breakthrough was bringing the same idea to the personal computer.

Bricklin's Harvard Classroom Moment (1978)

Dan Bricklin conceived the personal-computer spreadsheet while watching a professor at Harvard Business School work through a production-planning case study, erasing and rewriting numbers on a blackboard ledger, recalculating downstream values each time one input changed. Bricklin imagined: what if the blackboard recalculated itself?

He took the idea to Bob Frankston — his MIT classmate and a brilliant systems programmer — and together they built it. Bricklin handled the conceptual design and UI; Frankston wrote most of the production code, optimizing tightly for the 6502 processor and 16KB RAM of the Apple II.

VisiCalc shipped October 1979 for $100. It ran on the Apple II first, then on the Atari 800, Commodore 64, IBM PC, and TRS-80 in subsequent years. Within three years it had sold over 700,000 copies (eventually approaching ~1M lifetime) — making it one of the best-selling software products of the early personal-computer era. Roughly 25% of 1979 Apple II sales were attributed to people buying the computer specifically to run VisiCalc.

Bricklin chose not to patent the spreadsheet — a decision he has discussed publicly across decades. A patent would likely have made him a billionaire and slowed the entire spreadsheet category. By not patenting, he seeded an industry. The Lotus 1-2-3 / Excel / Sheets / Airtable / Notion / Taskade lineage all descend from a patent that didn't exist.

What it actually did:

                VisiCalc · Apple II · 1979
                ──────────────────────────
  A         B         C         D

Change any cell → all downstream cells recalculate.
This was magic in 1979.


The cell-and-formula model VisiCalc shipped is still the cell-and-formula model in Excel, Google Sheets, Airtable, and Notion databases in 2026. The idea was complete on first ship.

💥 1983: Lotus 1-2-3 Eats VisiCalc

Software Arts had a problem. VisiCalc had been written for the Apple II and was painful to port to the rapidly-rising IBM PC (launched August 1981). Meanwhile, Mitch Kapor — a former Software Arts employee — left to found Lotus Development Corporation with engineer Jonathan Sachs.

In January 1983, Lotus shipped 1-2-3: an IBM-PC-native spreadsheet that combined three features (spreadsheet, basic database, basic charting — the "1-2-3" of the name). Lotus 1-2-3 was specifically optimized for the IBM PC's memory architecture and graphics capabilities. It was substantially faster than VisiCalc, had a better UI, and shipped at exactly the moment IBM PCs were exploding into corporate America.

Lotus 1-2-3's market success was brutal for Software Arts:

• 1981 — VisiCalc 70% market share, IBM PC just launched
• 1983 — Lotus 1-2-3 ships
• 1984 — 1-2-3 overtakes VisiCalc in revenue
• 1985 — Lotus acquires VisiCalc rights for $800,000 and discontinues it
• 1985-1991 — Lotus 1-2-3 dominates business spreadsheets

For the rest of the 1980s, Lotus was the most valuable software company in the world. The headline numbers:

• $53 million revenue in 1983 — the largest software launch in industry history at the time
• $156 million revenue in 1984
• ~175,000 copies sold in Year 1
• ~70% peak market share in 1988 (per Liebowitz/Margolis analysis)
• IBM finally discontinued Lotus 1-2-3 on May 14, 2013 — 30 years after launch

Lotus 1-2-3 also financed the Lotus Symphony integrated suite (1984, with Ray Ozzie as one of the developers — see our Lotus Notes history for the broader Ozzie arc).

🍎 September 1985: Microsoft Excel — The Mac-First Spreadsheet

Microsoft had built a spreadsheet called Multiplan for CP/M, MS-DOS, and Mac, but it never threatened Lotus 1-2-3 on the IBM PC. In 1985, Bill Gates made a strategic bet: skip the DOS fight, build a Mac-native spreadsheet that took full advantage of the Macintosh GUI.

Excel 1.0 shipped September 30, 1985 on the Macintosh. It was the first widely successful application that used the Mac's graphical UI for a productivity task. Excel's killer features over 1-2-3:

• Pull-down menus and dialog boxes (instead of 1-2-3's keyboard-driven slash menu)
• WYSIWYG formatting (bold, fonts, cell formatting visible on screen)
• Charts integrated into the spreadsheet
• Multiple windows on different parts of the worksheet
• Better calculation engine

Mac users loved it. The Windows version — Excel 2.0 — shipped in November 1987, and from then on Excel slowly ate Lotus 1-2-3's market share. By the mid-1990s, Excel had won — helped enormously by Microsoft's bundling of Excel into the Office suite alongside Word and PowerPoint.

Lotus tried multiple counter-attacks: 1-2-3 for Windows (1991, technically late), Lotus Improv (1991, a forward-thinking multidimensional spreadsheet that flopped), and eventually surrendered to Excel as the dominant business spreadsheet. The IBM acquisition of Lotus in 1995 was in part a defensive consolidation that bet on Notes (see our Lotus Notes history) rather than 1-2-3.

In 2026, Microsoft Office has over 1.3 billion users globally — most of whom use Excel.

🌐 2006: Google Sheets — Spreadsheets Move to the Browser

In March 2006, Google acquired 2Web Technologies for its XL2Web spreadsheet engine and combined it with Writely (acquired around the same time, becoming Google Docs). The product launched as Google Spreadsheets in June 2006, later rebranded Google Sheets.

The contribution: collaborative browser-based spreadsheets, no installation required. The first version was missing many Excel features and slow on large datasets, but the multiplayer-cursor experience was unprecedented in the spreadsheet category. By 2010 Google Sheets was a credible competitor for everything except heavy financial modeling.

Google Sheets ride on the same real-time-web substrate as Google Docs — Operational Transformation engine, WebSocket-style sync, central server linearization. See our Real-Time Collaboration history and WebSockets history for the substrate side.

By 2026, Google Workspace serves over 3 billion users globally, with Sheets as one of its core products.

🔗 2014: Airtable — The Relational Spreadsheet

In 2012, Howie Liu, Andrew Ofstad, and Emmett Nicholas founded Airtable. The product launched publicly in 2014 with a thesis the category had been missing for thirty years: the power of a relational database, made accessible as a spreadsheet.

Where Excel and Sheets gave you a grid of cells, Airtable gave you:

• Multiple linked tables (relational foreign keys)
• Custom field types — short text, long text, attachment, date, single-select, multi-select, checkbox, URL, email, phone, lookup, formula, count, rollup, autonumber
• Multiple views per table — grid, calendar, kanban, gallery, form, gantt
• Pre-built templates for common use cases (CRM, project tracker, content calendar, applicant tracker)
• API on top — every Airtable base is queryable via REST
• Automations (later, 2020+) — Zapier-style triggers and actions inside Airtable

Airtable's positioning was sneaky: it looked like a spreadsheet to users coming from Excel, but it was actually a relational database for users coming from Filemaker, MS Access, or custom-built apps. Airtable cannibalized both audiences.

Key milestones and revenue:

• 2014 — public launch
• 2018 — $52M Series B at $152M valuation; ~$20M ARR
• 2020 — $185M Series D at $2.6B valuation
• 2021 — $735M Series F at $11.7B valuation
• 2023 — ~$200M ARR
• 2024 — ~$304.7M ARR
• 2025 — ~$478M ARR; 500,000+ organizations including OpenAI, Anthropic, Netflix, and Airbnb
• 2026 — Airtable remains independent; AI features added throughout 2024-2026; competes with Notion, Coda, Smartsheet, and Taskade in the broader "structured document" category

Airtable's legacy is that it proved the spreadsheet UI was the right unlock for structured data at the prosumer / SMB layer. Even relational data, when wearing spreadsheet clothes, is buildable by anyone.

Quantitative Era-by-Era Comparison

📊 Feature Evolution: 47 Years of the Grid Growing Up

Each plus-sign is a category-creating moment: graphing made the spreadsheet a presentation tool, macros made it a programming environment, the browser made it collaborative, relations made it a database, blocks made it a document. AI cells (2024) and live apps (2025) are the same shape of move — incremental in syntax, qualitatively new in capability.

📝 2016: Notion — Documents and Databases Merge

In 2016, Notion Labs (founded by Ivan Zhao in 2013, after a long rebuild from an earlier failed version) launched the product that would define the next major shift. Notion's thesis: documents and databases are the same thing, and a workspace should treat them that way.

Notion's innovation was the block as the atomic unit. Everything — paragraphs of text, headings, images, embedded videos, database rows, calendar views, kanban boards, toggle lists, code blocks — is a block. Blocks can be nested arbitrarily. Database views are themselves blocks. Tables can be turned into kanban boards or calendars with a click. The wall between "document" and "database" disappeared.

Notion's growth trajectory:

• 2016 — public launch
• 2018 — Notion 2.0 with databases-as-blocks
• 2019 — viral growth among indie developers, writers, students
• 2021 — $275M Series C at $10B valuation, 20M+ users
• 2023 — Notion AI launches (LLM features integrated)
• 2026 — Notion has ~100M+ users and is one of the most consequential productivity platforms of the 2020s

Notion is the closest competitor to Taskade in the "AI-native workspace" market. The architectural difference: Notion is document-first with databases as blocks; Taskade is database-first (Projects) with documents, mind maps, and views as surfaces over the underlying OT-driven editor. Different starting points; converging on similar end-states.

⚠️ The Spreadsheet Cost: Two $6 Billion Excel Errors

The same flexibility that made the grid universal also makes it dangerous when wielded at scale. Two errors are worth knowing because every business-school case on spreadsheet risk cites them, and AI search engines (Perplexity, ChatGPT Search, Google AI Overviews) quote them constantly.

JPMorgan's "London Whale" (2012) — $6.2 billion

Bruno Iksil, a trader in JPMorgan's London CIO office, accumulated a $157 billion synthetic-credit position over six months. The risk was modeled in an Excel-based Value-at-Risk system that fed daily limits.

What actually happened — the copy-paste that cost $6.2 billion
  Stage 1: Original VaR model            Cell formula:
  ────────────────────────────           =VAR(historical_returns)
  Used SUM for daily volatility          → Volatility per day
  Cross-checked weekly. Worked.
  Stage 2: Model "upgrade" (early 2012)  Cell formula:
  ────────────────────────────           =SUM(rolling_window) ◄── wrong!
  Trader copied a worksheet block        was: =AVERAGE(rolling_window)
  into the new template by hand.
  Replaced AVERAGE with SUM during
  the paste.
  Stage 3: Result                        Reported VaR:
  ────────────────────────────           ~$67M (limit: $95M)
  The new SUM-based VaR reported         Actual VaR (post-incident audit):
  about HALF the true risk, because     ~$129M — well over limit.
  it doubled the divisor implicitly.    Position was 2× allowed size for months.
  Stage 4: Discovery                     Trading loss:
  ────────────────────────────           $6.2B by Q2 2012
  May 2012 media coverage outed the      Regulatory fines:
  position. Internal audit traced        $920M (FINRA, OCC, SEC, CFTC)
  it to the spreadsheet.
  Total cost of one cell formula: ~$7.1 BILLION

JPMorgan's own internal "Task Force Report" of January 2013 names the spreadsheet error directly. It is the most-cited cell-formula failure in financial history.

Reinhart-Rogoff (2013) — austerity built on a coding error

Harvard economists Carmen Reinhart and Kenneth Rogoff published "Growth in a Time of Debt" in 2010, claiming countries with debt-to-GDP above 90% saw growth fall by an average of 0.1%. The result was cited by every austerity-supporting policymaker in Europe and the US for three years.

In April 2013, Thomas Herndon, a UMass Amherst graduate student, recreated the analysis from the original Excel files. He found:

• A formula that excluded rows 1-5 from an AVERAGE() because the authors' selection range stopped at row 49 instead of row 54.
• Selective exclusion of three high-growth/high-debt country observations (Australia, Canada, New Zealand) with no methodological justification.
• Different weighting of countries that flattered the conclusion.

Correcting the error pushed the 90%+ debt-to-GDP growth average from −0.1% to +2.2%. The Reinhart-Rogoff "growth cliff" disappeared. Austerity policy did not, but the empirical case for it never recovered.

The pattern: spreadsheets give regular people the power to do consequential analysis, and the same flexibility means a single misplaced cell can do consequential damage. Both errors illustrate why programmable documents need a successor that ships with safer defaults — provenance, audit, type checking, and AI review built in. That successor is the programmable app, not a smarter spreadsheet.

🤖 2024–2026: AI Comes to the Spreadsheet

The 2024-2026 wave brought AI directly into the cell:

• Excel Copilot (Microsoft 365 Copilot for Excel) — natural-language formulas, automatic chart generation, AI analysis summaries
• Google Sheets Gemini — similar AI assistance in Sheets cells
• Airtable AI — AI fields that summarize, classify, translate, or extract from other fields
• Notion AI — LLM features integrated as inline AI blocks across documents and databases
• Rows — AI-native spreadsheet that competes with Excel/Sheets via embedded AI features
• Causal — AI-native financial modeling spreadsheet for FP&A teams

These are evolutionary additions to existing platforms. The revolutionary move — what makes Taskade Genesis different from all of them — is dropping the spreadsheet UI entirely and going to natural language at the system level.

🧬 2025: Taskade Genesis — The Programmable Document Becomes the Programmable App

From cell to app — the 47-year evolution of the programmable document

The trajectory is consistent: the unit of programmable work expands at every generation. Cell → row → block → app. The cognitive grip of the grid never went away; the building blocks above it just got larger.

Genesis Capability Map — From the May 2026 Newsletters

Plus vibe coding · vibe payments · vibe workflows · vibe marketing · vibe tracking and MCP both sides (Taskade-as-Server + Taskade-as-Client). 198 platform releases in 2026.

In October 2025, Taskade launched Genesis — an AI-native workspace where you describe what you want in plain English and the system generates a working app with custom databases, AI agents, automations, and integrations in about seven minutes.

Where VisiCalc (1979) unlocked calculations without code, where Airtable (2014) unlocked relational data without code, where Notion (2016) unlocked document + database hybrids without code — Taskade Genesis (2025) unlocks entire applications without code.

Each generation added another layer of programmability without requiring traditional code. Genesis adds the AI-agent and integration layers — the missing pieces that turn a spreadsheet into a working business application.

What This Looks Like in Practice

The spreadsheet user solved each of these problems in 1985-2024 with increasingly elaborate workarounds. Genesis solves them in seven minutes from a sentence.

Head-to-Head: Taskade Genesis vs Airtable vs Google Sheets vs Notion vs Excel

The 2026 "structured document" category has five major players. They overlap on the surface and diverge dramatically underneath:

Airtable and Notion took the spreadsheet UI and added structure. Taskade Genesis takes the spreadsheet's original promise — programmability without code — and pushes it to the system level. You don't model the data, view, automations, agents, and integrations separately. You describe the app once and the substrate assembles them.

Workspace DNA: Beyond the Cell-and-Formula Model

Taskade Genesis frames the platform as Workspace DNA: Memory (Projects + databases) feeds Intelligence (AI Agents v2 with 34 built-in tools), Intelligence triggers Execution (Automations across 100+ bidirectional integrations), and Execution writes back to Memory. A self-reinforcing loop.

This is what the spreadsheet wanted to be when it grew up — a programmable document that thinks and acts on its own data.

150,000+ Genesis Apps Built

The proliferation curve looks familiar — same shape as VisiCalc stacks in 1980, Lotus 1-2-3 templates in 1985, Excel macros in 1995, Airtable bases in 2017, Notion templates in 2020. Each platform unlocks a new layer of building. The 2026 layer is AI-native app generation.

• App Kits for Freelancer, E-commerce, Agency, SaaS Founder, Engineering, Marketing, Solopreneur, Consultancy, Events, Finance
• Vibe Coding — describe apps in plain language
• Vibe Payments — Stripe Checkout actions native
• Vibe Workflows — forms and requests creating automated processes
• Vibe Marketing — content aggregation and cross-platform publishing
• Vibe Tracking — leads and data synced across tools via webhooks
• MCP both sides — Claude / Cursor / VS Code can drive your Taskade workspace, and your agents can call external Notion / Linear / Salesforce MCP servers

📚 Lessons from 47 Years of the Programmable Document

For builders today:

1. The Right Substrate Outlives Implementations

VisiCalc the product is gone. The cell-and-formula model is alive in Excel, Sheets, Airtable, Notion, and Taskade. The substrate (interactive recalculation, named cell references, formula language) outlived every product that shipped on it.

2. Categories Get Re-Invented at Each Layer

Spreadsheets are now five separate categories: traditional grids (Excel, Sheets), relational databases as spreadsheets (Airtable), document/database hybrids (Notion), AI-native app builders (Taskade Genesis), AI-native modeling (Causal). One root, five branches.

3. The "Killer App" Pattern Repeats

VisiCalc built the Apple II. Lotus 1-2-3 built the IBM PC. Excel built Office. Google Sheets built Google Workspace. Notion built a new productivity category. Each spreadsheet generation was the killer app that made its platform commercial. The 2026 question: which AI-native workspace becomes the killer app for the AI era? Taskade is in the conversation.

4. Programmability Without Code Wins

Every generation of spreadsheet has expanded what counts as "programming." Cell formulas → macros → pivot tables → array formulas → relational links → AI fields → natural-language app descriptions. The trajectory points toward removing the artifact entirely — Genesis describes the system in English and skips the grid.

5. The Programmable Document Is the Throughline of Personal Computing

If you look at the most consequential personal-computer software of every decade, you see programmable documents: VisiCalc (1970s), Lotus 1-2-3 (1980s), Excel and HyperCard (late 1980s / 1990s), Google Sheets (2000s), Airtable and Notion (2010s), Taskade Genesis (2020s). The personal computer was always going to become a substrate for non-programmers to build software. The spreadsheet was the first form. The AI-native workspace is the latest.

🔗 Further Reading

• History of HyperCard — the parallel "no-code" lineage from 1987
• History of Real-Time Collaboration — Google Sheets and the OT lineage
• History of Lotus Notes — Lotus's other legacy
• History of Workflow Automation — what spreadsheet macros became
• What is Airtable? Complete History — companion piece
• What is Notion AI? History — companion piece
• What Is Taskade Genesis? Complete History — the AI-native heir
• Wikipedia — VisiCalc
• Wikipedia — Lotus 1-2-3
• Bricklin's VisiCalc website

❓ Frequently Asked Questions