As models & controls improve, SaaS orchestration through natural-language APIs will massively improve productivity.

Potential errors could be grave (imagine updating all Hubspot CRM records’ owner to a current user). Monitoring, testing, & rollback/undo buttons will be critical.

If everyone in a company will have the power to orchestrate software through a natural language API, they will also require equivalent tooling for application development.

Over the past few decades, developers have built systems like this for code : code review, testing, monitoring, isolation.

How would security leads ensure the workflow is executed by an authorized person, on permitted data, using sanctioned models, & the process complies with international data regulations? Second, English APIs (aka LLMs) are probabilistic. They err, just like humans.

"Find my top 10 expansion leads this week from Hubspot, then collect their logos, & for each one create an image in Adobe Firefly that puts the company’s logo on a racecar alongside my company’s logo as an attachment to the email.

Then draft an email to each prospect using data from their customer support tickets & save them in my drafts folder." That’s a level of automation rarely seen within the enterprise without custom coding. That workflow alone - if it worked - might save a day’s work.

Examples like thees illustrate conceptually how AI could unlock 1000x the GDP growth of the PC. This vision glosses over many important considerations. First, data security & data loss prevention.

Anyone can write English to enter some input (e.g., “What are my top leads for this week?”) & receive an answer. When every major SaaS application has an English-language API, users of all kinds will be able to ask & receive answers of their applications.

That is a meaningful innovation. English language APIs allow for another layer of abstraction. Imagine being able to write :

What do all of these launches & many other similar ones have in common?

What if Every SaaS App Spoke English? In the last month, The largest SaaS companies in the world have released English-language APIs. Hubspot launched ChatSpot. Zendesk AI automates customer support. Salesforce launched AI Cloud. Adobe’s released Firefly.

Web3 startups will sell software to web2 buyers & in the process become web2 SaaS businesses with unique & defensible architectural advantages & a different capital structure.

Web3 software & infrastructure companies yearning to thrive will need to look beyond the web3 buyer base to new markets with larger willingness to spend. The cloud (web2 software & infrastructure) has captured 40%+ of a $1.5t annual spend on software.

But web3-to-web2 sales must be the future for companies to scale. In that transition, web3 software & infrastructure companies will shed their language of wallets, blockchains, & tokens for terms most buyers understand : accounts, databases, & credits.

Those billions are the future of web3 startup growth. Web3 has created novel technologies that custody data, enable faster & more secure ways of moving money, guarantee provenance, & enable proofs of many things (identity, funds, insurance, presence to name a few).

If the current pace continues, web3 startup fundraising will fall by 73% in 2023. The absence of those dollars will flatten & shrink the already modest addressable market for web3 software & infrastructure.

In the mid-2010s, every web1 company became a web2 company. This time it’s different. Many web3 companies will become web2 companies, too. Why will web3 companies dress like SaaS mutton?

A mental model like this should guide both software procurement on the buyer side & product management on the vendor side especially in an procurement environment where every software vendor must justify their cost-savings or revenue increase.

There will be exceptions, especially where a user base is predominantly individual users or the willingness to pay for the software outweighs the productivity gains.

This 2x2 matrix captures these ideas. Startups should focus on AI for Execs (perhaps why legal has been a heavily funded category) & AI for common workstreams (code autocompletion).

This thought experiment highlights a few ideas to validate : 1. AI initiatives focused on cost-reduction should start with engineering.

AI’s contribution to engineering productivity is 5x+ greater than other teams because of the scale of the gain & the company’s team composition.

Workstreams within SaaS companies are specialized : content marketing, a valuable GTM strategy for many businesses, represents a low-single digit percentage of the employee population.

3. AI startups building next-generation offerings should prioritize common workflows shared across a significant fraction of the employee population or workflows for highly paid employees.